Tag zfs-0.7.13

META file and changelog updated.

Signed-off-by: Tony Hutter <hutter2@llnl.gov>

.github/CONTRIBUTING.md

@@ -161,7 +161,7 @@ coding convention.
 ### Commit Message Formats
 #### New Changes
 Commit messages for new changes must meet the following guidelines:
-* In 50 characters or less, provide a summary of the change as the
+* In 72 characters or less, provide a summary of the change as the
 first line in the commit message.
 * A body which provides a description of the change. If necessary,
 please summarize important information such as why the proposed

META

@@ -1,7 +1,7 @@
 Meta:         1
 Name:         zfs
 Branch:       1.0
-Version:      0.7.12
+Version:      0.7.13
 Release:      1
 Release-Tags: relext
 License:      CDDL

cmd/mount_zfs/Makefile.am

@@ -7,6 +7,8 @@ DEFAULT_INCLUDES += \
 #
 # Ignore the prefix for the mount helper.  It must be installed in /sbin/
 # because this path is hardcoded in the mount(8) for security reasons.
+# However, if needed, the configure option --with-mounthelperdir= can be used
+# to override the default install location.
 #
 sbindir=$(mounthelperdir)
 sbin_PROGRAMS = mount.zfs

cmd/vdev_id/vdev_id

@@ -100,10 +100,11 @@ usage() {
 	cat << EOF
 Usage: vdev_id [-h]
        vdev_id <-d device> [-c config_file] [-p phys_per_port]
-               [-g sas_direct|sas_switch] [-m]
+               [-g sas_direct|sas_switch|scsi] [-m]
 
   -c    specify name of alernate config file [default=$CONFIG]
   -d    specify basename of device (i.e. sda)
+  -e    Create enclose device symlinks only (/dev/by-enclosure)
   -g    Storage network topology [default="$TOPOLOGY"]
   -m    Run in multipath mode
   -p    number of phy's per switch port [default=$PHYS_PER_PORT]
@@ -135,7 +136,7 @@ map_channel() {
 		MAPPED_CHAN=`awk "\\$1 == \"channel\" && \\$2 == ${PORT} \
 			{ print \\$3; exit }" $CONFIG`
 		;;
-		"sas_direct")
+		"sas_direct"|"scsi")
 		MAPPED_CHAN=`awk "\\$1 == \"channel\" && \
 			\\$2 == \"${PCI_ID}\" && \\$3 == ${PORT} \
 			{ print \\$4; exit }" $CONFIG`
@@ -276,6 +277,23 @@ sas_handler() {
 		d=$(eval echo \${$i})
 		SLOT=`echo $d | sed -e 's/^.*://'`
 		;;
+	"ses")
+		# look for this SAS path in all SCSI Enclosure Services
+		# (SES) enclosures
+		sas_address=`cat $end_device_dir/sas_address 2>/dev/null`
+		enclosures=`lsscsi -g | \
+			sed -n -e '/enclosu/s/^.* \([^ ][^ ]*\) *$/\1/p'`
+		for enclosure in $enclosures; do
+			set -- $(sg_ses -p aes $enclosure | \
+				awk "/device slot number:/{slot=\$12} \
+					/SAS address: $sas_address/\
+					{print slot}")
+			SLOT=$1
+			if [ -n "$SLOT" ] ; then
+				break
+			fi
+		done
+		;;
 	esac
 	if [ -z "$SLOT" ] ; then
 		return
@@ -289,6 +307,156 @@ sas_handler() {
 	echo ${CHAN}${SLOT}${PART}
 }
 
+scsi_handler() {
+	if [ -z "$FIRST_BAY_NUMBER" ] ; then
+		FIRST_BAY_NUMBER=`awk "\\$1 == \"first_bay_number\" \
+			{print \\$2; exit}" $CONFIG`
+	fi
+	FIRST_BAY_NUMBER=${FIRST_BAY_NUMBER:-0}
+
+	if [ -z "$PHYS_PER_PORT" ] ; then
+		PHYS_PER_PORT=`awk "\\$1 == \"phys_per_port\" \
+			{print \\$2; exit}" $CONFIG`
+	fi
+	PHYS_PER_PORT=${PHYS_PER_PORT:-4}
+	if ! echo $PHYS_PER_PORT | grep -q -E '^[0-9]+$' ; then
+		echo "Error: phys_per_port value $PHYS_PER_PORT is non-numeric"
+		exit 1
+	fi
+
+	if [ -z "$MULTIPATH_MODE" ] ; then
+		MULTIPATH_MODE=`awk "\\$1 == \"multipath\" \
+			{print \\$2; exit}" $CONFIG`
+	fi
+
+	# Use first running component device if we're handling a dm-mpath device
+	if [ "$MULTIPATH_MODE" = "yes" ] ; then
+		# If udev didn't tell us the UUID via DM_NAME, check /dev/mapper
+		if [ -z "$DM_NAME" ] ; then
+			DM_NAME=`ls -l --full-time /dev/mapper |
+				awk "/\/$DEV$/{print \\$9}"`
+		fi
+
+		# For raw disks udev exports DEVTYPE=partition when
+		# handling partitions, and the rules can be written to
+		# take advantage of this to append a -part suffix.  For
+		# dm devices we get DEVTYPE=disk even for partitions so
+		# we have to append the -part suffix directly in the
+		# helper.
+		if [ "$DEVTYPE" != "partition" ] ; then
+			PART=`echo $DM_NAME | awk -Fp '/p/{print "-part"$2}'`
+		fi
+
+		# Strip off partition information.
+		DM_NAME=`echo $DM_NAME | sed 's/p[0-9][0-9]*$//'`
+		if [ -z "$DM_NAME" ] ; then
+			return
+		fi
+
+		# Get the raw scsi device name from multipath -ll. Strip off
+		# leading pipe symbols to make field numbering consistent.
+		DEV=`multipath -ll $DM_NAME |
+			awk '/running/{gsub("^[|]"," "); print $3 ; exit}'`
+		if [ -z "$DEV" ] ; then
+			return
+		fi
+	fi
+
+	if echo $DEV | grep -q ^/devices/ ; then
+		sys_path=$DEV
+	else
+		sys_path=`udevadm info -q path -p /sys/block/$DEV 2>/dev/null`
+	fi
+
+	# expect sys_path like this, for example:
+	# /devices/pci0000:00/0000:00:0b.0/0000:09:00.0/0000:0a:05.0/0000:0c:00.0/host3/target3:1:0/3:1:0:21/block/sdv
+
+	# Use positional parameters as an ad-hoc array
+	set -- $(echo "$sys_path" | tr / ' ')
+	num_dirs=$#
+	scsi_host_dir="/sys"
+
+	# Get path up to /sys/.../hostX
+	i=1
+	while [ $i -le $num_dirs ] ; do
+		d=$(eval echo \${$i})
+		scsi_host_dir="$scsi_host_dir/$d"
+		echo $d | grep -q -E '^host[0-9]+$' && break
+		i=$(($i + 1))
+	done
+
+	if [ $i = $num_dirs ] ; then
+		return
+	fi
+
+	PCI_ID=$(eval echo \${$(($i -1))} | awk -F: '{print $2":"$3}')
+
+	# In scsi mode, the directory two levels beneath
+	# /sys/.../hostX reveals the port and slot.
+	port_dir=$scsi_host_dir
+	j=$(($i + 2))
+
+	i=$(($i + 1))
+	while [ $i -le $j ] ; do
+		port_dir="$port_dir/$(eval echo \${$i})"
+		i=$(($i + 1))
+	done
+
+	set -- $(echo $port_dir | sed -e 's/^.*:\([^:]*\):\([^:]*\)$/\1 \2/')
+	PORT=$1
+	SLOT=$(($2 + $FIRST_BAY_NUMBER))
+
+	if [ -z "$SLOT" ] ; then
+		return
+	fi
+
+	CHAN=`map_channel $PCI_ID $PORT`
+	SLOT=`map_slot $SLOT $CHAN`
+	if [ -z "$CHAN" ] ; then
+		return
+	fi
+	echo ${CHAN}${SLOT}${PART}
+}
+
+# Figure out the name for the enclosure symlink
+enclosure_handler () {
+	# We get all the info we need from udev's DEVPATH variable:
+	#
+	# DEVPATH=/sys/devices/pci0000:00/0000:00:03.0/0000:05:00.0/host0/subsystem/devices/0:0:0:0/scsi_generic/sg0
+
+	# Get the enclosure ID ("0:0:0:0")
+	ENC=$(basename $(readlink -m "/sys/$DEVPATH/../.."))
+	if [ ! -d /sys/class/enclosure/$ENC ] ; then
+		# Not an enclosure, bail out
+		return
+	fi
+
+	# Get the long sysfs device path to our enclosure. Looks like:
+	# /devices/pci0000:00/0000:00:03.0/0000:05:00.0/host0/port-0:0/ ... /enclosure/0:0:0:0
+
+	ENC_DEVICE=$(readlink /sys/class/enclosure/$ENC)
+
+	# Grab the full path to the hosts port dir:
+	# /devices/pci0000:00/0000:00:03.0/0000:05:00.0/host0/port-0:0
+	PORT_DIR=$(echo $ENC_DEVICE | grep -Eo '.+host[0-9]+/port-[0-9]+:[0-9]+')
+
+	# Get the port number
+	PORT_ID=$(echo $PORT_DIR | grep -Eo "[0-9]+$")
+
+	# The PCI directory is two directories up from the port directory
+	# /sys/devices/pci0000:00/0000:00:03.0/0000:05:00.0
+	PCI_ID_LONG=$(basename $(readlink -m "/sys/$PORT_DIR/../.."))
+
+	# Strip down the PCI address from 0000:05:00.0 to 05:00.0
+	PCI_ID=$(echo "$PCI_ID_LONG" | sed -r 's/^[0-9]+://g')
+
+	# Name our device according to vdev_id.conf (like "L0" or "U1").
+	NAME=$(awk "/channel/{if (\$1 == \"channel\" && \$2 == \"$PCI_ID\" && \
+		\$3 == \"$PORT_ID\") {print \$4int(count[\$4])}; count[\$4]++}" $CONFIG)
+
+	echo "${NAME}"
+}
+
 alias_handler () {
 	# Special handling is needed to correctly append a -part suffix
 	# to partitions of device mapper devices.  The DEVTYPE attribute
@@ -344,7 +512,7 @@ alias_handler () {
 	done
 }
 
