/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright 2011 Nexenta Systems, Inc. All rights reserved.
- * Copyright (c) 2012 by Delphix. All rights reserved.
+ * Copyright (c) 2013 by Delphix. All rights reserved.
* Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
*/
#include <sys/zfs_context.h>
#include <sys/arc.h>
#include <sys/dmu.h>
+#include <sys/dmu_send.h>
#include <sys/dmu_impl.h>
#include <sys/dbuf.h>
#include <sys/dmu_objset.h>
void *tag, dmu_buf_impl_t **dbp, int depth);
static int __dbuf_hold_impl(struct dbuf_hold_impl_data *dh);
+/*
+ * Number of times that zfs_free_range() took the slow path while doing
+ * a zfs receive. A nonzero value indicates a potential performance problem.
+ */
+uint64_t zfs_free_range_recv_miss;
+
static void dbuf_destroy(dmu_buf_impl_t *db);
static boolean_t dbuf_undirty(dmu_buf_impl_t *db, dmu_tx_t *tx);
static void dbuf_write(dbuf_dirty_record_t *dr, arc_buf_t *data, dmu_tx_t *tx);
retry:
h->hash_table_mask = hsize - 1;
#if defined(_KERNEL) && defined(HAVE_SPL)
- /* Large allocations which do not require contiguous pages
- * should be using vmem_alloc() in the linux kernel */
+ /*
+ * Large allocations which do not require contiguous pages
+ * should be using vmem_alloc() in the linux kernel
+ */
h->hash_table = vmem_zalloc(hsize * sizeof (void *), KM_PUSHPAGE);
#else
h->hash_table = kmem_zalloc(hsize * sizeof (void *), KM_NOSLEEP);
for (i = 0; i < DBUF_MUTEXES; i++)
mutex_destroy(&h->hash_mutexes[i]);
#if defined(_KERNEL) && defined(HAVE_SPL)
- /* Large allocations which do not require contiguous pages
- * should be using vmem_free() in the linux kernel */
+ /*
+ * Large allocations which do not require contiguous pages
+ * should be using vmem_free() in the linux kernel
+ */
vmem_free(h->hash_table, (h->hash_table_mask + 1) * sizeof (void *));
#else
kmem_free(h->hash_table, (h->hash_table_mask + 1) * sizeof (void *));
ASSERT(!refcount_is_zero(&db->db_holds));
if (db->db_state == DB_NOFILL)
- return (EIO);
+ return (SET_ERROR(EIO));
DB_DNODE_ENTER(db);
dn = DB_DNODE(db);
if (!havepzio)
err = zio_wait(zio);
} else {
+ /*
+ * Another reader came in while the dbuf was in flight
+ * between UNCACHED and CACHED. Either a writer will finish
+ * writing the buffer (sending the dbuf to CACHED) or the
+ * first reader's request will reach the read_done callback
+ * and send the dbuf to CACHED. Otherwise, a failure
+ * occurred and the dbuf went to UNCACHED.
+ */
mutex_exit(&db->db_mtx);
if (prefetch)
dmu_zfetch(&dn->dn_zfetch, db->db.db_offset,
rw_exit(&dn->dn_struct_rwlock);
DB_DNODE_EXIT(db);
+ /* Skip the wait per the caller's request. */
mutex_enter(&db->db_mtx);
if ((flags & DB_RF_NEVERWAIT) == 0) {
while (db->db_state == DB_READ ||
cv_wait(&db->db_changed, &db->db_mtx);
}
if (db->db_state == DB_UNCACHED)
- err = EIO;
+ err = SET_ERROR(EIO);
}
mutex_exit(&db->db_mtx);
}
ASSERT(db->db_data_pending != dr);
/* free this block */
- if (!BP_IS_HOLE(bp)) {
+ if (!BP_IS_HOLE(bp) && !dr->dt.dl.dr_nopwrite) {
spa_t *spa;
DB_GET_SPA(&spa, db);
zio_free(spa, txg, bp);
}
dr->dt.dl.dr_override_state = DR_NOT_OVERRIDDEN;
+ dr->dt.dl.dr_nopwrite = B_FALSE;
+
/*
* Release the already-written buffer, so we leave it in
* a consistent dirty state. Note that all callers are
/*
* Evict (if its unreferenced) or clear (if its referenced) any level-0
* data blocks in the free range, so that any future readers will find
- * empty blocks. Also, if we happen accross any level-1 dbufs in the
+ * empty blocks. Also, if we happen across any level-1 dbufs in the
* range that have not already been marked dirty, mark them dirty so
* they stay in memory.
+ *
+ * This is a no-op if the dataset is in the middle of an incremental
+ * receive; see comment below for details.
