Fix ENOSPC in "Handle zap_add() failures in ..."

Commit cc63068 caused ENOSPC error when copy a large amount of files
between two directories. The reason is that the patch limits zap leaf
expansion to 2 retries, and return ENOSPC when failed.

The intent for limiting retries is to prevent pointlessly growing table
to max size when adding a block full of entries with same name in
different case in mixed mode. However, it turns out we cannot use any
limit on the retry. When we copy files from one directory in readdir
order, we are copying in hash order, one leaf block at a time. Which
means that if the leaf block in source directory has expanded 6 times,
and you copy those entries in that block, by the time you need to expand
the leaf in destination directory, you need to expand it 6 times in one
go. So any limit on the retry will result in error where it shouldn't.

Note that while we do use different salt for different directories, it
seems that the salt/hash function doesn't provide enough randomization
to the hash distance to prevent this from happening.

Since cc63068 has already been reverted. This patch adds it back and
removes the retry limit.

Also, as it turn out, failing on zap_add() has a serious side effect for
mzap_upgrade(). When upgrading from micro zap to fat zap, it will
call zap_add() to transfer entries one at a time. If it hit any error
halfway through, the remaining entries will be lost, causing those files
to become orphan. This patch add a VERIFY to catch it.

Reviewed-by: Sanjeev Bagewadi <[email protected]>
Reviewed-by: Richard Yao <[email protected]>
Reviewed-by: Tony Hutter <[email protected]>
Reviewed-by: Albert Lee <[email protected]>
Reviewed-by: Brian Behlendorf <[email protected]>
Reviewed by: Matthew Ahrens <[email protected]>
Signed-off-by: Chunwei Chen <[email protected]>
Closes #7401 
Closes #7421 