-while getopts 'c:d:g:mp:h' OPTION; do
+while getopts 'c:d:eg:mp:h' OPTION; do
 	case ${OPTION} in
 	c)
 		CONFIG=${OPTARG}
@@ -352,6 +520,16 @@ while getopts 'c:d:g:mp:h' OPTION; do
 	d)
 		DEV=${OPTARG}
 		;;
+	e)
+	# When udev sees a scsi_generic device, it calls this script with -e to
+	# create the enclosure device symlinks only.  We also need
+	# "enclosure_symlinks yes" set in vdev_id.config to actually create the
+	# symlink.
+	ENCLOSURE_MODE=$(awk '{if ($1 == "enclosure_symlinks") print $2}' $CONFIG)
+	if [ "$ENCLOSURE_MODE" != "yes" ] ; then
+		exit 0
+	fi
+		;;
 	g)
 		TOPOLOGY=$OPTARG
 		;;
@@ -371,7 +549,7 @@ if [ ! -r $CONFIG ] ; then
 	exit 0
 fi
 
-if [ -z "$DEV" ] ; then
+if [ -z "$DEV" -a -z "$ENCLOSURE_MODE" ] ; then
 	echo "Error: missing required option -d"
 	exit 1
 fi
@@ -384,16 +562,37 @@ if [ -z "$BAY" ] ; then
 	BAY=`awk "\\$1 == \"slot\" {print \\$2; exit}" $CONFIG`
 fi
 
+TOPOLOGY=${TOPOLOGY:-sas_direct}
+
+# Should we create /dev/by-enclosure symlinks?
+if [ "$ENCLOSURE_MODE" = "yes" -a "$TOPOLOGY" = "sas_direct" ] ; then
+	ID_ENCLOSURE=$(enclosure_handler)
+	if [ -z "$ID_ENCLOSURE" ] ; then
+		exit 0
+	fi
+
+	# Just create the symlinks to the enclosure devices and then exit.
+	ENCLOSURE_PREFIX=$(awk '/enclosure_symlinks_prefix/{print $2}' $CONFIG)
+	if [ -z "$ENCLOSURE_PREFIX" ] ; then
+		ENCLOSURE_PREFIX="enc"
+	fi
+	echo "ID_ENCLOSURE=$ID_ENCLOSURE"
+	echo "ID_ENCLOSURE_PATH=by-enclosure/$ENCLOSURE_PREFIX-$ID_ENCLOSURE"
+	exit 0
+fi
+
 # First check if an alias was defined for this device.
 ID_VDEV=`alias_handler`
 
 if [ -z "$ID_VDEV" ] ; then
 	BAY=${BAY:-bay}
-	TOPOLOGY=${TOPOLOGY:-sas_direct}
 	case $TOPOLOGY in
 		sas_direct|sas_switch)
 			ID_VDEV=`sas_handler`
 			;;
+		scsi)
+			ID_VDEV=`scsi_handler`
+			;;
 		*)
 			echo "Error: unknown topology $TOPOLOGY"
 			exit 1

cmd/ztest/ztest.c

@@ -171,8 +171,8 @@ typedef struct ztest_shared_opts {
 } ztest_shared_opts_t;
 
 static const ztest_shared_opts_t ztest_opts_defaults = {
-	.zo_pool = { 'z', 't', 'e', 's', 't', '\0' },
-	.zo_dir = { '/', 't', 'm', 'p', '\0' },
+	.zo_pool = "ztest",
+	.zo_dir = "/tmp",
 	.zo_alt_ztest = { '\0' },
 	.zo_alt_libpath = { '\0' },
 	.zo_vdevs = 5,

config/kernel-access-ok-type.m4

@@ -0,0 +1,21 @@
+dnl #
+dnl # Linux 5.0: access_ok() drops 'type' parameter:
+dnl #
+dnl # - access_ok(type, addr, size)
+dnl # + access_ok(addr, size)
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_ACCESS_OK_TYPE], [
+	AC_MSG_CHECKING([whether access_ok() has 'type' parameter])
+	ZFS_LINUX_TRY_COMPILE([
+		#include <linux/uaccess.h>
+	],[
+		const void __user __attribute__((unused)) *addr = (void *) 0xdeadbeef;
+		unsigned long __attribute__((unused)) size = 1;
+		int error __attribute__((unused)) = access_ok(0, addr, size);
+	],[
+		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_ACCESS_OK_TYPE, 1, [kernel has access_ok with 'type' parameter])
+	],[
+		AC_MSG_RESULT(no)
+	])
+])

config/kernel-bio_set_dev.m4

@@ -1,10 +1,10 @@
 dnl #
 dnl # Linux 4.14 API,
 dnl #
-dnl # The bio_set_dev() helper was introduced as part of the transition
+dnl # The bio_set_dev() helper macro was introduced as part of the transition
 dnl # to have struct gendisk in struct bio. 
 dnl #
-AC_DEFUN([ZFS_AC_KERNEL_BIO_SET_DEV], [
+AC_DEFUN([ZFS_AC_KERNEL_BIO_SET_DEV_MACRO], [
 	AC_MSG_CHECKING([whether bio_set_dev() exists])
 	ZFS_LINUX_TRY_COMPILE([
 		#include <linux/bio.h>
@@ -20,3 +20,34 @@ AC_DEFUN([ZFS_AC_KERNEL_BIO_SET_DEV], [
 		AC_MSG_RESULT(no)
 	])
 ])
+
+dnl #
+dnl # Linux 5.0 API,
+dnl #
+dnl # The bio_set_dev() helper macro was updated to internally depend on
+dnl # bio_associate_blkg() symbol which is exported GPL-only.
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_BIO_SET_DEV_GPL_ONLY], [
+	AC_MSG_CHECKING([whether bio_set_dev() is GPL-only])
+	ZFS_LINUX_TRY_COMPILE([
+		#include <linux/module.h>
+		#include <linux/bio.h>
+		#include <linux/fs.h>
+		MODULE_LICENSE("$ZFS_META_LICENSE");
+	],[
+		struct block_device *bdev = NULL;
+		struct bio *bio = NULL;
+		bio_set_dev(bio, bdev);
+	],[
+		AC_MSG_RESULT(no)
+	],[
+		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_BIO_SET_DEV_GPL_ONLY, 1,
+		    [bio_set_dev() GPL-only])
+	])
+])
+
+AC_DEFUN([ZFS_AC_KERNEL_BIO_SET_DEV], [
+	ZFS_AC_KERNEL_BIO_SET_DEV_MACRO
+	ZFS_AC_KERNEL_BIO_SET_DEV_GPL_ONLY
+])

config/kernel-fpu.m4

@@ -1,18 +1,41 @@
 dnl # 
-dnl # 4.2 API change
-dnl # asm/i387.h is replaced by asm/fpu/api.h
+dnl # Handle differences in kernel FPU code.
+dnl #
+dnl # Kernel
+dnl # 5.0:	All kernel fpu functions are GPL only, so we can't use them.
+dnl #		(nothing defined)
+dnl #
+dnl # 4.2:	Use __kernel_fpu_{begin,end}()
+dnl #		HAVE_UNDERSCORE_KERNEL_FPU & KERNEL_EXPORTS_X86_FPU
+dnl #
+dnl # Pre-4.2:	Use kernel_fpu_{begin,end}()
+dnl #		HAVE_KERNEL_FPU & KERNEL_EXPORTS_X86_FPU
 dnl #
 AC_DEFUN([ZFS_AC_KERNEL_FPU], [
-	AC_MSG_CHECKING([whether asm/fpu/api.h exists])
+	AC_MSG_CHECKING([which kernel_fpu function to use])
+	ZFS_LINUX_TRY_COMPILE([
+		#include <asm/i387.h>
+		#include <asm/xcr.h>
+	],[
+		kernel_fpu_begin();
+		kernel_fpu_end();
+	],[
+		AC_MSG_RESULT(kernel_fpu_*)
+		AC_DEFINE(HAVE_KERNEL_FPU, 1, [kernel has kernel_fpu_* functions])
+		AC_DEFINE(KERNEL_EXPORTS_X86_FPU, 1, [kernel exports FPU functions])
+	],[
 		ZFS_LINUX_TRY_COMPILE([
 			#include <linux/kernel.h>
 			#include <asm/fpu/api.h>
 		],[
 			__kernel_fpu_begin();
+			__kernel_fpu_end();
 		],[
-		AC_MSG_RESULT(yes)
-		AC_DEFINE(HAVE_FPU_API_H, 1, [kernel has <asm/fpu/api.h> interface])
+			AC_MSG_RESULT(__kernel_fpu_*)
+			AC_DEFINE(HAVE_UNDERSCORE_KERNEL_FPU, 1, [kernel has __kernel_fpu_* functions])
+			AC_DEFINE(KERNEL_EXPORTS_X86_FPU, 1, [kernel exports FPU functions])
 		],[
-		AC_MSG_RESULT(no)
+			AC_MSG_RESULT(not exported)
+		])
 	])
 ])

config/kernel-misc-minor.m4

@@ -0,0 +1,26 @@
+dnl #
+dnl # Determine an available miscellaneous minor number which can be used
+dnl # for the /dev/zfs device.  This is needed because kernel module
+dnl # auto-loading depends on registering a reserved non-conflicting minor
+dnl # number.  Start with a large known available unreserved minor and work
+dnl # our way down to lower value if a collision is detected.
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_MISC_MINOR], [
+	AC_MSG_CHECKING([for available /dev/zfs minor])
+
+	for i in $(seq 249 -1 200); do
+		if ! grep -q "^#define\s\+.*_MINOR\s\+.*$i" \
+		    ${LINUX}/include/linux/miscdevice.h; then
+			ZFS_DEVICE_MINOR="$i"
+			AC_MSG_RESULT($ZFS_DEVICE_MINOR)
+			AC_DEFINE_UNQUOTED([ZFS_DEVICE_MINOR],
+			    [$ZFS_DEVICE_MINOR], [/dev/zfs minor])
+			break
+		fi
+	done
+
+	AS_IF([ test -z "$ZFS_DEVICE_MINOR"], [
+		AC_MSG_ERROR([
+	*** No available misc minor numbers available for use.])
+	])
+])

config/kernel.m4

@@ -5,7 +5,9 @@ AC_DEFUN([ZFS_AC_CONFIG_KERNEL], [
 	ZFS_AC_KERNEL
 	ZFS_AC_SPL
 	ZFS_AC_QAT
+	ZFS_AC_KERNEL_ACCESS_OK_TYPE
 	ZFS_AC_TEST_MODULE
+	ZFS_AC_KERNEL_MISC_MINOR
 	ZFS_AC_KERNEL_OBJTOOL
 	ZFS_AC_KERNEL_CONFIG
 	ZFS_AC_KERNEL_DECLARE_EVENT_CLASS
@@ -255,7 +257,7 @@ AC_DEFUN([ZFS_AC_KERNEL], [
 	AS_IF([test "$utsrelease"], [
 		kernsrcver=`(echo "#include <$utsrelease>";
 		             echo "kernsrcver=UTS_RELEASE") |
-		             cpp -I $kernelbuild/include |
+		             ${CPP} -I $kernelbuild/include - |
 		             grep "^kernsrcver=" | cut -d \" -f 2`
 
