OpenZFS 7247 - zfs receive of deduplicated stream fails

Authored by: Chris Williamson <chris.williamson@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Reviewed-by: loli10K <ezomori.nozomu@gmail.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Ported-by: George Melikov <mail@gmelikov.ru>

OpenZFS-issue: https://www.illumos.org/issues/7247
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/2ad25b4
Closes #5689 

Porting notes:
- tests/zfs-tests/tests/functional/cli_root/zfs_receive/zfs_receive_013_pos.ksh
  renamed as zfs_receive_015_pos.ksh, zfs_receive_013_pos.ksh is now
  used for OpenZFS test.
- libzfs_sendrecv.c: SMALLEST_POSSIBLE_MAX_DDT_MB is always used
  for all 32-bit builds.

lib/libzfs/libzfs_sendrecv.c

@@ -245,12 +245,16 @@ cksummer(void *arg)
 	int outfd;
 	dedup_table_t ddt;
 	zio_cksum_t stream_cksum;
-	uint64_t physmem = sysconf(_SC_PHYS_PAGES) * sysconf(_SC_PAGESIZE);
 	uint64_t numbuckets;
 
+#ifdef _ILP32
+	ddt.max_ddt_size = SMALLEST_POSSIBLE_MAX_DDT_MB << 20;
+#else
+	uint64_t physmem = sysconf(_SC_PHYS_PAGES) * sysconf(_SC_PAGESIZE);
 	ddt.max_ddt_size =
 	    MAX((physmem * MAX_DDT_PHYSMEM_PERCENT) / 100,
 	    SMALLEST_POSSIBLE_MAX_DDT_MB << 20);
+#endif
 
 	numbuckets = ddt.max_ddt_size / (sizeof (dedup_entry_t));
 

module/zfs/dmu_send.c

@@ -3230,6 +3230,9 @@ dmu_recv_end_sync(void *arg, dmu_tx_t *tx)
 		dsl_dataset_phys(origin_head)->ds_flags &=
 		    ~DS_FLAG_INCONSISTENT;
 
+		drc->drc_newsnapobj =
+		    dsl_dataset_phys(origin_head)->ds_prev_snap_obj;
+
 		dsl_dataset_rele(origin_head, FTAG);
 		dsl_destroy_head_sync_impl(drc->drc_ds, tx);
 
@@ -3265,8 +3268,9 @@ dmu_recv_end_sync(void *arg, dmu_tx_t *tx)
 			(void) zap_remove(dp->dp_meta_objset, ds->ds_object,
 			    DS_FIELD_RESUME_TONAME, tx);
 		}
+		drc->drc_newsnapobj =
+		    dsl_dataset_phys(drc->drc_ds)->ds_prev_snap_obj;
 	}
-	drc->drc_newsnapobj = dsl_dataset_phys(drc->drc_ds)->ds_prev_snap_obj;
 	zvol_create_minors(dp->dp_spa, drc->drc_tofs, B_TRUE);
 	/*
 	 * Release the hold from dmu_recv_begin.  This must be done before
@@ -3310,8 +3314,6 @@ static int dmu_recv_end_modified_blocks = 3;
 static int
 dmu_recv_existing_end(dmu_recv_cookie_t *drc)
 {
-	int error;
-
 #ifdef _KERNEL
 	/*
 	 * We will be destroying the ds; make sure its origin is unmounted if
@@ -3322,23 +3324,30 @@ dmu_recv_existing_end(dmu_recv_cookie_t *drc)
 	zfs_destroy_unmount_origin(name);
 #endif
 