9 files changed, 353 insertions, 0 deletions

diff --git a/tests/zfs-tests/tests/functional/Makefile.am b/tests/zfs-tests/tests/functional/Makefile.am
index a5bbb36e0..6388f2d0f 100644
--- a/tests/zfs-tests/tests/functional/Makefile.am
+++ b/tests/zfs-tests/tests/functional/Makefile.am
@@ -13,6 +13,7 @@ SUBDIRS = \
 	cli_root \
 	cli_user \
 	compression \
+	cp_files \
 	ctime \
 	deadman \
 	delegate \
diff --git a/tests/zfs-tests/tests/functional/casenorm/Makefile.am b/tests/zfs-tests/tests/functional/casenorm/Makefile.am
index 65dd156e7..b284a2560 100644
--- a/tests/zfs-tests/tests/functional/casenorm/Makefile.am
+++ b/tests/zfs-tests/tests/functional/casenorm/Makefile.am
@@ -7,6 +7,7 @@ dist_pkgdata_SCRIPTS = \
 	insensitive_formd_lookup.ksh \
 	insensitive_none_delete.ksh \
 	insensitive_none_lookup.ksh \
+	mixed_create_failure.ksh \
 	mixed_formd_delete.ksh \
 	mixed_formd_lookup_ci.ksh \
 	mixed_formd_lookup.ksh \
diff --git a/tests/zfs-tests/tests/functional/casenorm/mixed_create_failure.ksh b/tests/zfs-tests/tests/functional/casenorm/mixed_create_failure.ksh
new file mode 100755
index 000000000..51b5bb3f6
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/casenorm/mixed_create_failure.ksh
@@ -0,0 +1,136 @@
+#!/bin/ksh -p
+#
+#
+# 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.
+#
+#
+# Copyright 2018 Nutanix Inc.  All rights reserved.
+#
+
+. $STF_SUITE/tests/functional/casenorm/casenorm.kshlib
+
+# DESCRIPTION:
+# For the filesystem with casesensitivity=mixed, normalization=none,
+# when multiple files with the same name (differing only in case) are created,
+# the number of files is limited to what can fit in a fatzap leaf-block.
+# And beyond that, it fails with ENOSPC.
+#
+# Ensure that the create/rename operations fail gracefully and not trigger an
+# ASSERT.
+#
+# STRATEGY:
+# Repeat the below steps for objects: files, directories, symlinks and hardlinks
+# 1. Create objects with same name but varying in case.
+#    E.g. 'abcdefghijklmnop', 'Abcdefghijklmnop', 'ABcdefghijklmnop' etc.
+#    The create should fail with ENOSPC.
+# 2. Create an object with name 'tmp_obj' and try to rename it to name that we
+#    failed to add in step 1 above.
+#    This should fail as well.
+
+verify_runnable "global"
+
+function cleanup
+{
+        destroy_testfs
+}
+
+log_onexit cleanup
+log_assert "With mixed mode: ensure create fails with ENOSPC beyond a certain limit"
+
+create_testfs "-o casesensitivity=mixed -o normalization=none"
+
+# Different object types
+obj_type=('file' 'dir' 'symlink' 'hardlink')
+
+# Commands to create different object types
+typeset -A ops
+ops['file']='touch'
+ops['dir']='mkdir'
+ops['symlink']='ln -s'
+ops['hardlink']='ln'
+
+# This function tests the following for a give object type :
+# - Create multiple objects with the same name (varying only in case).
+#   Ensure that it eventually fails once the leaf-block limit is exceeded.
+# - Create another object with a different name. And attempt rename it to the
+#   name (for which the create had failed in the previous step).
+#   This should fail as well.
+# Args :
+#   $1 - object type (file/dir/symlink/hardlink)
+#   $2 - test directory
+#
+function test_ops
+{
+	typeset obj_type=$1
+	typeset testdir=$2
+
+	target_obj='target-file'
+
+	op="${ops[$obj_type]}"
+
+	log_note "The op : $op"
+	log_note "testdir=$testdir obj_type=$obj_type"
+
+	test_path="$testdir/$obj_type"
+	mkdir $test_path
+	log_note "Created test dir $test_path"
+
+	if [[ $obj_type = "symlink" || $obj_type = "hardlink" ]]; then
+		touch $test_path/$target_obj
+		log_note "Created target: $test_path/$target_obj"
+		op="$op $test_path/$target_obj"
+	fi
+
+	log_note "op : $op"
+	names='{a,A}{b,B}{c,C}{d,D}{e,E}{f,F}{g,G}{h,H}{i,I}{j,J}{k,K}{l,L}'
+	for name in $names; do
+		cmd="$op $test_path/$name"
+		out=$($cmd 2>&1)
+		ret=$?
+		log_note "cmd: $cmd ret: $ret out=$out"
+		if (($ret != 0)); then
+			if [[ $out = *@(No space left on device)* ]]; then
+				save_name="$test_path/$name"
+				break;
+			else
+				log_err "$cmd failed with unexpected error : $out"
+			fi
+		fi
+	done
+
+	log_note 'Test rename \"sample_name\" rename'
+	TMP_OBJ="$test_path/tmp_obj"
+	cmd="$op $TMP_OBJ"
+	out=$($cmd 2>&1)
+	ret=$?
+	if (($ret != 0)); then
+		log_err "cmd:$cmd failed out:$out"
+	fi
+
+	# Now, try to rename the tmp_obj to the name which we failed to add earlier.
+	# This should fail as well.
+	out=$(mv $TMP_OBJ $save_name 2>&1)
+	ret=$?
+	if (($ret != 0)); then
+		if [[ $out = *@(No space left on device)* ]]; then
+			log_note "$cmd failed as expected : $out"
+		else
+			log_err "$cmd failed with : $out"
+		fi
+	fi
+}
+
+for obj_type in ${obj_type[*]};
+do
+	log_note "Testing create of $obj_type"
+	test_ops $obj_type $TESTDIR
+done
+
+log_pass "Mixed mode FS: Ops on large number of colliding names fail gracefully"