 		AS_IF([test -z "$kernsrcver"], [

configure.ac

@@ -122,6 +122,9 @@ AC_CONFIG_FILES([
 	contrib/dracut/02zfsexpandknowledge/Makefile
 	contrib/dracut/90zfs/Makefile
 	contrib/initramfs/Makefile
+	contrib/initramfs/hooks/Makefile
+	contrib/initramfs/scripts/Makefile
+	contrib/initramfs/scripts/local-top/Makefile
 	module/Makefile
 	module/avl/Makefile
 	module/nvpair/Makefile

contrib/dracut/90zfs/Makefile.am

@@ -24,6 +24,7 @@ $(pkgdracut_SCRIPTS):%:%.in
 		-e 's,@udevruledir\@,$(udevruledir),g' \
 		-e 's,@sysconfdir\@,$(sysconfdir),g' \
 		-e 's,@systemdunitdir\@,$(systemdunitdir),g' \
+		-e 's,@mounthelperdir\@,$(mounthelperdir),g' \
 		$< >'$@'
 
 distclean-local::

contrib/dracut/90zfs/module-setup.sh.in

@@ -5,7 +5,7 @@ check() {
 	[ "${1}" = "-d" ] && return 0
 
 	# Verify the zfs tool chain
-	for tool in "@sbindir@/zpool" "@sbindir@/zfs" "@sbindir@/mount.zfs" ; do
+	for tool in "@sbindir@/zpool" "@sbindir@/zfs" "@mounthelperdir@/mount.zfs" ; do
 		test -x "$tool" || return 1
 	done
 	# Verify grep exists
@@ -53,7 +53,7 @@ install() {
 		# Fallback: Guess the path and include all matches
 		dracut_install /usr/lib/gcc/*/*/libgcc_s.so*
 	fi
-	dracut_install @sbindir@/mount.zfs
+	dracut_install @mounthelperdir@/mount.zfs
 	dracut_install @udevdir@/vdev_id
 	dracut_install awk
 	dracut_install head

contrib/initramfs/Makefile.am

@@ -3,12 +3,11 @@ initrddir = $(datarootdir)/initramfs-tools
 initrd_SCRIPTS = \
 	conf.d/zfs conf-hooks.d/zfs hooks/zfs scripts/zfs scripts/local-top/zfs
 
+SUBDIRS = hooks scripts
+
 EXTRA_DIST = \
 	$(top_srcdir)/contrib/initramfs/conf.d/zfs \
 	$(top_srcdir)/contrib/initramfs/conf-hooks.d/zfs \
-	$(top_srcdir)/contrib/initramfs/hooks/zfs \
-	$(top_srcdir)/contrib/initramfs/scripts/zfs \
-	$(top_srcdir)/contrib/initramfs/scripts/local-top/zfs \
 	$(top_srcdir)/contrib/initramfs/README.initramfs.markdown
 
 install-initrdSCRIPTS: $(EXTRA_DIST)

contrib/initramfs/hooks/.gitignore

@@ -0,0 +1 @@
+zfs

contrib/initramfs/hooks/Makefile.am

@@ -0,0 +1,21 @@
+hooksdir = $(datarootdir)/initramfs-tools/hooks
+
+hooks_SCRIPTS = \
+	zfs
+
+EXTRA_DIST = \
+	$(top_srcdir)/contrib/initramfs/hooks/zfs.in
+
+$(hooks_SCRIPTS):%:%.in
+	-$(SED) -e 's,@sbindir\@,$(sbindir),g' \
+		-e 's,@sysconfdir\@,$(sysconfdir),g' \
+		-e 's,@udevdir\@,$(udevdir),g' \
+		-e 's,@udevruledir\@,$(udevruledir),g' \
+		-e 's,@mounthelperdir\@,$(mounthelperdir),g' \
+		$< >'$@'
+
+clean-local::
+	-$(RM) $(hooks_SCRIPTS)
+
+distclean-local::
+	-$(RM) $(hooks_SCRIPTS)

contrib/initramfs/hooks/zfs.in

@@ -8,11 +8,13 @@ PREREQ="zdev"
 
 # These prerequisites are provided by the zfsutils package. The zdb utility is
 # not strictly required, but it can be useful at the initramfs recovery prompt.
-COPY_EXEC_LIST="/sbin/zdb /sbin/zpool /sbin/zfs /sbin/mount.zfs"
-COPY_EXEC_LIST="$COPY_EXEC_LIST /usr/bin/dirname /lib/udev/vdev_id"
-COPY_FILE_LIST="/etc/hostid /etc/zfs/zpool.cache /etc/default/zfs"
-COPY_FILE_LIST="$COPY_FILE_LIST /etc/zfs/zfs-functions /etc/zfs/vdev_id.conf"
-COPY_FILE_LIST="$COPY_FILE_LIST /lib/udev/rules.d/69-vdev.rules"
+COPY_EXEC_LIST="@sbindir@/zdb @sbindir@/zpool @sbindir@/zfs"
+COPY_EXEC_LIST="$COPY_EXEC_LIST @mounthelperdir@/mount.zfs @udevdir@/vdev_id"
+COPY_FILE_LIST="/etc/hostid @sysconfdir@/zfs/zpool.cache"
+COPY_FILE_LIST="$COPY_FILE_LIST @sysconfdir@/default/zfs"
+COPY_FILE_LIST="$COPY_FILE_LIST @sysconfdir@/zfs/zfs-functions"
+COPY_FILE_LIST="$COPY_FILE_LIST @sysconfdir@/zfs/vdev_id.conf"
+COPY_FILE_LIST="$COPY_FILE_LIST @udevruledir@/69-vdev.rules"
 
 # These prerequisites are provided by the base system.
 COPY_EXEC_LIST="$COPY_EXEC_LIST /usr/bin/dirname /bin/hostname /sbin/blkid"

contrib/initramfs/scripts/.gitignore

@@ -0,0 +1 @@
+zfs

contrib/initramfs/scripts/Makefile.am

@@ -0,0 +1,20 @@
+scriptsdir = $(datarootdir)/initramfs-tools/scripts
+
+scripts_SCRIPTS = \
+	zfs
+
+SUBDIRS = local-top
+
+EXTRA_DIST = \
+	$(top_srcdir)/contrib/initramfs/scripts/zfs.in
+
+$(scripts_SCRIPTS):%:%.in
+	-$(SED) -e 's,@sbindir\@,$(sbindir),g' \
+		-e 's,@sysconfdir\@,$(sysconfdir),g' \
+		$< >'$@'
+
+clean-local::
+	-$(RM) $(scripts_SCRIPTS)
+
+distclean-local::
+	-$(RM) $(scripts_SCRIPTS)

contrib/initramfs/scripts/local-top/Makefile.am

@@ -0,0 +1,3 @@
+localtopdir = $(datarootdir)/initramfs-tools/scripts/local-top
+
+EXTRA_DIST = zfs

contrib/initramfs/scripts/zfs.in

@@ -11,9 +11,9 @@
 
 # Paths to what we need - in the initrd, these paths are hardcoded,
 # so override the defines in zfs-functions.
-ZFS="/sbin/zfs"
-ZPOOL="/sbin/zpool"
-ZPOOL_CACHE="/etc/zfs/zpool.cache"
+ZFS="@sbindir@/zfs"
+ZPOOL="@sbindir@/zpool"
+ZPOOL_CACHE="@sysconfdir@/zfs/zpool.cache"
 export ZFS ZPOOL ZPOOL_CACHE
 
 # This runs any scripts that should run before we start importing

etc/modules-load.d/zfs.conf.in

@@ -1,3 +1,3 @@
-# Always load kernel modules at boot.  The default behavior is to load the
-# kernel modules in the zfs-import-*.service or when blkid(8) detects a pool.
+# The default behavior is to allow udev to load the kernel modules on demand.
+# Uncomment the following line to unconditionally load them at boot.
 #zfs

etc/systemd/system/zfs-import-cache.service.in

@@ -12,7 +12,6 @@ ConditionPathExists=@sysconfdir@/zfs/zpool.cache
 [Service]
 Type=oneshot
 RemainAfterExit=yes
-ExecStartPre=-/sbin/modprobe zfs
 ExecStart=@sbindir@/zpool import -c @sysconfdir@/zfs/zpool.cache -aN
 
 [Install]

etc/systemd/system/zfs-import-scan.service.in

@@ -11,7 +11,6 @@ ConditionPathExists=!@sysconfdir@/zfs/zpool.cache
 [Service]
 Type=oneshot
 RemainAfterExit=yes
-ExecStartPre=-/sbin/modprobe zfs
 ExecStart=@sbindir@/zpool import -aN -o cachefile=none
 
 [Install]

etc/zfs/Makefile.am

@@ -4,6 +4,7 @@ pkgsysconf_DATA = \
 	vdev_id.conf.alias.example \
 	vdev_id.conf.sas_direct.example \
 	vdev_id.conf.sas_switch.example \
-	vdev_id.conf.multipath.example
+	vdev_id.conf.multipath.example \
+	vdev_id.conf.scsi.example
 