-	error = dsl_sync_task(drc->drc_tofs,
+	return (dsl_sync_task(drc->drc_tofs,
 	    dmu_recv_end_check, dmu_recv_end_sync, drc,
-	    dmu_recv_end_modified_blocks, ZFS_SPACE_CHECK_NORMAL);
+	    dmu_recv_end_modified_blocks, ZFS_SPACE_CHECK_NORMAL));
-
-	if (error != 0)
-		dmu_recv_cleanup_ds(drc);
-	return (error);
 }
 
 static int
 dmu_recv_new_end(dmu_recv_cookie_t *drc)
+{
+	return (dsl_sync_task(drc->drc_tofs,
+	    dmu_recv_end_check, dmu_recv_end_sync, drc,
+	    dmu_recv_end_modified_blocks, ZFS_SPACE_CHECK_NORMAL));
+}
+
+int
+dmu_recv_end(dmu_recv_cookie_t *drc, void *owner)
 {
 	int error;
 
-	error = dsl_sync_task(drc->drc_tofs,
+	drc->drc_owner = owner;
-	    dmu_recv_end_check, dmu_recv_end_sync, drc,
+
-	    dmu_recv_end_modified_blocks, ZFS_SPACE_CHECK_NORMAL);
+	if (drc->drc_newfs)
+		error = dmu_recv_new_end(drc);
+	else
+		error = dmu_recv_existing_end(drc);
 
 	if (error != 0) {
 		dmu_recv_cleanup_ds(drc);
@@ -3350,17 +3359,6 @@ dmu_recv_new_end(dmu_recv_cookie_t *drc)
 	return (error);
 }
 
-int
-dmu_recv_end(dmu_recv_cookie_t *drc, void *owner)
-{
-	drc->drc_owner = owner;
-
-	if (drc->drc_newfs)
-		return (dmu_recv_new_end(drc));
-	else
-		return (dmu_recv_existing_end(drc));
-}
-
 /*
  * Return TRUE if this objset is currently being received into.
  */

tests/runfiles/linux.run

@@ -150,7 +150,7 @@ tests = ['zfs_receive_001_pos', 'zfs_receive_002_pos', 'zfs_receive_003_pos',
     'zfs_receive_005_neg', 'zfs_receive_006_pos',
     'zfs_receive_007_neg', 'zfs_receive_008_pos', 'zfs_receive_009_neg',
     'zfs_receive_010_pos', 'zfs_receive_011_pos', 'zfs_receive_012_pos',
-    'zfs_receive_013_pos', 'zfs_receive_014_pos']
+    'zfs_receive_013_pos', 'zfs_receive_014_pos', 'zfs_receive_015_pos']
 
 # DISABLED:
 # zfs_rename_006_pos - https://github.com/zfsonlinux/zfs/issues/5647

tests/zfs-tests/tests/functional/cli_root/zfs_receive/Makefile.am

@@ -15,4 +15,5 @@ dist_pkgdata_SCRIPTS = \
 	zfs_receive_011_pos.ksh \
 	zfs_receive_012_pos.ksh \
 	zfs_receive_013_pos.ksh \
-	zfs_receive_014_pos.ksh
+	zfs_receive_014_pos.ksh \
+	zfs_receive_015_pos.ksh

tests/zfs-tests/tests/functional/cli_root/zfs_receive/zfs_receive_013_pos.ksh

@@ -15,69 +15,59 @@
 #
 
 #
-# Copyright 2016, loli10K. All rights reserved.
+# Copyright (c) 2015 by Delphix. All rights reserved.
 #
 
-. $STF_SUITE/include/libtest.shlib
+. $STF_SUITE/tests/functional/cli_root/cli_common.kshlib
-. $STF_SUITE/tests/functional/cli_root/zfs_set/zfs_set_common.kshlib
 
 #
 # DESCRIPTION:
-#	Verify ZFS can receive custom properties on both filesystems and
+#   Verifying 'zfs receive' works correctly on deduplicated streams
-#	snapshots from full and incremental streams.
 #
 # STRATEGY:
-#	1. Create a filesystem.
+#   1. Create some snapshots with duplicated data
-#	2. Snapshot the filesystem.
+#   2. Send a deduplicated stream of the last snapshot
-#	3. Set custom properties on both the fs and snapshots.
+#   3. Attempt to receive the deduplicated stream
-#	4. Create different send streams with the properties.
-#	5. Receive the send streams and verify the properties.
 #
 
