File Manager

003 File Manager

Current Path: /usr/src/sys/contrib/openzfs/module/zfs

usr / src / sys / contrib / openzfs / module / zfs /

📁 ..
📄 Makefile.in(4.57 KB)
📄 THIRDPARTYLICENSE.cityhash(1.03 KB)
📄 THIRDPARTYLICENSE.cityhash.descrip(39 B)
📄 abd.c(31.72 KB)
📄 aggsum.c(8.46 KB)
📄 arc.c(330.91 KB)
📄 blkptr.c(4.26 KB)
📄 bplist.c(2.38 KB)
📄 bpobj.c(27.82 KB)
📄 bptree.c(8.2 KB)
📄 bqueue.c(4.7 KB)
📄 btree.c(65.19 KB)
📄 dataset_kstats.c(6.32 KB)
📄 dbuf.c(141.84 KB)
📄 dbuf_stats.c(6.05 KB)
📄 ddt.c(28.01 KB)
📄 ddt_zap.c(4.38 KB)
📄 dmu.c(60.44 KB)
📄 dmu_diff.c(6.25 KB)
📄 dmu_object.c(14.73 KB)
📄 dmu_objset.c(79.38 KB)
📄 dmu_recv.c(98.69 KB)
📄 dmu_redact.c(37.09 KB)
📄 dmu_send.c(95.51 KB)
📄 dmu_traverse.c(21.98 KB)
📄 dmu_tx.c(40.13 KB)
📄 dmu_zfetch.c(13.56 KB)
📄 dnode.c(71.32 KB)
📄 dnode_sync.c(24.93 KB)
📄 dsl_bookmark.c(51.22 KB)
📄 dsl_crypt.c(78.06 KB)
📄 dsl_dataset.c(141.61 KB)
📄 dsl_deadlist.c(27.94 KB)
📄 dsl_deleg.c(19.92 KB)
📄 dsl_destroy.c(36.73 KB)
📄 dsl_dir.c(65.45 KB)
📄 dsl_pool.c(42.92 KB)
📄 dsl_prop.c(32.88 KB)
📄 dsl_scan.c(134.47 KB)
📄 dsl_synctask.c(7.94 KB)
📄 dsl_userhold.c(18.26 KB)
📄 edonr_zfs.c(3.21 KB)
📄 fm.c(39.75 KB)
📄 gzip.c(2.57 KB)
📄 hkdf.c(4.5 KB)
📄 lz4.c(26.46 KB)
📄 lzjb.c(3.85 KB)
📄 metaslab.c(190.62 KB)
📄 mmp.c(24.93 KB)
📄 multilist.c(12.21 KB)
📄 objlist.c(2.58 KB)
📄 pathname.c(2.52 KB)
📄 range_tree.c(24.9 KB)
📄 refcount.c(7.82 KB)
📄 rrwlock.c(10.82 KB)
📄 sa.c(59.23 KB)
📄 sha256.c(2.89 KB)
📄 skein_zfs.c(2.92 KB)
📄 spa.c(272.81 KB)
📄 spa_boot.c(1.28 KB)
📄 spa_checkpoint.c(21.54 KB)
📄 spa_config.c(16.97 KB)
📄 spa_errlog.c(11.35 KB)
📄 spa_history.c(17.75 KB)
📄 spa_log_spacemap.c(47.13 KB)
📄 spa_misc.c(75.4 KB)
📄 spa_stats.c(27.08 KB)
📄 space_map.c(31.45 KB)
📄 space_reftree.c(4.21 KB)
📄 txg.c(27.49 KB)
📄 uberblock.c(2.17 KB)
📄 unique.c(2.53 KB)
📄 vdev.c(143.91 KB)
📄 vdev_cache.c(11.71 KB)
📄 vdev_draid.c(96.86 KB)
📄 vdev_draid_rand.c(1.16 KB)
📄 vdev_indirect.c(60.83 KB)
📄 vdev_indirect_births.c(5.88 KB)
📄 vdev_indirect_mapping.c(18.06 KB)
📄 vdev_initialize.c(22.38 KB)
📄 vdev_label.c(57.52 KB)
📄 vdev_mirror.c(26.76 KB)
📄 vdev_missing.c(3.74 KB)
📄 vdev_queue.c(37.61 KB)
📄 vdev_raidz.c(76.09 KB)
📄 vdev_raidz_math.c(17.18 KB)
📄 vdev_raidz_math_aarch64_neon.c(112.13 KB)
📄 vdev_raidz_math_aarch64_neon_common.h(23.64 KB)
📄 vdev_raidz_math_aarch64_neonx2.c(5.39 KB)
📄 vdev_raidz_math_avx2.c(10.64 KB)
📄 vdev_raidz_math_avx512bw.c(10.93 KB)
📄 vdev_raidz_math_avx512f.c(17.9 KB)
📄 vdev_raidz_math_impl.h(34.78 KB)
📄 vdev_raidz_math_powerpc_altivec.c(213.46 KB)
📄 vdev_raidz_math_powerpc_altivec_common.h(22.62 KB)
📄 vdev_raidz_math_scalar.c(9.54 KB)
📄 vdev_raidz_math_sse2.c(25.84 KB)
📄 vdev_raidz_math_ssse3.c(118 KB)
📄 vdev_rebuild.c(34.65 KB)
📄 vdev_removal.c(72.08 KB)
📄 vdev_root.c(4.28 KB)
📄 vdev_trim.c(50.92 KB)
📄 zap.c(35.35 KB)
📄 zap_leaf.c(22.75 KB)
📄 zap_micro.c(42.18 KB)
📄 zcp.c(42.29 KB)
📄 zcp_get.c(21.01 KB)
📄 zcp_global.c(1.9 KB)
📄 zcp_iter.c(18.21 KB)
📄 zcp_set.c(2.37 KB)
📄 zcp_synctask.c(13.69 KB)
📄 zfeature.c(17.6 KB)
📄 zfs_byteswap.c(5.64 KB)
📄 zfs_fm.c(40.58 KB)
📄 zfs_fuid.c(19.69 KB)
📄 zfs_ioctl.c(195.48 KB)
📄 zfs_log.c(20.6 KB)
📄 zfs_onexit.c(5.09 KB)
📄 zfs_quota.c(12.93 KB)
📄 zfs_ratelimit.c(2.35 KB)
📄 zfs_replay.c(26.11 KB)
📄 zfs_rlock.c(20.38 KB)
📄 zfs_sa.c(13.09 KB)
📄 zfs_vnops.c(22 KB)
📄 zil.c(111.32 KB)
📄 zio.c(144.72 KB)
📄 zio_checksum.c(17.28 KB)
📄 zio_compress.c(5.92 KB)
📄 zio_inject.c(25.79 KB)
📄 zle.c(2.52 KB)
📄 zrlock.c(4.45 KB)
📄 zthr.c(16.41 KB)
📄 zvol.c(42.96 KB)