 EXTRA_DIST = $(pkgsysconf_DATA)

etc/zfs/vdev_id.conf.sas_direct.example

@@ -2,6 +2,9 @@ multipath     no
 topology      sas_direct
 phys_per_port 4
 
+# Additionally create /dev/by-enclousure/ symlinks for enclosure devices
+enclosure_symlinks	yes
+
 #       PCI_ID  HBA PORT  CHANNEL NAME
 channel 85:00.0 1         A
 channel 85:00.0 0         B

etc/zfs/vdev_id.conf.scsi.example

@@ -0,0 +1,9 @@
+multipath     no
+topology      scsi
+phys_per_port 1
+# Usually scsi disks are numbered from 0, but this can be offset, to
+# match the physical bay numbers, as follows:
+first_bay_number 1
+
+#       PCI_ID  HBA PORT  CHANNEL NAME
+channel 0c:00.0 0         Y

include/linux/kmap_compat.h

@@ -27,6 +27,7 @@
 #define	_ZFS_KMAP_H
 
 #include <linux/highmem.h>
+#include <linux/uaccess.h>
 
 #ifdef HAVE_1ARG_KMAP_ATOMIC
 /* 2.6.37 API change */
@@ -37,4 +38,11 @@
 #define	zfs_kunmap_atomic(addr, km_type)	kunmap_atomic(addr, km_type)
 #endif
 
+/* 5.0 API change - no more 'type' argument for access_ok() */
+#ifdef HAVE_ACCESS_OK_TYPE
+#define	zfs_access_ok(type, addr, size)	access_ok(type, addr, size)
+#else
+#define	zfs_access_ok(type, addr, size)	access_ok(addr, size)
+#endif
+
 #endif	/* _ZFS_KMAP_H */

include/linux/simd_x86.h

@@ -81,7 +81,7 @@
 #endif
 
 #if defined(_KERNEL)
-#if defined(HAVE_FPU_API_H)
+#if defined(HAVE_UNDERSCORE_KERNEL_FPU)
 #include <asm/fpu/api.h>
 #include <asm/fpu/internal.h>
 #define	kfpu_begin()		\
@@ -94,12 +94,18 @@
 	__kernel_fpu_end();		\
 	preempt_enable();		\
 }
-#else
+#elif defined(HAVE_KERNEL_FPU)
 #include <asm/i387.h>
 #include <asm/xcr.h>
 #define	kfpu_begin()	kernel_fpu_begin()
 #define	kfpu_end()		kernel_fpu_end()
-#endif /* defined(HAVE_FPU_API_H) */
+#else
+/* Kernel doesn't export any kernel_fpu_* functions */
+#include <asm/fpu/internal.h>	/* For kernel xgetbv() */
+#define	kfpu_begin() 	panic("This code should never run")
+#define	kfpu_end() 	panic("This code should never run")
+#endif /* defined(HAVE_KERNEL_FPU) */
+
 #else
 /*
  * fpu dummy methods for userspace
@@ -278,11 +284,13 @@ __simd_state_enabled(const uint64_t state)
 	boolean_t has_osxsave;
 	uint64_t xcr0;
 
-#if defined(_KERNEL) && defined(X86_FEATURE_OSXSAVE)
+#if defined(_KERNEL)
+#if defined(X86_FEATURE_OSXSAVE) && defined(KERNEL_EXPORTS_X86_FPU)
 	has_osxsave = !!boot_cpu_has(X86_FEATURE_OSXSAVE);
-#elif defined(_KERNEL) && !defined(X86_FEATURE_OSXSAVE)
-	has_osxsave = B_FALSE;
 #else
+	has_osxsave = B_FALSE;
+#endif
+#elif !defined(_KERNEL)
 	has_osxsave = __cpuid_has_osxsave();
 #endif
 
@@ -307,8 +315,12 @@ static inline boolean_t
 zfs_sse_available(void)
 {
 #if defined(_KERNEL)
+#if defined(KERNEL_EXPORTS_X86_FPU)
 	return (!!boot_cpu_has(X86_FEATURE_XMM));
 #else
+	return (B_FALSE);
+#endif
+#elif !defined(_KERNEL)
 	return (__cpuid_has_sse());
 #endif
 }
@@ -320,8 +332,12 @@ static inline boolean_t
 zfs_sse2_available(void)
 {
 #if defined(_KERNEL)
+#if defined(KERNEL_EXPORTS_X86_FPU)
 	return (!!boot_cpu_has(X86_FEATURE_XMM2));
 #else
+	return (B_FALSE);
+#endif
+#elif !defined(_KERNEL)
 	return (__cpuid_has_sse2());
 #endif
 }
@@ -333,8 +349,12 @@ static inline boolean_t
 zfs_sse3_available(void)
 {
 #if defined(_KERNEL)
+#if defined(KERNEL_EXPORTS_X86_FPU)
 	return (!!boot_cpu_has(X86_FEATURE_XMM3));
 #else
+	return (B_FALSE);
+#endif
+#elif !defined(_KERNEL)
 	return (__cpuid_has_sse3());
 #endif
 }
@@ -346,8 +366,12 @@ static inline boolean_t
 zfs_ssse3_available(void)
 {
 #if defined(_KERNEL)
+#if defined(KERNEL_EXPORTS_X86_FPU)
 	return (!!boot_cpu_has(X86_FEATURE_SSSE3));
 #else
+	return (B_FALSE);
+#endif
+#elif !defined(_KERNEL)
 	return (__cpuid_has_ssse3());
 #endif
 }
@@ -359,8 +383,12 @@ static inline boolean_t
 zfs_sse4_1_available(void)
 {
 #if defined(_KERNEL)
+#if defined(KERNEL_EXPORTS_X86_FPU)
 	return (!!boot_cpu_has(X86_FEATURE_XMM4_1));
 #else
+	return (B_FALSE);
+#endif
+#elif !defined(_KERNEL)
 	return (__cpuid_has_sse4_1());
 #endif
 }
@@ -372,8 +400,12 @@ static inline boolean_t
 zfs_sse4_2_available(void)
 {
 #if defined(_KERNEL)
+#if defined(KERNEL_EXPORTS_X86_FPU)
 	return (!!boot_cpu_has(X86_FEATURE_XMM4_2));
 #else
+	return (B_FALSE);
+#endif
+#elif !defined(_KERNEL)
 	return (__cpuid_has_sse4_2());
 #endif
 }
@@ -386,8 +418,12 @@ zfs_avx_available(void)
 {
 	boolean_t has_avx;
 #if defined(_KERNEL)
+#if defined(KERNEL_EXPORTS_X86_FPU)
 	has_avx = !!boot_cpu_has(X86_FEATURE_AVX);
 #else
+	has_avx = B_FALSE;
+#endif
+#elif !defined(_KERNEL)
 	has_avx = __cpuid_has_avx();
 #endif
 
@@ -401,11 +437,13 @@ static inline boolean_t
 zfs_avx2_available(void)
 {
 	boolean_t has_avx2;
-#if defined(_KERNEL) && defined(X86_FEATURE_AVX2)
+#if defined(_KERNEL)
+#if defined(X86_FEATURE_AVX2) && defined(KERNEL_EXPORTS_X86_FPU)
 	has_avx2 = !!boot_cpu_has(X86_FEATURE_AVX2);
-#elif defined(_KERNEL) && !defined(X86_FEATURE_AVX2)
-	has_avx2 = B_FALSE;
 #else
+	has_avx2 = B_FALSE;
+#endif
+#elif !defined(_KERNEL)
 	has_avx2 = __cpuid_has_avx2();
 #endif
 
@@ -418,11 +456,13 @@ zfs_avx2_available(void)
 static inline boolean_t
 zfs_bmi1_available(void)
 {
-#if defined(_KERNEL) && defined(X86_FEATURE_BMI1)
+#if defined(_KERNEL)
+#if defined(X86_FEATURE_BMI1) && defined(KERNEL_EXPORTS_X86_FPU)
 	return (!!boot_cpu_has(X86_FEATURE_BMI1));
-#elif defined(_KERNEL) && !defined(X86_FEATURE_BMI1)
-	return (B_FALSE);
 #else
+	return (B_FALSE);
+#endif
+#elif !defined(_KERNEL)
 	return (__cpuid_has_bmi1());
 #endif
 }
@@ -433,16 +473,17 @@ zfs_bmi1_available(void)
 static inline boolean_t
 zfs_bmi2_available(void)
 {
-#if defined(_KERNEL) && defined(X86_FEATURE_BMI2)
+#if defined(_KERNEL)
+#if defined(X86_FEATURE_BMI2) && defined(KERNEL_EXPORTS_X86_FPU)
 	return (!!boot_cpu_has(X86_FEATURE_BMI2));
-#elif defined(_KERNEL) && !defined(X86_FEATURE_BMI2)
-	return (B_FALSE);
 #else
+	return (B_FALSE);
+#endif
+#elif !defined(_KERNEL)
 	return (__cpuid_has_bmi2());
 #endif
 }
 
-
 /*
  * AVX-512 family of instruction sets:
  *
@@ -466,8 +507,12 @@ zfs_avx512f_available(void)
 {
 	boolean_t has_avx512 = B_FALSE;
 
-#if defined(_KERNEL) && defined(X86_FEATURE_AVX512F)
+#if defined(_KERNEL)
+#if defined(X86_FEATURE_AVX512F) && defined(KERNEL_EXPORTS_X86_FPU)
 	has_avx512 = !!boot_cpu_has(X86_FEATURE_AVX512F);
+#else
+	has_avx512 = B_FALSE;
+#endif
 #elif !defined(_KERNEL)
 	has_avx512 = __cpuid_has_avx512f();
 #endif
@@ -481,9 +526,13 @@ zfs_avx512cd_available(void)
 {
 	boolean_t has_avx512 = B_FALSE;
 
-#if defined(_KERNEL) && defined(X86_FEATURE_AVX512CD)
+#if defined(_KERNEL)
+#if defined(X86_FEATURE_AVX512CD) && defined(KERNEL_EXPORTS_X86_FPU)
 	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
 	    boot_cpu_has(X86_FEATURE_AVX512CD);
+#else
+	has_avx512 = B_FALSE;
+#endif
 #elif !defined(_KERNEL)
 	has_avx512 = __cpuid_has_avx512cd();
 #endif
@@ -497,9 +546,13 @@ zfs_avx512er_available(void)
 {
 	boolean_t has_avx512 = B_FALSE;
 