*/
void
dbuf_free_range(dnode_t *dn, uint64_t start, uint64_t end, dmu_tx_t *tx)
last_l1 = end >> epbs;
}
dprintf_dnode(dn, "start=%llu end=%llu\n", start, end);
+
mutex_enter(&dn->dn_dbufs_mtx);
- for (db = list_head(&dn->dn_dbufs); db; db = db_next) {
+ if (start >= dn->dn_unlisted_l0_blkid * dn->dn_datablksz) {
+ /* There can't be any dbufs in this range; no need to search. */
+ mutex_exit(&dn->dn_dbufs_mtx);
+ return;
+ } else if (dmu_objset_is_receiving(dn->dn_objset)) {
+ /*
+ * If we are receiving, we expect there to be no dbufs in
+ * the range to be freed, because receive modifies each
+ * block at most once, and in offset order. If this is
+ * not the case, it can lead to performance problems,
+ * so note that we unexpectedly took the slow path.
+ */
+ atomic_inc_64(&zfs_free_range_recv_miss);
+ }
+
+ for (db = list_head(&dn->dn_dbufs); db != NULL; db = db_next) {
db_next = list_next(&dn->dn_dbufs, db);
ASSERT(db->db_blkid != DMU_BONUS_BLKID);
sizeof (dbuf_dirty_record_t),
offsetof(dbuf_dirty_record_t, dr_dirty_node));
}
+ if (db->db_blkid != DMU_BONUS_BLKID && os->os_dsl_dataset != NULL)
+ dr->dr_accounted = db->db.db_size;
dr->dr_dbuf = db;
dr->dr_txg = tx->tx_txg;
dr->dr_next = *drp;
dbuf_rele(parent, FTAG);
mutex_enter(&db->db_mtx);
- /* possible race with dbuf_undirty() */
+ /*
+ * Since we've dropped the mutex, it's possible that
+ * dbuf_undirty() might have changed this out from under us.
+ */
if (db->db_last_dirty == dr ||
dn->dn_object == DMU_META_DNODE_OBJECT) {
mutex_enter(&di->dt.di.dr_mtx);
}
/*
- * Return TRUE if this evicted the dbuf.
+ * Undirty a buffer in the transaction group referenced by the given
+ * transaction. Return whether this evicted the dbuf.
*/
static boolean_t
dbuf_undirty(dmu_buf_impl_t *db, dmu_tx_t *tx)
ASSERT(db->db.db_size != 0);
- /* XXX would be nice to fix up dn_towrite_space[] */
+ /*
+ * Any space we accounted for in dp_dirty_* will be cleaned up by
+ * dsl_pool_sync(). This is relatively rare so the discrepancy
+ * is not a big deal.
+ */
*drp = dr->dr_next;
/*
* "Clear" the contents of this dbuf. This will mark the dbuf
- * EVICTING and clear *most* of its references. Unfortunetely,
+ * EVICTING and clear *most* of its references. Unfortunately,
* when we are not holding the dn_dbufs_mtx, we can't clear the
* entry in the dn_dbufs list. We have to wait until dbuf_destroy()
* in this case. For callers from the DMU we will usually see:
if (level >= nlevels ||
(blkid > (dn->dn_phys->dn_maxblkid >> (level * epbs)))) {
/* the buffer has no parent yet */
- return (ENOENT);
+ return (SET_ERROR(ENOENT));
} else if (level < nlevels-1) {
/* this block is referenced from an indirect block */
int err;
if (dh == NULL) {
err = dbuf_hold_impl(dn, level+1, blkid >> epbs,
fail_sparse, NULL, parentp);
- }
- else {
+ } else {
__dbuf_hold_impl_init(dh + 1, dn, dh->dh_level + 1,
blkid >> epbs, fail_sparse, NULL,
parentp, dh->dh_depth + 1);
db->db.db_offset = 0;
} else {
int blocksize =
- db->db_level ? 1<<dn->dn_indblkshift : dn->dn_datablksz;
+ db->db_level ? 1 << dn->dn_indblkshift : dn->dn_datablksz;
db->db.db_size = blocksize;
db->db.db_offset = db->db_blkid * blocksize;
}
return (odb);
}
list_insert_head(&dn->dn_dbufs, db);
+ if (db->db_level == 0 && db->db_blkid >=
+ dn->dn_unlisted_l0_blkid)
+ dn->dn_unlisted_l0_blkid = db->db_blkid + 1;
db->db_state = DB_UNCACHED;
mutex_exit(&dn->dn_dbufs_mtx);
arc_space_consume(sizeof (dmu_buf_impl_t), ARC_SPACE_OTHER);
}
void
-dbuf_prefetch(dnode_t *dn, uint64_t blkid)
+dbuf_prefetch(dnode_t *dn, uint64_t blkid, zio_priority_t prio)
{
dmu_buf_impl_t *db = NULL;
blkptr_t *bp = NULL;
if (dbuf_findbp(dn, 0, blkid, TRUE, &db, &bp, NULL) == 0) {
if (bp && !BP_IS_HOLE(bp)) {
- int priority = dn->dn_type == DMU_OT_DDT_ZAP ?