diff --git a/tests/zfs-tests/tests/functional/cp_files/.gitignore b/tests/zfs-tests/tests/functional/cp_files/.gitignore
new file mode 100644
index 000000000..eac05e155
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cp_files/.gitignore
@@ -0,0 +1 @@
+/cp_files
diff --git a/tests/zfs-tests/tests/functional/cp_files/Makefile.am b/tests/zfs-tests/tests/functional/cp_files/Makefile.am
new file mode 100644
index 000000000..06c31f5f3
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cp_files/Makefile.am
@@ -0,0 +1,13 @@
+include $(top_srcdir)/config/Rules.am
+
+pkgdatadir = $(datadir)/@PACKAGE@/zfs-tests/tests/functional/cp_files
+
+dist_pkgdata_SCRIPTS = \
+	cp_files_001_pos.ksh \
+	cleanup.ksh \
+	setup.ksh
+
+pkgexecdir = $(datadir)/@PACKAGE@/zfs-tests/tests/functional/cp_files
+
+pkgexec_PROGRAMS = cp_files
+cp_files_SOURCES= cp_files.c
diff --git a/tests/zfs-tests/tests/functional/cp_files/cleanup.ksh b/tests/zfs-tests/tests/functional/cp_files/cleanup.ksh
new file mode 100755
index 000000000..3166bd6ec
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cp_files/cleanup.ksh
@@ -0,0 +1,34 @@
+#!/bin/ksh -p
+#
+# 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 2007 Sun Microsystems, Inc.  All rights reserved.
+# Use is subject to license terms.
+#
+
+#
+# Copyright (c) 2013 by Delphix. All rights reserved.
+#
+
+. $STF_SUITE/include/libtest.shlib
+
+default_cleanup
diff --git a/tests/zfs-tests/tests/functional/cp_files/cp_files.c b/tests/zfs-tests/tests/functional/cp_files/cp_files.c
new file mode 100644
index 000000000..9af64a112
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cp_files/cp_files.c
@@ -0,0 +1,58 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include <dirent.h>
+#include <errno.h>
+#include <string.h>
+
+int
+main(int argc, char *argv[])
+{
+	int tfd;
+	DIR *sdir;
+	struct dirent *dirent;
+
+	if (argc != 3) {
+		fprintf(stderr, "Usage: %s SRC DST\n", argv[0]);
+		exit(1);
+	}
+
+	sdir = opendir(argv[1]);
+	if (sdir == NULL) {
+		fprintf(stderr, "Failed to open %s: %s\n",
+		    argv[1], strerror(errno));
+		exit(2);
+	}
+
+	tfd = open(argv[2], O_DIRECTORY);
+	if (tfd < 0) {
+		fprintf(stderr, "Failed to open %s: %s\n",
+		    argv[2], strerror(errno));
+		closedir(sdir);
+		exit(3);
+	}
+
+	while ((dirent = readdir(sdir)) != NULL) {
+		if (dirent->d_name[0] == '.' &&
+		    (dirent->d_name[1] == '.' || dirent->d_name[1] == '\0'))
+			continue;
+
+		int fd = openat(tfd, dirent->d_name, O_CREAT|O_WRONLY, 0666);
+		if (fd < 0) {
+			fprintf(stderr, "Failed to create %s/%s: %s\n",
+			    argv[2], dirent->d_name, strerror(errno));
+			closedir(sdir);
+			close(tfd);
+			exit(4);
+		}
+		close(fd);
+	}
+
+	closedir(sdir);
+	close(tfd);
+
+	return (0);
+}
diff --git a/tests/zfs-tests/tests/functional/cp_files/cp_files_001_pos.ksh b/tests/zfs-tests/tests/functional/cp_files/cp_files_001_pos.ksh
new file mode 100755
index 000000000..3e138cfc9
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cp_files/cp_files_001_pos.ksh
@@ -0,0 +1,74 @@
+#! /bin/ksh -p
+#
+# 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 (c) 2018 by Nutanix. All rights reserved.
+#
+
+. $STF_SUITE/include/libtest.shlib
+
+#
+# DESCRIPTION:
+# Copy a large number of files between 2 directories
+# within a zfs filesystem works without errors.
+# This make sure zap upgrading and expanding works.
+#
+# STRATEGY:
+#
+# 1. Create NR_FILES files in directory src
+# 2. Check the number of files is correct
+# 3. Copy files from src to dst in readdir order
+# 4. Check the number of files is correct
+#
+
+verify_runnable "global"
+
+function cleanup
+{
+	rm -rf $TESTDIR/src $TESTDIR/dst
+}
+
+log_assert "Copy a large number of files between 2 directories" \
+	"within a zfs filesystem works without errors"
+
+log_onexit cleanup
+
+NR_FILES=60000
+BATCH=1000
+
+log_must mkdir $TESTDIR/src
+log_must mkdir $TESTDIR/dst
+
+WD=$(pwd)
+cd $TESTDIR/src
+# create NR_FILES in BATCH at a time to prevent overflowing argument buffer
+for i in $(seq $(($NR_FILES/$BATCH))); do touch $(seq $((($i-1)*$BATCH+1)) $(($i*$BATCH))); done
+cd $WD
+
+log_must test $NR_FILES -eq $(ls -U $TESTDIR/src | wc -l)
+
+# copy files from src to dst, use cp_files to make sure we copy in readdir order
+log_must $STF_SUITE/tests/functional/cp_files/cp_files $TESTDIR/src $TESTDIR/dst
+
+log_must test $NR_FILES -eq $(ls -U $TESTDIR/dst | wc -l)
+
+log_pass
diff --git a/tests/zfs-tests/tests/functional/cp_files/setup.ksh b/tests/zfs-tests/tests/functional/cp_files/setup.ksh
new file mode 100755
index 000000000..fc5cec306
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cp_files/setup.ksh
@@ -0,0 +1,35 @@
+#!/bin/ksh -p
+#
+# 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 2007 Sun Microsystems, Inc.  All rights reserved.
+# Use is subject to license terms.
+#
+
+#
+# Copyright (c) 2013 by Delphix. All rights reserved.
+#
+
+. $STF_SUITE/include/libtest.shlib
+
+DISK=${DISKS%% *}
+default_setup $DISK