-#if defined(_KERNEL) && defined(X86_FEATURE_AVX512ER)
+#if defined(_KERNEL)
+#if defined(X86_FEATURE_AVX512ER) && defined(KERNEL_EXPORTS_X86_FPU)
 	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
 	    boot_cpu_has(X86_FEATURE_AVX512ER);
+#else
+	has_avx512 = B_FALSE;
+#endif
 #elif !defined(_KERNEL)
 	has_avx512 = __cpuid_has_avx512er();
 #endif
@@ -513,9 +566,13 @@ zfs_avx512pf_available(void)
 {
 	boolean_t has_avx512 = B_FALSE;
 
-#if defined(_KERNEL) && defined(X86_FEATURE_AVX512PF)
+#if defined(_KERNEL)
+#if defined(X86_FEATURE_AVX512PF) && defined(KERNEL_EXPORTS_X86_FPU)
 	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
 	    boot_cpu_has(X86_FEATURE_AVX512PF);
+#else
+	has_avx512 = B_FALSE;
+#endif
 #elif !defined(_KERNEL)
 	has_avx512 = __cpuid_has_avx512pf();
 #endif
@@ -529,9 +586,13 @@ zfs_avx512bw_available(void)
 {
 	boolean_t has_avx512 = B_FALSE;
 
-#if defined(_KERNEL) && defined(X86_FEATURE_AVX512BW)
+#if defined(_KERNEL)
+#if defined(X86_FEATURE_AVX512BW) && defined(KERNEL_EXPORTS_X86_FPU)
 	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
 	    boot_cpu_has(X86_FEATURE_AVX512BW);
+#else
+	has_avx512 = B_FALSE;
+#endif
 #elif !defined(_KERNEL)
 	has_avx512 = __cpuid_has_avx512bw();
 #endif
@@ -545,9 +606,13 @@ zfs_avx512dq_available(void)
 {
 	boolean_t has_avx512 = B_FALSE;
 
-#if defined(_KERNEL) && defined(X86_FEATURE_AVX512DQ)
+#if defined(_KERNEL)
+#if defined(X86_FEATURE_AVX512DQ) && defined(KERNEL_EXPORTS_X86_FPU)
 	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
 	    boot_cpu_has(X86_FEATURE_AVX512DQ);
+#else
+	has_avx512 = B_FALSE;
+#endif
 #elif !defined(_KERNEL)
 	has_avx512 = __cpuid_has_avx512dq();
 #endif
@@ -561,9 +626,13 @@ zfs_avx512vl_available(void)
 {
 	boolean_t has_avx512 = B_FALSE;
 
-#if defined(_KERNEL) && defined(X86_FEATURE_AVX512VL)
+#if defined(_KERNEL)
+#if defined(X86_FEATURE_AVX512VL) && defined(KERNEL_EXPORTS_X86_FPU)
 	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
 	    boot_cpu_has(X86_FEATURE_AVX512VL);
+#else
+	has_avx512 = B_FALSE;
+#endif
 #elif !defined(_KERNEL)
 	has_avx512 = __cpuid_has_avx512vl();
 #endif
@@ -577,9 +646,13 @@ zfs_avx512ifma_available(void)
 {
 	boolean_t has_avx512 = B_FALSE;
 
-#if defined(_KERNEL) && defined(X86_FEATURE_AVX512IFMA)
+#if defined(_KERNEL)
+#if defined(X86_FEATURE_AVX512IFMA) && defined(KERNEL_EXPORTS_X86_FPU)
 	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
 	    boot_cpu_has(X86_FEATURE_AVX512IFMA);
+#else
+	has_avx512 = B_FALSE;
+#endif
 #elif !defined(_KERNEL)
 	has_avx512 = __cpuid_has_avx512ifma();
 #endif
@@ -593,9 +666,13 @@ zfs_avx512vbmi_available(void)
 {
 	boolean_t has_avx512 = B_FALSE;
 
-#if defined(_KERNEL) && defined(X86_FEATURE_AVX512VBMI)
+#if defined(_KERNEL)
+#if defined(X86_FEATURE_AVX512VBMI) && defined(KERNEL_EXPORTS_X86_FPU)
 	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
 	    boot_cpu_has(X86_FEATURE_AVX512VBMI);
+#else
+	has_avx512 = B_FALSE;
+#endif
 #elif !defined(_KERNEL)
 	has_avx512 = __cpuid_has_avx512f() &&
 	    __cpuid_has_avx512vbmi();

include/sys/dbuf.h

@@ -20,7 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2015 by Delphix. All rights reserved.
+ * Copyright (c) 2012, 2018 by Delphix. All rights reserved.
  * Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
  * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
  */
@@ -294,7 +294,7 @@ boolean_t dbuf_try_add_ref(dmu_buf_t *db, objset_t *os, uint64_t obj,
 uint64_t dbuf_refcount(dmu_buf_impl_t *db);
 
 void dbuf_rele(dmu_buf_impl_t *db, void *tag);
-void dbuf_rele_and_unlock(dmu_buf_impl_t *db, void *tag);
+void dbuf_rele_and_unlock(dmu_buf_impl_t *db, void *tag, boolean_t evicting);
 
 dmu_buf_impl_t *dbuf_find(struct objset *os, uint64_t object, uint8_t level,
     uint64_t blkid);

include/sys/dnode.h

@@ -20,7 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2017 by Delphix. All rights reserved.
+ * Copyright (c) 2012, 2018 by Delphix. All rights reserved.
  * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
  */
 
@@ -339,7 +339,7 @@ int dnode_hold_impl(struct objset *dd, uint64_t object, int flag, int dn_slots,
     void *ref, dnode_t **dnp);
 boolean_t dnode_add_ref(dnode_t *dn, void *ref);
 void dnode_rele(dnode_t *dn, void *ref);
-void dnode_rele_and_unlock(dnode_t *dn, void *tag);
+void dnode_rele_and_unlock(dnode_t *dn, void *tag, boolean_t evicting);
 void dnode_setdirty(dnode_t *dn, dmu_tx_t *tx);
 void dnode_sync(dnode_t *dn, dmu_tx_t *tx);
 void dnode_allocate(dnode_t *dn, dmu_object_type_t ot, int blocksize, int ibs,

include/sys/uio_impl.h

@@ -42,7 +42,7 @@
 #include <sys/uio.h>
 
 extern int uiomove(void *, size_t, enum uio_rw, uio_t *);
-extern void uio_prefaultpages(ssize_t, uio_t *);
+extern int uio_prefaultpages(ssize_t, uio_t *);
 extern int uiocopy(void *, size_t, enum uio_rw, uio_t *, size_t *);
 extern void uioskip(uio_t *, size_t);
 

lib/libzfs/libzfs_util.c

@@ -963,13 +963,14 @@ libzfs_load_module(const char *module)
 				load = 0;
 		}
 
-		if (load && libzfs_run_process("/sbin/modprobe", argv, 0))
+		if (load) {
+			if (libzfs_run_process("/sbin/modprobe", argv, 0))
 				return (ENOEXEC);
-	}
 
-	/* Module loading is synchronous it must be available */
 			if (!libzfs_module_loaded(module))
 				return (ENXIO);
+		}
+	}
 
 	/*
 	 * Device creation by udev is asynchronous and waiting may be

man/man5/vdev_id.conf.5

@@ -38,6 +38,19 @@ defined by udev.  This may be an absolute path or the base filename.
 Maps a physical path to a channel name (typically representing a single
 disk enclosure).
 
+.TP
+\fIenclosure_symlinks\fR <yes|no>
+Additionally create /dev/by-enclosure symlinks to the disk enclosure
+sg devices using the naming scheme from from vdev_id.conf.
+\fIenclosure_symlinks\fR is only allowed for sas_direct mode.
+.TP
+\fIenclosure_symlinks_prefix\fR <prefix>
+Specify the prefix for the enclosure symlinks in the form of:
+
+/dev/by-enclosure/<prefix>-<channel><num>
+
+Defaults to "enc" if not specified.
+.TP
 \fIpci_slot\fR - specifies the PCI SLOT of the HBA
 hosting the disk enclosure being mapped, as found in the output of
 .BR lspci (8).
@@ -90,7 +103,7 @@ internally uses this value to determine which HBA or switch port a
 device is connected to.  The default is 4.
 
 .TP
-\fIslot\fR <bay|phy|port|id|lun>
+\fIslot\fR <bay|phy|port|id|lun|ses>
 Specifies from which element of a SAS identifier the slot number is
 taken.  The default is bay.
 
@@ -103,6 +116,12 @@ taken.  The default is bay.
 \fIid\fR - use the scsi id as the slot number.
 
 \fIlun\fR - use the scsi lun as the slot number.
+
+\fIses\fR - use the SCSI Enclosure Services (SES) enclosure device slot number,
+as reported by
+.BR sg_ses (8).
+This is intended for use only on systems where \fIbay\fR is unsupported,
+noting that \fIport\fR and \fIid\fR may be unstable across disk replacement.
 .SH EXAMPLES
 A non-multipath configuration with direct-attached SAS enclosures and an
 arbitrary slot re-mapping.
@@ -163,6 +182,27 @@ definitions - one per physical path.
 	channel 86:00.0  0         B
 .fi
 .P
+A configuration with enclosure_symlinks enabled.
+.P
+.nf
+	multipath yes
+	enclosure_symlinks yes
+
+	#          PCI_ID      HBA PORT     CHANNEL NAME
+	channel    05:00.0     1            U
+	channel    05:00.0     0            L
+	channel    06:00.0     1            U
+	channel    06:00.0     0            L
+.fi
+In addition to the disks symlinks, this configuration will create:
+.P
+.nf
+	/dev/by-enclosure/enc-L0
+	/dev/by-enclosure/enc-L1
+	/dev/by-enclosure/enc-U0
+	/dev/by-enclosure/enc-U1
+.fi
+.P
 A configuration using device link aliases.
 .P
 .nf

man/man8/zfs.8

@@ -29,7 +29,7 @@
 .\" Copyright 2016 Nexenta Systems, Inc.
 .\" Copyright 2016 Richard Laager. All rights reserved.
 .\"
-.Dd July 13, 2018
+.Dd Jan 05, 2019
 .Dt ZFS 8 SMM
 .Os Linux
 .Sh NAME
@@ -3981,9 +3981,9 @@ renames the remaining snapshots, and then creates a new snapshot, as follows:
 # zfs destroy -r pool/users@7daysago
 # zfs rename -r pool/users@6daysago @7daysago
 # zfs rename -r pool/users@5daysago @6daysago
-# zfs rename -r pool/users@yesterday @5daysago
-# zfs rename -r pool/users@yesterday @4daysago
-# zfs rename -r pool/users@yesterday @3daysago
+# zfs rename -r pool/users@4daysago @5daysago
+# zfs rename -r pool/users@3daysago @4daysago
+# zfs rename -r pool/users@2daysago @3daysago
 # zfs rename -r pool/users@yesterday @2daysago
 # zfs rename -r pool/users@today @yesterday
 # zfs snapshot -r pool/users@today