Editing: dmu_zfetch.c

/*
 * CDDL HEADER START
 *
 * The contents of this file are subject to the terms of the
 * Common Development and Distribution License (the "License").
 * You may not use this file except in compliance with the License.
 *
 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
 * or http://www.opensolaris.org/os/licensing.
 * See the License for the specific language governing permissions
 * and limitations under the License.
 *
 * When distributing Covered Code, include this CDDL HEADER in each
 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
 * If applicable, add the following below this CDDL HEADER, with the
 * fields enclosed by brackets "[]" replaced with your own identifying
 * information: Portions Copyright [yyyy] [name of copyright owner]
 *
 * CDDL HEADER END
 */
/*
 * Copyright 2009 Sun Microsystems, Inc.  All rights reserved.
 * Use is subject to license terms.
 */

/*
 * Copyright (c) 2013, 2017 by Delphix. All rights reserved.
 */

#include <sys/zfs_context.h>
#include <sys/dnode.h>
#include <sys/dmu_objset.h>
#include <sys/dmu_zfetch.h>
#include <sys/dmu.h>
#include <sys/dbuf.h>
#include <sys/kstat.h>

/*
 * This tunable disables predictive prefetch.  Note that it leaves "prescient"
 * prefetch (e.g. prefetch for zfs send) intact.  Unlike predictive prefetch,
 * prescient prefetch never issues i/os that end up not being needed,
 * so it can't hurt performance.
 */

int zfs_prefetch_disable = B_FALSE;