- ZIO_PRIORITY_DDT_PREFETCH : ZIO_PRIORITY_ASYNC_READ;
dsl_dataset_t *ds = dn->dn_objset->os_dsl_dataset;
uint32_t aflags = ARC_NOWAIT | ARC_PREFETCH;
zbookmark_t zb;
dn->dn_object, 0, blkid);
(void) arc_read(NULL, dn->dn_objset->os_spa,
- bp, NULL, NULL, priority,
+ bp, NULL, NULL, prio,
ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE,
&aflags, &zb);
}
}
}
-#define DBUF_HOLD_IMPL_MAX_DEPTH 20
+#define DBUF_HOLD_IMPL_MAX_DEPTH 20
/*
* Returns with db_holds incremented, and db_mtx not held.
dh->dh_fail_sparse, &dh->dh_parent,
&dh->dh_bp, dh);
if (dh->dh_fail_sparse) {
- if (dh->dh_err == 0 && dh->dh_bp && BP_IS_HOLE(dh->dh_bp))
- dh->dh_err = ENOENT;
+ if (dh->dh_err == 0 &&
+ dh->dh_bp && BP_IS_HOLE(dh->dh_bp))
+ dh->dh_err = SET_ERROR(ENOENT);
if (dh->dh_err) {
if (dh->dh_parent)
dbuf_rele(dh->dh_parent, NULL);
struct dbuf_hold_impl_data *dh;
int error;
- dh = kmem_zalloc(sizeof(struct dbuf_hold_impl_data) *
+ dh = kmem_zalloc(sizeof (struct dbuf_hold_impl_data) *
DBUF_HOLD_IMPL_MAX_DEPTH, KM_PUSHPAGE);
__dbuf_hold_impl_init(dh, dn, level, blkid, fail_sparse, tag, dbp, 0);
error = __dbuf_hold_impl(dh);
- kmem_free(dh, sizeof(struct dbuf_hold_impl_data) *
+ kmem_free(dh, sizeof (struct dbuf_hold_impl_data) *
DBUF_HOLD_IMPL_MAX_DEPTH);
return (error);
dnode_t *dn;
if (db->db_blkid != DMU_SPILL_BLKID)
- return (ENOTSUP);
+ return (SET_ERROR(ENOTSUP));
if (blksz == 0)
blksz = SPA_MINBLOCKSIZE;
if (blksz > SPA_MAXBLOCKSIZE)
return (res);
}
+blkptr_t *
+dmu_buf_get_blkptr(dmu_buf_t *db)
+{
+ dmu_buf_impl_t *dbi = (dmu_buf_impl_t *)db;
+ return (dbi->db_blkptr);
+}
+
static void
dbuf_check_blkptr(dnode_t *dn, dmu_buf_impl_t *db)
{
}
}
-/* dbuf_sync_indirect() is called recursively from dbuf_sync_list() so it
+/*
+ * dbuf_sync_indirect() is called recursively from dbuf_sync_list() so it
* is critical the we not allow the compiler to inline this function in to
* dbuf_sync_list() thereby drastically bloating the stack usage.
*/
ASSERT(db->db_level > 0);
DBUF_VERIFY(db);
+ /* Read the block if it hasn't been read yet. */
if (db->db_buf == NULL) {
mutex_exit(&db->db_mtx);
(void) dbuf_read(db, NULL, DB_RF_MUST_SUCCEED);
DB_DNODE_ENTER(db);
dn = DB_DNODE(db);
+ /* Indirect block size must match what the dnode thinks it is. */
ASSERT3U(db->db.db_size, ==, 1<<dn->dn_phys->dn_indblkshift);
dbuf_check_blkptr(dn, db);
DB_DNODE_EXIT(db);
+ /* Provide the pending dirty record to child dbufs */
db->db_data_pending = dr;
mutex_exit(&db->db_mtx);
zio_nowait(zio);
}
-/* dbuf_sync_leaf() is called recursively from dbuf_sync_list() so it is
+/*
+ * dbuf_sync_leaf() is called recursively from dbuf_sync_list() so it is
* critical the we not allow the compiler to inline this function in to
* dbuf_sync_list() thereby drastically bloating the stack usage.
*/
mutex_exit(&db->db_mtx);
}
+/*
+ * The SPA will call this callback several times for each zio - once
+ * for every physical child i/o (zio->io_phys_children times). This
+ * allows the DMU to monitor the progress of each logical i/o. For example,
+ * there may be 2 copies of an indirect block, or many fragments of a RAID-Z
+ * block. There may be a long delay before all copies/fragments are completed,
+ * so this callback allows us to retire dirty space gradually, as the physical
+ * i/os complete.