-verify_runnable "both"
+src_fs=$TESTPOOL/drecvsrc
-
+temppool=recvtank
-typeset streamfile_full=/var/tmp/streamfile_full.$$
+dst_fs=$temppool/drecvdest
-typeset streamfile_incr=/var/tmp/streamfile_incr.$$
+streamfile=/var/tmp/drecvstream.$$
-orig=$TESTPOOL/$TESTFS1
+tpoolfile=/temptank.$$
-dest=$TESTPOOL/$TESTFS2
-typeset user_prop=$(valid_user_property 8)
-typeset value=$(user_property_value 8)
 
 function cleanup
 {
-	log_must $RM $streamfile_full
+    for fs in $src_fs $dst_fs; do
-	log_must $RM $streamfile_incr
+        datasetexists $fs && log_must $ZFS destroy -rf $fs
-	log_must $ZFS destroy -rf $TESTPOOL/$TESTFS1
+    done
-	log_must $ZFS destroy -rf $TESTPOOL/$TESTFS2
+    $ZPOOL destroy $temppool
+    [[ -f $streamfile ]] && log_must $RM -f $streamfile
+    [[ -f $tpoolfile ]] && log_must $RM -f $tpoolfile
 }
 
-log_assert "ZFS can receive custom properties."
+log_assert "Verifying 'zfs receive' works correctly on deduplicated streams"
 log_onexit cleanup
 
-#	1. Create a filesystem.
+truncate -s 100M $tpoolfile
-log_must $ZFS create $orig
+log_must $ZPOOL create $temppool $tpoolfile
+log_must $ZFS create $src_fs
+src_mnt=$(get_prop mountpoint $src_fs) || log_fail "get_prop mountpoint $src_fs"
 
-#	2. Snapshot the filesystem.
+echo blah > $src_mnt/blah
-log_must $ZFS snapshot $orig@snap1
+$ZFS snapshot $src_fs@base
-log_must $ZFS snapshot $orig@snap2
-log_must $ZFS snapshot $orig@snap3
 
-#	3. Set custom properties on both the fs and snapshots.
+echo grumble > $src_mnt/grumble
-log_must eval "$ZFS set '$user_prop'='$value' $orig"
+echo blah > $src_mnt/blah2
-log_must eval "$ZFS set '$user_prop:snap1'='$value:snap1' $orig@snap1"
+$ZFS snapshot $src_fs@snap2
-log_must eval "$ZFS set '$user_prop:snap2'='$value:snap2' $orig@snap2"
-log_must eval "$ZFS set '$user_prop:snap3'='$value:snap3' $orig@snap3"
 
-#	4. Create different send streams with the properties.
+echo grumble > $src_mnt/mumble
-log_must eval "$ZFS send -p $orig@snap1 > $streamfile_full"
+echo blah > $src_mnt/blah3
-log_must eval "$ZFS send -p -I $orig@snap1 $orig@snap3 > $streamfile_incr"
+$ZFS snapshot $src_fs@snap3
 
-#	5. Receive the send streams and verify the properties.
+log_must eval "$ZFS send -D -R $src_fs@snap3 > $streamfile"
-log_must eval "$ZFS recv $dest < $streamfile_full"
+log_must eval "$ZFS receive -v $dst_fs < $streamfile"
-log_must eval "check_user_prop $dest $user_prop '$value'"
-log_must eval "check_user_prop $dest@snap1 '$user_prop:snap1' '$value:snap1'"
-log_must eval "$ZFS recv $dest < $streamfile_incr"
-log_must eval "check_user_prop $dest@snap2 '$user_prop:snap2' '$value:snap2'"
-log_must eval "check_user_prop $dest@snap3 '$user_prop:snap3' '$value:snap3'"
 