module/Makefile.in

@@ -36,12 +36,12 @@ modules:
 	list='$(SUBDIR_TARGETS)'; for targetdir in $$list; do \
 		$(MAKE) -C $$targetdir; \
 	done
-	$(MAKE) -C @LINUX_OBJ@ SUBDIRS=`pwd` @KERNELMAKE_PARAMS@ CONFIG_ZFS=m $@
+	$(MAKE) -C @LINUX_OBJ@ M=`pwd` @KERNELMAKE_PARAMS@ CONFIG_ZFS=m $@
 
 clean:
 	@# Only cleanup the kernel build directories when CONFIG_KERNEL
 	@# is defined.  This indicates that kernel modules should be built.
-@CONFIG_KERNEL_TRUE@	$(MAKE) -C @LINUX_OBJ@ SUBDIRS=`pwd` @KERNELMAKE_PARAMS@ $@
+@CONFIG_KERNEL_TRUE@	$(MAKE) -C @LINUX_OBJ@ M=`pwd` @KERNELMAKE_PARAMS@ $@
 
 	if [ -f @SPL_SYMBOLS@ ]; then $(RM) @SPL_SYMBOLS@; fi
 	if [ -f @LINUX_SYMBOLS@ ]; then $(RM) @LINUX_SYMBOLS@; fi
@@ -49,7 +49,7 @@ clean:
 
 modules_install:
 	@# Install the kernel modules
-	$(MAKE) -C @LINUX_OBJ@ SUBDIRS=`pwd` $@ \
+	$(MAKE) -C @LINUX_OBJ@ M=`pwd` $@ \
 		INSTALL_MOD_PATH=$(DESTDIR)$(INSTALL_MOD_PATH) \
 		INSTALL_MOD_DIR=$(INSTALL_MOD_DIR) \
 		KERNELRELEASE=@LINUX_VERSION@

module/zcommon/zfs_uio.c

@@ -50,6 +50,7 @@
 #include <sys/types.h>
 #include <sys/uio_impl.h>
 #include <linux/kmap_compat.h>
+#include <linux/uaccess.h>
 
 /*
  * Move "n" bytes at byte address "p"; "rw" indicates the direction
@@ -77,9 +78,24 @@ uiomove_iov(void *p, size_t n, enum uio_rw rw, struct uio *uio)
 				if (copy_to_user(iov->iov_base+skip, p, cnt))
 					return (EFAULT);
 			} else {
-				if (copy_from_user(p, iov->iov_base+skip, cnt))
+				if (uio->uio_fault_disable) {
+					if (!zfs_access_ok(VERIFY_READ,
+					    (iov->iov_base + skip), cnt)) {
 						return (EFAULT);
 					}
+					pagefault_disable();
+					if (__copy_from_user_inatomic(p,
+					    (iov->iov_base + skip), cnt)) {
+						pagefault_enable();
+						return (EFAULT);
+					}
+					pagefault_enable();
+				} else {
+					if (copy_from_user(p,
+					    (iov->iov_base + skip), cnt))
+						return (EFAULT);
+				}
+			}
 			break;
 		case UIO_SYSSPACE:
 			if (rw == UIO_READ)
@@ -156,7 +172,7 @@ EXPORT_SYMBOL(uiomove);
  * error will terminate the process as this is only a best attempt to get
  * the pages resident.
  */
-void
+int
 uio_prefaultpages(ssize_t n, struct uio *uio)
 {
 	const struct iovec *iov;
@@ -170,7 +186,7 @@ uio_prefaultpages(ssize_t n, struct uio *uio)
 	switch (uio->uio_segflg) {
 		case UIO_SYSSPACE:
 		case UIO_BVEC:
-			return;
+			return (0);
 		case UIO_USERSPACE:
 		case UIO_USERISPACE:
 			break;
@@ -194,7 +210,7 @@ uio_prefaultpages(ssize_t n, struct uio *uio)
 		p = iov->iov_base + skip;
 		while (cnt) {
 			if (fuword8((uint8_t *)p, &tmp))
-				return;
+				return (EFAULT);
 			incr = MIN(cnt, PAGESIZE);
 			p += incr;
 			cnt -= incr;
@@ -204,8 +220,10 @@ uio_prefaultpages(ssize_t n, struct uio *uio)
 		 */
 		p--;
 		if (fuword8((uint8_t *)p, &tmp))
-			return;
+			return (EFAULT);
 	}
+
+	return (0);
 }
 EXPORT_SYMBOL(uio_prefaultpages);
 

module/zfs/arc.c

@@ -4004,9 +4004,9 @@ arc_all_memory(void)
 {
 #ifdef _KERNEL
 #ifdef CONFIG_HIGHMEM
-	return (ptob(totalram_pages - totalhigh_pages));
+	return (ptob(zfs_totalram_pages - totalhigh_pages));
 #else
-	return (ptob(totalram_pages));
+	return (ptob(zfs_totalram_pages));
 #endif /* CONFIG_HIGHMEM */
 #else
 	return (ptob(physmem) / 2);

module/zfs/dbuf.c

@@ -72,8 +72,6 @@ static void __dbuf_hold_impl_init(struct dbuf_hold_impl_data *dh,
 	void *tag, dmu_buf_impl_t **dbp, int depth);
 static int __dbuf_hold_impl(struct dbuf_hold_impl_data *dh);
 
-uint_t zfs_dbuf_evict_key;
-
 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);
 
@@ -505,14 +503,6 @@ dbuf_evict_one(void)
 	dmu_buf_impl_t *db;
 	ASSERT(!MUTEX_HELD(&dbuf_evict_lock));
 
-	/*
-	 * Set the thread's tsd to indicate that it's processing evictions.
-	 * Once a thread stops evicting from the dbuf cache it will
-	 * reset its tsd to NULL.
-	 */
-	ASSERT3P(tsd_get(zfs_dbuf_evict_key), ==, NULL);
-	(void) tsd_set(zfs_dbuf_evict_key, (void *)B_TRUE);
-
 	db = multilist_sublist_tail(mls);
 	while (db != NULL && mutex_tryenter(&db->db_mtx) == 0) {
 		db = multilist_sublist_prev(mls, db);
@@ -530,7 +520,6 @@ dbuf_evict_one(void)
 	} else {
 		multilist_sublist_unlock(mls);
 	}
-	(void) tsd_set(zfs_dbuf_evict_key, NULL);
 }
 
 /*
@@ -583,29 +572,6 @@ dbuf_evict_thread(void)
 static void
 dbuf_evict_notify(void)
 {
-
-	/*
-	 * We use thread specific data to track when a thread has
-	 * started processing evictions. This allows us to avoid deeply
-	 * nested stacks that would have a call flow similar to this:
-	 *
-	 * dbuf_rele()-->dbuf_rele_and_unlock()-->dbuf_evict_notify()
-	 *	^						|
-	 *	|						|
-	 *	+-----dbuf_destroy()<--dbuf_evict_one()<--------+
-	 *
-	 * The dbuf_eviction_thread will always have its tsd set until
-	 * that thread exits. All other threads will only set their tsd
-	 * if they are participating in the eviction process. This only
-	 * happens if the eviction thread is unable to process evictions
-	 * fast enough. To keep the dbuf cache size in check, other threads
-	 * can evict from the dbuf cache directly. Those threads will set
-	 * their tsd values so that we ensure that they only evict one dbuf
-	 * from the dbuf cache.
-	 */
-	if (tsd_get(zfs_dbuf_evict_key) != NULL)
-		return;
-
 	/*
 	 * We check if we should evict without holding the dbuf_evict_lock,
 	 * because it's OK to occasionally make the wrong decision here,
@@ -681,7 +647,6 @@ retry:
 	    dbuf_cache_multilist_index_func);
 	zfs_refcount_create(&dbuf_cache_size);
 
-	tsd_create(&zfs_dbuf_evict_key, NULL);
 	dbuf_evict_thread_exit = B_FALSE;
 	mutex_init(&dbuf_evict_lock, NULL, MUTEX_DEFAULT, NULL);
 	cv_init(&dbuf_evict_cv, NULL, CV_DEFAULT, NULL);
@@ -718,7 +683,6 @@ dbuf_fini(void)
 		cv_wait(&dbuf_evict_cv, &dbuf_evict_lock);
 	}
 	mutex_exit(&dbuf_evict_lock);
-	tsd_destroy(&zfs_dbuf_evict_key);
 
 	mutex_destroy(&dbuf_evict_lock);
 	cv_destroy(&dbuf_evict_cv);
@@ -1004,7 +968,7 @@ dbuf_read_done(zio_t *zio, arc_buf_t *buf, void *vdb)
 		db->db_state = DB_UNCACHED;
 	}
 	cv_broadcast(&db->db_changed);
-	dbuf_rele_and_unlock(db, NULL);
+	dbuf_rele_and_unlock(db, NULL, B_FALSE);
 }
 
 static int
@@ -2178,7 +2142,8 @@ dbuf_destroy(dmu_buf_impl_t *db)
 		 * value in dnode_move(), since DB_DNODE_EXIT doesn't actually
 		 * release any lock.
 		 */
-		dnode_rele(dn, db);
+		mutex_enter(&dn->dn_mtx);
+		dnode_rele_and_unlock(dn, db, B_TRUE);
 		db->db_dnode_handle = NULL;
 
 		dbuf_hash_remove(db);
@@ -2204,8 +2169,10 @@ dbuf_destroy(dmu_buf_impl_t *db)
 	 * If this dbuf is referenced from an indirect dbuf,
 	 * decrement the ref count on the indirect dbuf.
 	 */
-	if (parent && parent != dndb)
-		dbuf_rele(parent, db);
+	if (parent && parent != dndb) {
+		mutex_enter(&parent->db_mtx);
+		dbuf_rele_and_unlock(parent, db, B_TRUE);
+	}
 }
 