/* max # of streams per zfetch */
unsigned int	zfetch_max_streams = 8;
/* min time before stream reclaim */
unsigned int	zfetch_min_sec_reap = 2;
/* max bytes to prefetch per stream (default 8MB) */
unsigned int	zfetch_max_distance = 8 * 1024 * 1024;
/* max bytes to prefetch indirects for per stream (default 64MB) */
unsigned int	zfetch_max_idistance = 64 * 1024 * 1024;
/* max number of bytes in an array_read in which we allow prefetching (1MB) */
unsigned long	zfetch_array_rd_sz = 1024 * 1024;

typedef struct zfetch_stats {
	kstat_named_t zfetchstat_hits;
	kstat_named_t zfetchstat_misses;
	kstat_named_t zfetchstat_max_streams;
	kstat_named_t zfetchstat_max_completion_us;
	kstat_named_t zfetchstat_last_completion_us;
	kstat_named_t zfetchstat_io_issued;
} zfetch_stats_t;

static zfetch_stats_t zfetch_stats = {
	{ "hits",			KSTAT_DATA_UINT64 },
	{ "misses",			KSTAT_DATA_UINT64 },
	{ "max_streams",		KSTAT_DATA_UINT64 },
	{ "max_completion_us",		KSTAT_DATA_UINT64 },
	{ "last_completion_us",		KSTAT_DATA_UINT64 },
	{ "io_issued",		KSTAT_DATA_UINT64 },
};

#define	ZFETCHSTAT_BUMP(stat) \
	atomic_inc_64(&zfetch_stats.stat.value.ui64)
#define	ZFETCHSTAT_ADD(stat, val)				\
	atomic_add_64(&zfetch_stats.stat.value.ui64, val)
#define	ZFETCHSTAT_SET(stat, val)				\
	zfetch_stats.stat.value.ui64 = val
#define	ZFETCHSTAT_GET(stat)					\
	zfetch_stats.stat.value.ui64

kstat_t		*zfetch_ksp;

void
zfetch_init(void)
{
	zfetch_ksp = kstat_create("zfs", 0, "zfetchstats", "misc",
	    KSTAT_TYPE_NAMED, sizeof (zfetch_stats) / sizeof (kstat_named_t),
	    KSTAT_FLAG_VIRTUAL);

if (zfetch_ksp != NULL) {
		zfetch_ksp->ks_data = &zfetch_stats;
		kstat_install(zfetch_ksp);
	}
}

void
zfetch_fini(void)
{
	if (zfetch_ksp != NULL) {
		kstat_delete(zfetch_ksp);
		zfetch_ksp = NULL;
	}
}

/*
 * This takes a pointer to a zfetch structure and a dnode.  It performs the
 * necessary setup for the zfetch structure, grokking data from the
 * associated dnode.
 */
void
dmu_zfetch_init(zfetch_t *zf, dnode_t *dno)
{
	if (zf == NULL)
		return;
	zf->zf_dnode = dno;
	zf->zf_numstreams = 0;

list_create(&zf->zf_stream, sizeof (zstream_t),
	    offsetof(zstream_t, zs_node));

mutex_init(&zf->zf_lock, NULL, MUTEX_DEFAULT, NULL);
}

static void
dmu_zfetch_stream_fini(zstream_t *zs)
{
	mutex_destroy(&zs->zs_lock);
	kmem_free(zs, sizeof (*zs));
}

static void
dmu_zfetch_stream_remove(zfetch_t *zf, zstream_t *zs)
{
	ASSERT(MUTEX_HELD(&zf->zf_lock));
	list_remove(&zf->zf_stream, zs);
	dmu_zfetch_stream_fini(zs);
	zf->zf_numstreams--;
}

static void
dmu_zfetch_stream_orphan(zfetch_t *zf, zstream_t *zs)
{
	ASSERT(MUTEX_HELD(&zf->zf_lock));
	list_remove(&zf->zf_stream, zs);
	zs->zs_fetch = NULL;
	zf->zf_numstreams--;
}

