Sync to HEAD@r272825.

6 files changed, 140 insertions, 75 deletions

diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/arc.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/arc.c
index 1a1d4d8..60d2ff9 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/arc.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/arc.c
@@ -5047,7 +5047,7 @@ l2arc_write_buffers(spa_t *spa, l2arc_dev_t *dev, uint64_t target_sz,
 		if (ab == NULL)
 			ARCSTAT_BUMP(arcstat_l2_write_buffer_list_null_iter);
 
-		headroom = target_sz * l2arc_headroom;
+		headroom = target_sz * l2arc_headroom * 2 / ARC_BUFC_NUMLISTS;
 		if (do_headroom_boost)
 			headroom = (headroom * l2arc_headroom_boost) / 100;
 
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu.c
index 4e67bf6..c86d3a7 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu.c
@@ -1021,8 +1021,8 @@ xuio_stat_wbuf_nocopy()
 }
 
 #ifdef _KERNEL
-int
-dmu_read_uio(objset_t *os, uint64_t object, uio_t *uio, uint64_t size)
+static int
+dmu_read_uio_dnode(dnode_t *dn, uio_t *uio, uint64_t size)
 {
 	dmu_buf_t **dbp;
 	int numbufs, i, err;
@@ -1032,8 +1032,8 @@ dmu_read_uio(objset_t *os, uint64_t object, uio_t *uio, uint64_t size)
 	 * NB: we could do this block-at-a-time, but it's nice
 	 * to be reading in parallel.
 	 */
-	err = dmu_buf_hold_array(os, object, uio->uio_loffset, size, TRUE, FTAG,
-	    &numbufs, &dbp);
+	err = dmu_buf_hold_array_by_dnode(dn, uio->uio_loffset, size,
+	    TRUE, FTAG, &numbufs, &dbp, 0);
 	if (err)
 		return (err);
 
@@ -1080,6 +1080,58 @@ dmu_read_uio(objset_t *os, uint64_t object, uio_t *uio, uint64_t size)
 	return (err);
 }
 
+/*
+ * Read 'size' bytes into the uio buffer.
+ * From object zdb->db_object.
+ * Starting at offset uio->uio_loffset.
+ *
+ * If the caller already has a dbuf in the target object
+ * (e.g. its bonus buffer), this routine is faster than dmu_read_uio(),
+ * because we don't have to find the dnode_t for the object.
+ */
+int
+dmu_read_uio_dbuf(dmu_buf_t *zdb, uio_t *uio, uint64_t size)
+{
+	dmu_buf_impl_t *db = (dmu_buf_impl_t *)zdb;
+	dnode_t *dn;
+	int err;
+
+	if (size == 0)
+		return (0);
+
+	DB_DNODE_ENTER(db);
+	dn = DB_DNODE(db);
+	err = dmu_read_uio_dnode(dn, uio, size);
+	DB_DNODE_EXIT(db);
+
+	return (err);
+}
+
+/*
+ * Read 'size' bytes into the uio buffer.
+ * From the specified object
+ * Starting at offset uio->uio_loffset.
+ */
+int
+dmu_read_uio(objset_t *os, uint64_t object, uio_t *uio, uint64_t size)
+{
+	dnode_t *dn;
+	int err;
+
+	if (size == 0)
+		return (0);
+
+	err = dnode_hold(os, object, FTAG, &dn);
+	if (err)
+		return (err);
+
+	err = dmu_read_uio_dnode(dn, uio, size);
+
+	dnode_rele(dn, FTAG);
+
+	return (err);
+}
+
 static int
 dmu_write_uio_dnode(dnode_t *dn, uio_t *uio, uint64_t size, dmu_tx_t *tx)
 {
@@ -1132,6 +1184,15 @@ dmu_write_uio_dnode(dnode_t *dn, uio_t *uio, uint64_t size, dmu_tx_t *tx)
 	return (err);
 }
 
+/*
+ * Write 'size' bytes from the uio buffer.
+ * To object zdb->db_object.
+ * Starting at offset uio->uio_loffset.
+ *
+ * If the caller already has a dbuf in the target object
+ * (e.g. its bonus buffer), this routine is faster than dmu_write_uio(),
+ * because we don't have to find the dnode_t for the object.
+ */
 int
 dmu_write_uio_dbuf(dmu_buf_t *zdb, uio_t *uio, uint64_t size,
     dmu_tx_t *tx)
@@ -1151,6 +1212,11 @@ dmu_write_uio_dbuf(dmu_buf_t *zdb, uio_t *uio, uint64_t size,
 	return (err);
 }
 