 /*
@@ -2912,7 +2879,7 @@ void
 dbuf_rele(dmu_buf_impl_t *db, void *tag)
 {
 	mutex_enter(&db->db_mtx);
-	dbuf_rele_and_unlock(db, tag);
+	dbuf_rele_and_unlock(db, tag, B_FALSE);
 }
 
 void
@@ -2923,10 +2890,19 @@ dmu_buf_rele(dmu_buf_t *db, void *tag)
 
 /*
  * dbuf_rele() for an already-locked dbuf.  This is necessary to allow
- * db_dirtycnt and db_holds to be updated atomically.
+ * db_dirtycnt and db_holds to be updated atomically.  The 'evicting'
+ * argument should be set if we are already in the dbuf-evicting code
+ * path, in which case we don't want to recursively evict.  This allows us to
+ * avoid deeply nested stacks that would have a call flow similar to this:
+ *
+ * dbuf_rele()-->dbuf_rele_and_unlock()-->dbuf_evict_notify()
+ *	^						|
+ *	|						|
+ *	+-----dbuf_destroy()<--dbuf_evict_one()<--------+
+ *
  */
 void
-dbuf_rele_and_unlock(dmu_buf_impl_t *db, void *tag)
+dbuf_rele_and_unlock(dmu_buf_impl_t *db, void *tag, boolean_t evicting)
 {
 	int64_t holds;
 
@@ -3021,6 +2997,7 @@ dbuf_rele_and_unlock(dmu_buf_impl_t *db, void *tag)
 				    db->db.db_size, db);
 				mutex_exit(&db->db_mtx);
 
+				if (!evicting)
 					dbuf_evict_notify();
 			}
 
@@ -3314,7 +3291,7 @@ dbuf_sync_leaf(dbuf_dirty_record_t *dr, dmu_tx_t *tx)
 		kmem_free(dr, sizeof (dbuf_dirty_record_t));
 		ASSERT(db->db_dirtycnt > 0);
 		db->db_dirtycnt -= 1;
-		dbuf_rele_and_unlock(db, (void *)(uintptr_t)txg);
+		dbuf_rele_and_unlock(db, (void *)(uintptr_t)txg, B_FALSE);
 		return;
 	}
 
@@ -3670,7 +3647,7 @@ dbuf_write_done(zio_t *zio, arc_buf_t *buf, void *vdb)
 	ASSERT(db->db_dirtycnt > 0);
 	db->db_dirtycnt -= 1;
 	db->db_data_pending = NULL;
-	dbuf_rele_and_unlock(db, (void *)(uintptr_t)tx->tx_txg);
+	dbuf_rele_and_unlock(db, (void *)(uintptr_t)tx->tx_txg, B_FALSE);
 }
 
 static void

module/zfs/dnode.c

@@ -1533,11 +1533,11 @@ void
 dnode_rele(dnode_t *dn, void *tag)
 {
 	mutex_enter(&dn->dn_mtx);
-	dnode_rele_and_unlock(dn, tag);
+	dnode_rele_and_unlock(dn, tag, B_FALSE);
 }
 
 void
-dnode_rele_and_unlock(dnode_t *dn, void *tag)
+dnode_rele_and_unlock(dnode_t *dn, void *tag, boolean_t evicting)
 {
 	uint64_t refs;
 	/* Get while the hold prevents the dnode from moving. */
@@ -1568,7 +1568,8 @@ dnode_rele_and_unlock(dnode_t *dn, void *tag)
 		 * that the handle has zero references, but that will be
 		 * asserted anyway when the handle gets destroyed.
 		 */
-		dbuf_rele(db, dnh);
+		mutex_enter(&db->db_mtx);
+		dbuf_rele_and_unlock(db, dnh, evicting);
 	}
 }
 

module/zfs/dnode_sync.c

@@ -21,7 +21,7 @@
 
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2017 by Delphix. All rights reserved.
+ * Copyright (c) 2012, 2018 by Delphix. All rights reserved.
  * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
  */
 
@@ -429,6 +429,19 @@ dnode_evict_dbufs(dnode_t *dn)
 			avl_insert_here(&dn->dn_dbufs, db_marker, db,
 			    AVL_BEFORE);
 
+			/*
+			 * We need to use the "marker" dbuf rather than
+			 * simply getting the next dbuf, because
+			 * dbuf_destroy() may actually remove multiple dbufs.
+			 * It can call itself recursively on the parent dbuf,
+			 * which may also be removed from dn_dbufs.  The code
+			 * flow would look like:
+			 *
+			 * dbuf_destroy():
+			 *   dnode_rele_and_unlock(parent_dbuf, evicting=TRUE):
+			 *	if (!cacheable || pending_evict)
+			 *	  dbuf_destroy()
+			 */
 			dbuf_destroy(db);
 
 			db_next = AVL_NEXT(&dn->dn_dbufs, db_marker);
@@ -489,7 +502,7 @@ dnode_undirty_dbufs(list_t *list)
 			list_destroy(&dr->dt.di.dr_children);
 		}
 		kmem_free(dr, sizeof (dbuf_dirty_record_t));
-		dbuf_rele_and_unlock(db, (void *)(uintptr_t)txg);
+		dbuf_rele_and_unlock(db, (void *)(uintptr_t)txg, B_FALSE);
 	}
 }
 

module/zfs/vdev_disk.c

@@ -35,6 +35,7 @@
 #include <sys/zio.h>
 #include <sys/sunldi.h>
 #include <linux/mod_compat.h>
+#include <linux/vfs_compat.h>
 
 char *zfs_vdev_scheduler = VDEV_SCHEDULER;
 static void *zfs_vdev_holder = VDEV_HOLDER;
@@ -76,7 +77,7 @@ vdev_bdev_mode(int smode)
 	ASSERT3S(smode & (FREAD | FWRITE), !=, 0);
 
 	if ((smode & FREAD) && !(smode & FWRITE))
-		mode = MS_RDONLY;
+		mode = SB_RDONLY;
 
 	return (mode);
 }
@@ -501,13 +502,38 @@ vdev_submit_bio_impl(struct bio *bio)
 #endif
 }
 
-#ifndef HAVE_BIO_SET_DEV
+#ifdef HAVE_BIO_SET_DEV
+#if defined(CONFIG_BLK_CGROUP) && defined(HAVE_BIO_SET_DEV_GPL_ONLY)
+/*
+ * The Linux 5.0 kernel updated the bio_set_dev() macro so it calls the
+ * GPL-only bio_associate_blkg() symbol thus inadvertently converting
+ * the entire macro.  Provide a minimal version which always assigns the
+ * request queue's root_blkg to the bio.
+ */
+static inline void
+vdev_bio_associate_blkg(struct bio *bio)
+{
+	struct request_queue *q = bio->bi_disk->queue;
+
+	ASSERT3P(q, !=, NULL);
+	ASSERT3P(q->root_blkg, !=, NULL);
+	ASSERT3P(bio->bi_blkg, ==, NULL);
+
+	if (blkg_tryget(q->root_blkg))
+		bio->bi_blkg = q->root_blkg;
+}
+#define	bio_associate_blkg vdev_bio_associate_blkg
+#endif
+#else
+/*
+ * Provide a bio_set_dev() helper macro for pre-Linux 4.14 kernels.
+ */
 static inline void
 bio_set_dev(struct bio *bio, struct block_device *bdev)
 {
 	bio->bi_bdev = bdev;
 }
-#endif /* !HAVE_BIO_SET_DEV */
+#endif /* HAVE_BIO_SET_DEV */
 
 static inline void
 vdev_submit_bio(struct bio *bio)

module/zfs/zfs_ioctl.c

@@ -6634,11 +6634,14 @@ static const struct file_operations zfsdev_fops = {
 };
 
 static struct miscdevice zfs_misc = {
-	.minor		= MISC_DYNAMIC_MINOR,
+	.minor		= ZFS_DEVICE_MINOR,
 	.name		= ZFS_DRIVER,
 	.fops		= &zfsdev_fops,
 };
 
+MODULE_ALIAS_MISCDEV(ZFS_DEVICE_MINOR);
+MODULE_ALIAS("devname:zfs");
+
 static int
 zfs_attach(void)
 {
@@ -6649,12 +6652,24 @@ zfs_attach(void)
 	zfsdev_state_list->zs_minor = -1;
 
 	error = misc_register(&zfs_misc);
-	if (error != 0) {
-		printk(KERN_INFO "ZFS: misc_register() failed %d\n", error);
-		return (error);
+	if (error == -EBUSY) {
+		/*
+		 * Fallback to dynamic minor allocation in the event of a
+		 * collision with a reserved minor in linux/miscdevice.h.
+		 * In this case the kernel modules must be manually loaded.
+		 */
+		printk(KERN_INFO "ZFS: misc_register() with static minor %d "
+		    "failed %d, retrying with MISC_DYNAMIC_MINOR\n",
+		    ZFS_DEVICE_MINOR, error);
+
+		zfs_misc.minor = MISC_DYNAMIC_MINOR;
+		error = misc_register(&zfs_misc);
 	}
 
-	return (0);
+	if (error)
+		printk(KERN_INFO "ZFS: misc_register() failed %d\n", error);
+
+	return (error);
 }
 
 static void

module/zfs/zfs_vfsops.c

@@ -66,6 +66,7 @@
 #include <sys/dmu_objset.h>
 #include <sys/spa_boot.h>
 #include <sys/zpl.h>
+#include <linux/vfs_compat.h>
 #include "zfs_comutil.h"
 
 enum {
@@ -259,7 +260,7 @@ zfsvfs_parse_options(char *mntopts, vfs_t **vfsp)
 boolean_t
 zfs_is_readonly(zfsvfs_t *zfsvfs)
 {
-	return (!!(zfsvfs->z_sb->s_flags & MS_RDONLY));
+	return (!!(zfsvfs->z_sb->s_flags & SB_RDONLY));
 }
 