/*
 * Clean-up state associated with a zfetch structure (e.g. destroy the
 * streams).  This doesn't free the zfetch_t itself, that's left to the caller.
 */
void
dmu_zfetch_fini(zfetch_t *zf)
{
	zstream_t *zs;

mutex_enter(&zf->zf_lock);
	while ((zs = list_head(&zf->zf_stream)) != NULL) {
		if (zfs_refcount_count(&zs->zs_blocks) != 0)
			dmu_zfetch_stream_orphan(zf, zs);
		else
			dmu_zfetch_stream_remove(zf, zs);
	}
	mutex_exit(&zf->zf_lock);
	list_destroy(&zf->zf_stream);
	mutex_destroy(&zf->zf_lock);

zf->zf_dnode = NULL;
}

/*
 * If there aren't too many streams already, create a new stream.
 * The "blkid" argument is the next block that we expect this stream to access.
 * While we're here, clean up old streams (which haven't been
 * accessed for at least zfetch_min_sec_reap seconds).
 */
static void
dmu_zfetch_stream_create(zfetch_t *zf, uint64_t blkid)
{
	zstream_t *zs_next;
	hrtime_t now = gethrtime();

ASSERT(MUTEX_HELD(&zf->zf_lock));

/*
	 * Clean up old streams.
	 */
	for (zstream_t *zs = list_head(&zf->zf_stream);
	    zs != NULL; zs = zs_next) {
		zs_next = list_next(&zf->zf_stream, zs);
		/*
		 * Skip gethrtime() call if there are still references
		 */
		if (zfs_refcount_count(&zs->zs_blocks) != 0)
			continue;
		if (((now - zs->zs_atime) / NANOSEC) >
		    zfetch_min_sec_reap)
			dmu_zfetch_stream_remove(zf, zs);
	}

/*
	 * The maximum number of streams is normally zfetch_max_streams,
	 * but for small files we lower it such that it's at least possible
	 * for all the streams to be non-overlapping.
	 *
	 * If we are already at the maximum number of streams for this file,
	 * even after removing old streams, then don't create this stream.
	 */
	uint32_t max_streams = MAX(1, MIN(zfetch_max_streams,
	    zf->zf_dnode->dn_maxblkid * zf->zf_dnode->dn_datablksz /
	    zfetch_max_distance));
	if (zf->zf_numstreams >= max_streams) {
		ZFETCHSTAT_BUMP(zfetchstat_max_streams);
		return;
	}

zstream_t *zs = kmem_zalloc(sizeof (*zs), KM_SLEEP);
	zs->zs_blkid = blkid;
	zs->zs_pf_blkid = blkid;
	zs->zs_ipf_blkid = blkid;
	zs->zs_atime = now;
	zs->zs_fetch = zf;
	zfs_refcount_create(&zs->zs_blocks);
	mutex_init(&zs->zs_lock, NULL, MUTEX_DEFAULT, NULL);
	zf->zf_numstreams++;
	list_insert_head(&zf->zf_stream, zs);
}

static void
dmu_zfetch_stream_done(void *arg, boolean_t io_issued)
{
	zstream_t *zs = arg;

if (zs->zs_start_time && io_issued) {
		hrtime_t now = gethrtime();
		hrtime_t delta = NSEC2USEC(now - zs->zs_start_time);

zs->zs_start_time = 0;
		ZFETCHSTAT_SET(zfetchstat_last_completion_us, delta);
		if (delta > ZFETCHSTAT_GET(zfetchstat_max_completion_us))
			ZFETCHSTAT_SET(zfetchstat_max_completion_us, delta);
	}

if (zfs_refcount_remove(&zs->zs_blocks, NULL) != 0)
		return;

/*
	 * The parent fetch structure has gone away
	 */
	if (zs->zs_fetch == NULL)
		dmu_zfetch_stream_fini(zs);
}