+ */
+/* ARGSUSED */
+static void
+dbuf_write_physdone(zio_t *zio, arc_buf_t *buf, void *arg)
+{
+ dmu_buf_impl_t *db = arg;
+ objset_t *os = db->db_objset;
+ dsl_pool_t *dp = dmu_objset_pool(os);
+ dbuf_dirty_record_t *dr;
+ int delta = 0;
+
+ dr = db->db_data_pending;
+ ASSERT3U(dr->dr_txg, ==, zio->io_txg);
+
+ /*
+ * The callback will be called io_phys_children times. Retire one
+ * portion of our dirty space each time we are called. Any rounding
+ * error will be cleaned up by dsl_pool_sync()'s call to
+ * dsl_pool_undirty_space().
+ */
+ delta = dr->dr_accounted / zio->io_phys_children;
+ dsl_pool_undirty_space(dp, delta, zio->io_txg);
+}
+
/* ARGSUSED */
static void
dbuf_write_done(zio_t *zio, arc_buf_t *buf, void *vdb)
ASSERT0(zio->io_error);
ASSERT(db->db_blkptr == bp);
- if (zio->io_flags & ZIO_FLAG_IO_REWRITE) {
+ /*
+ * For nopwrites and rewrites we ensure that the bp matches our
+ * original and bypass all the accounting.
+ */
+ if (zio->io_flags & (ZIO_FLAG_IO_REWRITE | ZIO_FLAG_NOPWRITE)) {
ASSERT(BP_EQUAL(bp, bp_orig));
} else {
objset_t *os;
ASSERT(db->db_dirtycnt > 0);
db->db_dirtycnt -= 1;
db->db_data_pending = NULL;
+
dbuf_rele_and_unlock(db, (void *)(uintptr_t)txg);
}
dbuf_write_done(zio, NULL, db);
}
+/* Issue I/O to commit a dirty buffer to disk. */
static void
dbuf_write(dbuf_dirty_record_t *dr, arc_buf_t *data, dmu_tx_t *tx)
{
}
if (parent != dn->dn_dbuf) {
+ /* Our parent is an indirect block. */
+ /* We have a dirty parent that has been scheduled for write. */
ASSERT(parent && parent->db_data_pending);
+ /* Our parent's buffer is one level closer to the dnode. */
ASSERT(db->db_level == parent->db_level-1);
+ /*
+ * We're about to modify our parent's db_data by modifying
+ * our block pointer, so the parent must be released.
+ */
ASSERT(arc_released(parent->db_buf));
zio = parent->db_data_pending->dr_zio;
} else {
+ /* Our parent is the dnode itself. */
ASSERT((db->db_level == dn->dn_phys->dn_nlevels-1 &&
db->db_blkid != DMU_SPILL_BLKID) ||
(db->db_blkid == DMU_SPILL_BLKID && db->db_level == 0));
ASSERT(db->db_state != DB_NOFILL);
dr->dr_zio = zio_write(zio, os->os_spa, txg,
db->db_blkptr, data->b_data, arc_buf_size(data), &zp,
- dbuf_write_override_ready, dbuf_write_override_done, dr,
- ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED, &zb);
+ dbuf_write_override_ready, NULL, dbuf_write_override_done,
+ dr, ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED, &zb);
mutex_enter(&db->db_mtx);
dr->dt.dl.dr_override_state = DR_NOT_OVERRIDDEN;
zio_write_override(dr->dr_zio, &dr->dt.dl.dr_overridden_by,
- dr->dt.dl.dr_copies);
+ dr->dt.dl.dr_copies, dr->dt.dl.dr_nopwrite);
mutex_exit(&db->db_mtx);
} else if (db->db_state == DB_NOFILL) {
ASSERT(zp.zp_checksum == ZIO_CHECKSUM_OFF);
dr->dr_zio = zio_write(zio, os->os_spa, txg,
db->db_blkptr, NULL, db->db.db_size, &zp,
- dbuf_write_nofill_ready, dbuf_write_nofill_done, db,
+ dbuf_write_nofill_ready, NULL, dbuf_write_nofill_done, db,
ZIO_PRIORITY_ASYNC_WRITE,
ZIO_FLAG_MUSTSUCCEED | ZIO_FLAG_NODATA, &zb);
} else {
dr->dr_zio = arc_write(zio, os->os_spa, txg,
db->db_blkptr, data, DBUF_IS_L2CACHEABLE(db),
DBUF_IS_L2COMPRESSIBLE(db), &zp, dbuf_write_ready,
- dbuf_write_done, db, ZIO_PRIORITY_ASYNC_WRITE,
- ZIO_FLAG_MUSTSUCCEED, &zb);
+ dbuf_write_physdone, dbuf_write_done, db,
+ ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED, &zb);
}
}