+/*
+ * Write 'size' bytes from the uio buffer.
+ * To the specified object.
+ * Starting at offset uio->uio_loffset.
+ */
 int
 dmu_write_uio(objset_t *os, uint64_t object, uio_t *uio, uint64_t size,
     dmu_tx_t *tx)
@@ -1852,6 +1918,7 @@ dmu_object_info_from_dnode(dnode_t *dn, dmu_object_info_t *doi)
 	doi->doi_indirection = dn->dn_nlevels;
 	doi->doi_checksum = dn->dn_checksum;
 	doi->doi_compress = dn->dn_compress;
+	doi->doi_nblkptr = dn->dn_nblkptr;
 	doi->doi_physical_blocks_512 = (DN_USED_BYTES(dnp) + 256) >> 9;
 	doi->doi_max_offset = (dn->dn_maxblkid + 1) * dn->dn_datablksz;
 	doi->doi_fill_count = 0;
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_object.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_object.c
index 5b512ad..808864a 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_object.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_object.c
@@ -20,7 +20,8 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2013 by Delphix. All rights reserved.
+ * Copyright (c) 2013, 2014 by Delphix. All rights reserved.
+ * Copyright 2014 HybridCluster. All rights reserved.
  */
 
 #include <sys/dmu.h>
@@ -107,11 +108,9 @@ dmu_object_claim(objset_t *os, uint64_t object, dmu_object_type_t ot,
 
 int
 dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot,
-    int blocksize, dmu_object_type_t bonustype, int bonuslen)
+    int blocksize, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx)
 {
 	dnode_t *dn;
-	dmu_tx_t *tx;
-	int nblkptr;
 	int err;
 
 	if (object == DMU_META_DNODE_OBJECT)
@@ -122,44 +121,9 @@ dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot,
 	if (err)
 		return (err);
 
-	if (dn->dn_type == ot && dn->dn_datablksz == blocksize &&
-	    dn->dn_bonustype == bonustype && dn->dn_bonuslen == bonuslen) {
-		/* nothing is changing, this is a noop */
-		dnode_rele(dn, FTAG);
-		return (0);
-	}
-
-	if (bonustype == DMU_OT_SA) {
-		nblkptr = 1;
-	} else {
-		nblkptr = 1 + ((DN_MAX_BONUSLEN - bonuslen) >> SPA_BLKPTRSHIFT);
-	}
-
-	/*
-	 * If we are losing blkptrs or changing the block size this must
-	 * be a new file instance.   We must clear out the previous file
-	 * contents before we can change this type of metadata in the dnode.
-	 */
-	if (dn->dn_nblkptr > nblkptr || dn->dn_datablksz != blocksize) {
-		err = dmu_free_long_range(os, object, 0, DMU_OBJECT_END);
-		if (err)
-			goto out;
-	}
-
-	tx = dmu_tx_create(os);
-	dmu_tx_hold_bonus(tx, object);
-	err = dmu_tx_assign(tx, TXG_WAIT);
-	if (err) {
-		dmu_tx_abort(tx);
-		goto out;
-	}
-
 	dnode_reallocate(dn, ot, blocksize, bonustype, bonuslen, tx);
 
-	dmu_tx_commit(tx);
-out:
 	dnode_rele(dn, FTAG);
-
 	return (err);
 }
 
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_send.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_send.c
index 314e73c..1a0cab5 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_send.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_send.c
@@ -24,6 +24,7 @@
  * Copyright (c) 2011, 2014 by Delphix. All rights reserved.
  * Copyright (c) 2014, Joyent, Inc. All rights reserved.
  * Copyright (c) 2012, Martin Matuska <mm@FreeBSD.org>. All rights reserved.
+ * Copyright 2014 HybridCluster. All rights reserved.
  */
 
 #include <sys/dmu.h>
@@ -1392,12 +1393,25 @@ backup_byteswap(dmu_replay_record_t *drr)
 #undef DO32
 }
 