-log_pass "ZFS can receive custom properties passed."
+cleanup
+
+log_pass "Verifying 'zfs receive' works correctly on deduplicated streams"

tests/zfs-tests/tests/functional/cli_root/zfs_receive/zfs_receive_015_pos.ksh

@@ -0,0 +1,83 @@
+#!/bin/ksh -p
+#
+# CDDL HEADER START
+#
+# This file and its contents are supplied under the terms of the
+# Common Development and Distribution License ("CDDL"), version 1.0.
+# You may only use this file in accordance with the terms of version
+# 1.0 of the CDDL.
+#
+# A full copy of the text of the CDDL should have accompanied this
+# source.  A copy of the CDDL is also available via the Internet at
+# http://www.illumos.org/license/CDDL.
+#
+# CDDL HEADER END
+#
+
+#
+# Copyright 2016, loli10K. All rights reserved.
+#
+
+. $STF_SUITE/include/libtest.shlib
+. $STF_SUITE/tests/functional/cli_root/zfs_set/zfs_set_common.kshlib
+
+#
+# DESCRIPTION:
+#	Verify ZFS can receive custom properties on both filesystems and
+#	snapshots from full and incremental streams.
+#
+# STRATEGY:
+#	1. Create a filesystem.
+#	2. Snapshot the filesystem.
+#	3. Set custom properties on both the fs and snapshots.
+#	4. Create different send streams with the properties.
+#	5. Receive the send streams and verify the properties.
+#
+
+verify_runnable "both"
+
+typeset streamfile_full=/var/tmp/streamfile_full.$$
+typeset streamfile_incr=/var/tmp/streamfile_incr.$$
+orig=$TESTPOOL/$TESTFS1
+dest=$TESTPOOL/$TESTFS2
+typeset user_prop=$(valid_user_property 8)
+typeset value=$(user_property_value 8)
+
+function cleanup
+{
+	log_must $RM $streamfile_full
+	log_must $RM $streamfile_incr
+	log_must $ZFS destroy -rf $TESTPOOL/$TESTFS1
+	log_must $ZFS destroy -rf $TESTPOOL/$TESTFS2
+}
+
+log_assert "ZFS can receive custom properties."
+log_onexit cleanup
+
+#	1. Create a filesystem.
+log_must $ZFS create $orig
+
+#	2. Snapshot the filesystem.
+log_must $ZFS snapshot $orig@snap1
+log_must $ZFS snapshot $orig@snap2
+log_must $ZFS snapshot $orig@snap3
+
+#	3. Set custom properties on both the fs and snapshots.
+log_must eval "$ZFS set '$user_prop'='$value' $orig"
+log_must eval "$ZFS set '$user_prop:snap1'='$value:snap1' $orig@snap1"
+log_must eval "$ZFS set '$user_prop:snap2'='$value:snap2' $orig@snap2"
+log_must eval "$ZFS set '$user_prop:snap3'='$value:snap3' $orig@snap3"
+
+#	4. Create different send streams with the properties.
+log_must eval "$ZFS send -p $orig@snap1 > $streamfile_full"
+log_must eval "$ZFS send -p -I $orig@snap1 $orig@snap3 > $streamfile_incr"
+
+#	5. Receive the send streams and verify the properties.
+log_must eval "$ZFS recv $dest < $streamfile_full"
+log_must eval "check_user_prop $dest $user_prop '$value'"
+log_must eval "check_user_prop $dest@snap1 '$user_prop:snap1' '$value:snap1'"
+log_must eval "$ZFS recv $dest < $streamfile_incr"
+log_must eval "check_user_prop $dest@snap2 '$user_prop:snap2' '$value:snap2'"
+log_must eval "check_user_prop $dest@snap3 '$user_prop:snap3' '$value:snap3'"
+
+log_pass "ZFS can receive custom properties passed."