/*
 * This is the predictive prefetch entry point.  It associates dnode access
 * specified with blkid and nblks arguments with prefetch stream, predicts
 * further accesses based on that stats and initiates speculative prefetch.
 * fetch_data argument specifies whether actual data blocks should be fetched:
 *   FALSE -- prefetch only indirect blocks for predicted data blocks;
 *   TRUE -- prefetch predicted data blocks plus following indirect blocks.
 */
void
dmu_zfetch(zfetch_t *zf, uint64_t blkid, uint64_t nblks, boolean_t fetch_data,
    boolean_t have_lock)
{
	zstream_t *zs;
	int64_t pf_start, ipf_start, ipf_istart, ipf_iend;
	int64_t pf_ahead_blks, max_blks;
	int epbs, max_dist_blks, pf_nblks, ipf_nblks, issued;
	uint64_t end_of_access_blkid;
	end_of_access_blkid = blkid + nblks;
	spa_t *spa = zf->zf_dnode->dn_objset->os_spa;

if (zfs_prefetch_disable)
		return;
	/*
	 * If we haven't yet loaded the indirect vdevs' mappings, we
	 * can only read from blocks that we carefully ensure are on
	 * concrete vdevs (or previously-loaded indirect vdevs).  So we
	 * can't allow the predictive prefetcher to attempt reads of other
	 * blocks (e.g. of the MOS's dnode object).
	 */
	if (!spa_indirect_vdevs_loaded(spa))
		return;

/*
	 * As a fast path for small (single-block) files, ignore access
	 * to the first block.
	 */
	if (!have_lock && blkid == 0)
		return;

if (!have_lock)
		rw_enter(&zf->zf_dnode->dn_struct_rwlock, RW_READER);

/*
	 * A fast path for small files for which no prefetch will
	 * happen.
	 */
	if (zf->zf_dnode->dn_maxblkid < 2) {
		if (!have_lock)
			rw_exit(&zf->zf_dnode->dn_struct_rwlock);
		return;
	}
	mutex_enter(&zf->zf_lock);

/*
	 * Find matching prefetch stream.  Depending on whether the accesses
	 * are block-aligned, first block of the new access may either follow
	 * the last block of the previous access, or be equal to it.
	 */
	for (zs = list_head(&zf->zf_stream); zs != NULL;
	    zs = list_next(&zf->zf_stream, zs)) {
		if (blkid == zs->zs_blkid || blkid + 1 == zs->zs_blkid) {
			mutex_enter(&zs->zs_lock);
			/*
			 * zs_blkid could have changed before we
			 * acquired zs_lock; re-check them here.
			 */
			if (blkid == zs->zs_blkid) {
				break;
			} else if (blkid + 1 == zs->zs_blkid) {
				blkid++;
				nblks--;
				if (nblks == 0) {
					/* Already prefetched this before. */
					mutex_exit(&zs->zs_lock);
					mutex_exit(&zf->zf_lock);
					if (!have_lock) {
						rw_exit(&zf->zf_dnode->
						    dn_struct_rwlock);
					}
					return;
				}
				break;
			}
			mutex_exit(&zs->zs_lock);
		}
	}

if (zs == NULL) {
		/*
		 * This access is not part of any existing stream.  Create
		 * a new stream for it.
		 */
		ZFETCHSTAT_BUMP(zfetchstat_misses);

dmu_zfetch_stream_create(zf, end_of_access_blkid);
		mutex_exit(&zf->zf_lock);
		if (!have_lock)
			rw_exit(&zf->zf_dnode->dn_struct_rwlock);
		return;
	}

/*
	 * This access was to a block that we issued a prefetch for on
	 * behalf of this stream. Issue further prefetches for this stream.
	 *
	 * Normally, we start prefetching where we stopped
	 * prefetching last (zs_pf_blkid).  But when we get our first
	 * hit on this stream, zs_pf_blkid == zs_blkid, we don't
	 * want to prefetch the block we just accessed.  In this case,
	 * start just after the block we just accessed.
	 */
	pf_start = MAX(zs->zs_pf_blkid, end_of_access_blkid);

/*
	 * Double our amount of prefetched data, but don't let the
	 * prefetch get further ahead than zfetch_max_distance.
	 */
	if (fetch_data) {
		max_dist_blks =
		    zfetch_max_distance >> zf->zf_dnode->dn_datablkshift;
		/*
		 * Previously, we were (zs_pf_blkid - blkid) ahead.  We
		 * want to now be double that, so read that amount again,
		 * plus the amount we are catching up by (i.e. the amount
		 * read just now).
		 */
		pf_ahead_blks = zs->zs_pf_blkid - blkid + nblks;
		max_blks = max_dist_blks - (pf_start - end_of_access_blkid);
		pf_nblks = MIN(pf_ahead_blks, max_blks);
	} else {
		pf_nblks = 0;
	}

zs->zs_pf_blkid = pf_start + pf_nblks;