+static inline uint8_t
+deduce_nblkptr(dmu_object_type_t bonus_type, uint64_t bonus_size)
+{
+	if (bonus_type == DMU_OT_SA) {
+		return (1);
+	} else {
+		return (1 +
+		    ((DN_MAX_BONUSLEN - bonus_size) >> SPA_BLKPTRSHIFT));
+	}
+}
+
 static int
 restore_object(struct restorearg *ra, objset_t *os, struct drr_object *drro)
 {
-	int err;
+	dmu_object_info_t doi;
 	dmu_tx_t *tx;
 	void *data = NULL;
+	uint64_t object;
+	int err;
 
 	if (drro->drr_type == DMU_OT_NONE ||
 	    !DMU_OT_IS_VALID(drro->drr_type) ||
@@ -1411,10 +1425,11 @@ restore_object(struct restorearg *ra, objset_t *os, struct drr_object *drro)
 		return (SET_ERROR(EINVAL));
 	}
 
-	err = dmu_object_info(os, drro->drr_object, NULL);
+	err = dmu_object_info(os, drro->drr_object, &doi);
 
 	if (err != 0 && err != ENOENT)
 		return (SET_ERROR(EINVAL));
+	object = err == 0 ? drro->drr_object : DMU_NEW_OBJECT;
 
 	if (drro->drr_bonuslen) {
 		data = restore_read(ra, P2ROUNDUP(drro->drr_bonuslen, 8), NULL);
@@ -1422,37 +1437,53 @@ restore_object(struct restorearg *ra, objset_t *os, struct drr_object *drro)
 			return (ra->err);
 	}
 
-	if (err == ENOENT) {
-		/* currently free, want to be allocated */
-		tx = dmu_tx_create(os);
-		dmu_tx_hold_bonus(tx, DMU_NEW_OBJECT);
-		err = dmu_tx_assign(tx, TXG_WAIT);
-		if (err != 0) {
-			dmu_tx_abort(tx);
-			return (err);
+	/*
+	 * If we are losing blkptrs or changing the block size this must
+	 * be a new file instance.  We must clear out the previous file
+	 * contents before we can change this type of metadata in the dnode.
+	 */
+	if (err == 0) {
+		int nblkptr;
+
+		nblkptr = deduce_nblkptr(drro->drr_bonustype,
+		    drro->drr_bonuslen);
+
+		if (drro->drr_blksz != doi.doi_data_block_size ||
+		    nblkptr < doi.doi_nblkptr) {
+			err = dmu_free_long_range(os, drro->drr_object,
+			    0, DMU_OBJECT_END);
+			if (err != 0)
+				return (SET_ERROR(EINVAL));
 		}
+	}
+
+	tx = dmu_tx_create(os);
+	dmu_tx_hold_bonus(tx, object);
+	err = dmu_tx_assign(tx, TXG_WAIT);
+	if (err != 0) {
+		dmu_tx_abort(tx);
+		return (err);
+	}
+
+	if (object == DMU_NEW_OBJECT) {
+		/* currently free, want to be allocated */
 		err = dmu_object_claim(os, drro->drr_object,
 		    drro->drr_type, drro->drr_blksz,
 		    drro->drr_bonustype, drro->drr_bonuslen, tx);
-		dmu_tx_commit(tx);
-	} else {
-		/* currently allocated, want to be allocated */
+	} else if (drro->drr_type != doi.doi_type ||
+	    drro->drr_blksz != doi.doi_data_block_size ||
+	    drro->drr_bonustype != doi.doi_bonus_type ||
+	    drro->drr_bonuslen != doi.doi_bonus_size) {
+		/* currently allocated, but with different properties */
 		err = dmu_object_reclaim(os, drro->drr_object,
 		    drro->drr_type, drro->drr_blksz,
-		    drro->drr_bonustype, drro->drr_bonuslen);
+		    drro->drr_bonustype, drro->drr_bonuslen, tx);
 	}
 	if (err != 0) {
+		dmu_tx_commit(tx);
 		return (SET_ERROR(EINVAL));
 	}
 
-	tx = dmu_tx_create(os);
-	dmu_tx_hold_bonus(tx, drro->drr_object);
-	err = dmu_tx_assign(tx, TXG_WAIT);
-	if (err != 0) {
-		dmu_tx_abort(tx);
-		return (err);
-	}
-
 	dmu_object_set_checksum(os, drro->drr_object, drro->drr_checksumtype,
 	    tx);
 	dmu_object_set_compress(os, drro->drr_object, drro->drr_compress, tx);
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dmu.h b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dmu.h
index 2c0b5ae..ad19266 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dmu.h
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dmu.h
@@ -25,6 +25,7 @@
  * 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.
  */
 
 /* Portions Copyright 2010 Robert Milkowski */
@@ -341,7 +342,7 @@ uint64_t dmu_object_alloc(objset_t *os, dmu_object_type_t ot,
 int dmu_object_claim(objset_t *os, uint64_t object, dmu_object_type_t ot,
     int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx);
 int dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot,
-    int blocksize, dmu_object_type_t bonustype, int bonuslen);
+    int blocksize, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *txp);
 
 /*
  * Free an object from this objset.
@@ -616,6 +617,7 @@ void dmu_write(objset_t *os, uint64_t object, uint64_t offset, uint64_t size,
 void dmu_prealloc(objset_t *os, uint64_t object, uint64_t offset, uint64_t size,
 	dmu_tx_t *tx);
 int dmu_read_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size);
+int dmu_read_uio_dbuf(dmu_buf_t *zdb, struct uio *uio, uint64_t size);
 int dmu_write_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size,
     dmu_tx_t *tx);
 int dmu_write_uio_dbuf(dmu_buf_t *zdb, struct uio *uio, uint64_t size,
@@ -661,7 +663,8 @@ typedef struct dmu_object_info {
 	uint8_t doi_indirection;		/* 2 = dnode->indirect->data */
 	uint8_t doi_checksum;
 	uint8_t doi_compress;
-	uint8_t doi_pad[5];
+	uint8_t doi_nblkptr;
+	uint8_t doi_pad[4];
 	uint64_t doi_physical_blocks_512;	/* data + metadata, 512b blks */
 	uint64_t doi_max_offset;
 	uint64_t doi_fill_count;		/* number of non-empty blocks */
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c
index 109a339..5fec709 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c
@@ -582,7 +582,6 @@ static int
 mappedread(vnode_t *vp, int nbytes, uio_t *uio)
 {
 	znode_t *zp = VTOZ(vp);
-	objset_t *os = zp->z_zfsvfs->z_os;
 	vm_object_t obj;
 	int64_t start;
 	caddr_t va;
@@ -613,7 +612,8 @@ mappedread(vnode_t *vp, int nbytes, uio_t *uio)
 			page_unhold(pp);
 		} else {
 			zfs_vmobject_wunlock(obj);
-			error = dmu_read_uio(os, zp->z_id, uio, bytes);
+			error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl),
+			    uio, bytes);
 			zfs_vmobject_wlock(obj);
 		}
 		len -= bytes;
@@ -650,7 +650,6 @@ zfs_read(vnode_t *vp, uio_t *uio, int ioflag, cred_t *cr, caller_context_t *ct)
 {
 	znode_t		*zp = VTOZ(vp);
 	zfsvfs_t	*zfsvfs = zp->z_zfsvfs;
-	objset_t	*os;
 	ssize_t		n, nbytes;
 	int		error = 0;
 	rl_t		*rl;
@@ -658,7 +657,6 @@ zfs_read(vnode_t *vp, uio_t *uio, int ioflag, cred_t *cr, caller_context_t *ct)
 
 	ZFS_ENTER(zfsvfs);
 	ZFS_VERIFY_ZP(zp);
-	os = zfsvfs->z_os;
 
 	if (zp->z_pflags & ZFS_AV_QUARANTINED) {
 		ZFS_EXIT(zfsvfs);
@@ -756,10 +754,12 @@ zfs_read(vnode_t *vp, uio_t *uio, int ioflag, cred_t *cr, caller_context_t *ct)
 			error = mappedread_sf(vp, nbytes, uio);
 		else
 #endif /* __FreeBSD__ */
-		if (vn_has_cached_data(vp))
+		if (vn_has_cached_data(vp)) {
 			error = mappedread(vp, nbytes, uio);
-		else
-			error = dmu_read_uio(os, zp->z_id, uio, nbytes);
+		} else {
+			error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl),
+			    uio, nbytes);
+		}
 		if (error) {
 			/* convert checksum errors into IO errors */
 			if (error == ECKSUM)