/*
	 * Do the same for indirects, starting from where we stopped last,
	 * or where we will stop reading data blocks (and the indirects
	 * that point to them).
	 */
	ipf_start = MAX(zs->zs_ipf_blkid, zs->zs_pf_blkid);
	max_dist_blks = zfetch_max_idistance >> zf->zf_dnode->dn_datablkshift;
	/*
	 * We want to double our distance ahead of the data prefetch
	 * (or reader, if we are not prefetching data).  Previously, we
	 * were (zs_ipf_blkid - blkid) ahead.  To double that, we read
	 * that amount again, plus the amount we are catching up by
	 * (i.e. the amount read now + the amount of data prefetched now).
	 */
	pf_ahead_blks = zs->zs_ipf_blkid - blkid + nblks + pf_nblks;
	max_blks = max_dist_blks - (ipf_start - end_of_access_blkid);
	ipf_nblks = MIN(pf_ahead_blks, max_blks);
	zs->zs_ipf_blkid = ipf_start + ipf_nblks;

epbs = zf->zf_dnode->dn_indblkshift - SPA_BLKPTRSHIFT;
	ipf_istart = P2ROUNDUP(ipf_start, 1 << epbs) >> epbs;
	ipf_iend = P2ROUNDUP(zs->zs_ipf_blkid, 1 << epbs) >> epbs;

zs->zs_atime = gethrtime();
	/* no prior reads in progress */
	if (zfs_refcount_count(&zs->zs_blocks) == 0)
		zs->zs_start_time = zs->zs_atime;
	zs->zs_blkid = end_of_access_blkid;
	zfs_refcount_add_many(&zs->zs_blocks, pf_nblks + ipf_iend - ipf_istart,
	    NULL);
	mutex_exit(&zs->zs_lock);
	mutex_exit(&zf->zf_lock);
	issued = 0;

/*
	 * dbuf_prefetch() is asynchronous (even when it needs to read
	 * indirect blocks), but we still prefer to drop our locks before
	 * calling it to reduce the time we hold them.
	 */

for (int i = 0; i < pf_nblks; i++) {
		issued += dbuf_prefetch_impl(zf->zf_dnode, 0, pf_start + i,
		    ZIO_PRIORITY_ASYNC_READ, ARC_FLAG_PREDICTIVE_PREFETCH,
		    dmu_zfetch_stream_done, zs);
	}
	for (int64_t iblk = ipf_istart; iblk < ipf_iend; iblk++) {
		issued += dbuf_prefetch_impl(zf->zf_dnode, 1, iblk,
		    ZIO_PRIORITY_ASYNC_READ, ARC_FLAG_PREDICTIVE_PREFETCH,
		    dmu_zfetch_stream_done, zs);
	}
	if (!have_lock)
		rw_exit(&zf->zf_dnode->dn_struct_rwlock);
	ZFETCHSTAT_BUMP(zfetchstat_hits);

if (issued)
		ZFETCHSTAT_ADD(zfetchstat_io_issued, issued);
}

/* BEGIN CSTYLED */
ZFS_MODULE_PARAM(zfs_prefetch, zfs_prefetch_, disable, INT, ZMOD_RW,
	"Disable all ZFS prefetching");

ZFS_MODULE_PARAM(zfs_prefetch, zfetch_, max_streams, UINT, ZMOD_RW,
	"Max number of streams per zfetch");

ZFS_MODULE_PARAM(zfs_prefetch, zfetch_, min_sec_reap, UINT, ZMOD_RW,
	"Min time before stream reclaim");

ZFS_MODULE_PARAM(zfs_prefetch, zfetch_, max_distance, UINT, ZMOD_RW,
	"Max bytes to prefetch per stream");

ZFS_MODULE_PARAM(zfs_prefetch, zfetch_, max_idistance, UINT, ZMOD_RW,
	"Max bytes to prefetch indirects for per stream");

ZFS_MODULE_PARAM(zfs_prefetch, zfetch_, array_rd_sz, ULONG, ZMOD_RW,
	"Number of bytes in a array_read");
/* END CSTYLED */

003 File Manager

Editing: dmu_zfetch.c

Upload File

Create Folder