 /*ARGSUSED*/
@@ -353,15 +354,15 @@ acltype_changed_cb(void *arg, uint64_t newval)
 	switch (newval) {
 	case ZFS_ACLTYPE_OFF:
 		zfsvfs->z_acl_type = ZFS_ACLTYPE_OFF;
-		zfsvfs->z_sb->s_flags &= ~MS_POSIXACL;
+		zfsvfs->z_sb->s_flags &= ~SB_POSIXACL;
 		break;
 	case ZFS_ACLTYPE_POSIXACL:
 #ifdef CONFIG_FS_POSIX_ACL
 		zfsvfs->z_acl_type = ZFS_ACLTYPE_POSIXACL;
-		zfsvfs->z_sb->s_flags |= MS_POSIXACL;
+		zfsvfs->z_sb->s_flags |= SB_POSIXACL;
 #else
 		zfsvfs->z_acl_type = ZFS_ACLTYPE_OFF;
-		zfsvfs->z_sb->s_flags &= ~MS_POSIXACL;
+		zfsvfs->z_sb->s_flags &= ~SB_POSIXACL;
 #endif /* CONFIG_FS_POSIX_ACL */
 		break;
 	default:
@@ -390,9 +391,9 @@ readonly_changed_cb(void *arg, uint64_t newval)
 		return;
 
 	if (newval)
-		sb->s_flags |= MS_RDONLY;
+		sb->s_flags |= SB_RDONLY;
 	else
-		sb->s_flags &= ~MS_RDONLY;
+		sb->s_flags &= ~SB_RDONLY;
 }
 
 static void
@@ -420,9 +421,9 @@ nbmand_changed_cb(void *arg, uint64_t newval)
 		return;
 
 	if (newval == TRUE)
-		sb->s_flags |= MS_MANDLOCK;
+		sb->s_flags |= SB_MANDLOCK;
 	else
-		sb->s_flags &= ~MS_MANDLOCK;
+		sb->s_flags &= ~SB_MANDLOCK;
 }
 
 static void
@@ -1763,8 +1764,8 @@ zfs_remount(struct super_block *sb, int *flags, zfs_mnt_t *zm)
 	int error;
 
 	if ((issnap || !spa_writeable(dmu_objset_spa(zfsvfs->z_os))) &&
-	    !(*flags & MS_RDONLY)) {
-		*flags |= MS_RDONLY;
+	    !(*flags & SB_RDONLY)) {
+		*flags |= SB_RDONLY;
 		return (EROFS);
 	}
 

module/zfs/zfs_vnops.c

@@ -675,7 +675,10 @@ zfs_write(struct inode *ip, uio_t *uio, int ioflag, cred_t *cr)
 		xuio = (xuio_t *)uio;
 	else
 #endif
-		uio_prefaultpages(MIN(n, max_blksz), uio);
+		if (uio_prefaultpages(MIN(n, max_blksz), uio)) {
+			ZFS_EXIT(zfsvfs);
+			return (SET_ERROR(EFAULT));
+		}
 
 	/*
 	 * If in append mode, set the io offset pointer to eof.
@@ -820,8 +823,19 @@ zfs_write(struct inode *ip, uio_t *uio, int ioflag, cred_t *cr)
 
 		if (abuf == NULL) {
 			tx_bytes = uio->uio_resid;
+			uio->uio_fault_disable = B_TRUE;
 			error = dmu_write_uio_dbuf(sa_get_db(zp->z_sa_hdl),
 			    uio, nbytes, tx);
+			if (error == EFAULT) {
+				dmu_tx_commit(tx);
+				if (uio_prefaultpages(MIN(n, max_blksz), uio)) {
+					break;
+				}
+				continue;
+			} else if (error != 0) {
+				dmu_tx_commit(tx);
+				break;
+			}
 			tx_bytes -= uio->uio_resid;
 		} else {
 			tx_bytes = nbytes;
@@ -921,8 +935,12 @@ zfs_write(struct inode *ip, uio_t *uio, int ioflag, cred_t *cr)
 		ASSERT(tx_bytes == nbytes);
 		n -= nbytes;
 
-		if (!xuio && n > 0)
-			uio_prefaultpages(MIN(n, max_blksz), uio);
+		if (!xuio && n > 0) {
+			if (uio_prefaultpages(MIN(n, max_blksz), uio)) {
+				error = EFAULT;
+				break;
+			}
+		}
 	}
 
 	zfs_inode_update(zp);

rpm/generic/zfs-kmod.spec.in

@@ -195,6 +195,9 @@ chmod u+x ${RPM_BUILD_ROOT}%{kmodinstdir_prefix}/*/extra/*/*/*
 rm -rf $RPM_BUILD_ROOT
 
 %changelog
+* Fri Feb 22 2019 Tony Hutter <hutter2@llnl.gov> - 0.7.13-1
+- Released 0.7.13-1, detailed release notes are available at:
+- https://github.com/zfsonlinux/zfs/releases/tag/zfs-0.7.13
 * Thu Nov 08 2018 Tony Hutter <hutter2@llnl.gov> - 0.7.12-1
 - Released 0.7.12-1, detailed release notes are available at:
 - https://github.com/zfsonlinux/zfs/releases/tag/zfs-0.7.12

rpm/generic/zfs.spec.in

@@ -283,6 +283,15 @@ fi
 %endif
 exit 0
 
+# On RHEL/CentOS 7 the static nodes aren't refreshed by default after
+# installing a package.  This is the default behavior for Fedora.
+%posttrans
+%if 0%{?rhel} == 7 || 0%{?centos} == 7
+systemctl restart kmod-static-nodes
+systemctl restart systemd-tmpfiles-setup-dev
+udevadm trigger
+%endif
+
 %preun
 %if 0%{?_systemd}
 %if 0%{?systemd_preun:1}
@@ -372,6 +381,9 @@ systemctl --system daemon-reload >/dev/null || true
 %endif
 
 %changelog
+* Fri Feb 22 2019 Tony Hutter <hutter2@llnl.gov> - 0.7.13-1
+- Released 0.7.13-1, detailed release notes are available at:
+- https://github.com/zfsonlinux/zfs/releases/tag/zfs-0.7.13
 * Thu Nov 08 2018 Tony Hutter <hutter2@llnl.gov> - 0.7.12-1
 - Released 0.7.12-1, detailed release notes are available at:
 - https://github.com/zfsonlinux/zfs/releases/tag/zfs-0.7.12

scripts/commitcheck.sh

@@ -50,10 +50,10 @@ function new_change_commit()
 {
     error=0
 
-    # subject is not longer than 50 characters
-    long_subject=$(git log -n 1 --pretty=%s "$REF" | grep -E -m 1 '.{51}')
+    # subject is not longer than 72 characters
+    long_subject=$(git log -n 1 --pretty=%s "$REF" | grep -E -m 1 '.{73}')
     if [ -n "$long_subject" ]; then
-        echo "error: commit subject over 50 characters"
+        echo "error: commit subject over 72 characters"
         error=1
     fi
 

scripts/dkms.mkconf

@@ -65,6 +65,10 @@ PRE_BUILD="configure
       then
         echo --enable-debug-dmu-tx
       fi
+      if [[ \${ZFS_DKMS_ENABLE_DEBUGINFO,,} == @(y|yes) ]]
+      then
+        echo --enable-debuginfo
+      fi
     }
   )
 "

tests/test-runner/cmd/test-runner.py

@@ -12,13 +12,19 @@
 #
 
 #
-# Copyright (c) 2012, 2015 by Delphix. All rights reserved.
+# Copyright (c) 2012, 2018 by Delphix. All rights reserved.
 # Copyright (c) 2017 Datto Inc.
 #
 
-import ConfigParser
+# some python 2.7 system don't have a configparser shim
+try:
+    import configparser
+except ImportError:
+    import ConfigParser as configparser
+
 import os
-import logging
+import sys
+
 from datetime import datetime
 from optparse import OptionParser
 from pwd import getpwnam
@@ -26,8 +32,6 @@ from pwd import getpwuid
 from select import select
 from subprocess import PIPE
 from subprocess import Popen
-from sys import argv
-from sys import maxint
 from threading import Timer
 from time import time
 
@@ -36,6 +40,10 @@ TESTDIR = '/usr/share/zfs/'
 KILL = 'kill'
 TRUE = 'true'
 SUDO = 'sudo'
+LOG_FILE = 'LOG_FILE'
+LOG_OUT = 'LOG_OUT'
+LOG_ERR = 'LOG_ERR'
+LOG_FILE_OBJ = None
 
 
 class Result(object):
@@ -79,7 +87,7 @@ class Output(object):
     """
     def __init__(self, stream):
         self.stream = stream
-        self._buf = ''
+        self._buf = b''
         self.lines = []
 
     def fileno(self):
@@ -104,15 +112,15 @@ class Output(object):
         buf = os.read(fd, 4096)
         if not buf:
             return None
-        if '\n' not in buf:
+        if b'\n' not in buf:
             self._buf += buf
             return []
 
         buf = self._buf + buf
-        tmp, rest = buf.rsplit('\n', 1)
+        tmp, rest = buf.rsplit(b'\n', 1)
         self._buf = rest
         now = datetime.now()
-        rows = tmp.split('\n')
+        rows = tmp.split(b'\n')
         self.lines += [(now, r) for r in rows]
 
 
@@ -204,23 +212,23 @@ class Cmd(object):
         if needed. Run the command, and update the result object.
         """
         if options.dryrun is True:
-            print self
+            print(self)
             return
 
         privcmd = self.update_cmd_privs(self.pathname, self.user)
         try:
             old = os.umask(0)
             if not os.path.isdir(self.outputdir):
-                os.makedirs(self.outputdir, mode=0777)
+                os.makedirs(self.outputdir, mode=0o777)
             os.umask(old)
-        except OSError, e:
+        except OSError as e:
             fail('%s' % e)
 
         self.result.starttime = time()
         proc = Popen(privcmd, stdout=PIPE, stderr=PIPE)
         # Allow a special timeout value of 0 to mean infinity
         if int(self.timeout) == 0:
-            self.timeout = maxint
+            self.timeout = sys.maxsize
         t = Timer(int(self.timeout), self.kill_cmd, [proc])
 
         try:
@@ -247,50 +255,52 @@ class Cmd(object):
         self.result.runtime = '%02d:%02d' % (m, s)
         self.result.result = 'SKIP'
 
-    def log(self, logger, options):
+    def log(self, options):
         """
         This function is responsible for writing all output. This includes
         the console output, the logfile of all results (with timestamped
         merged stdout and stderr), and for each test, the unmodified
         stdout/stderr/merged in it's own file.
         """
-        if logger is None:
-            return
 
         logname = getpwuid(os.getuid()).pw_name
         user = ' (run as %s)' % (self.user if len(self.user) else logname)
         msga = 'Test: %s%s ' % (self.pathname, user)
-        msgb = '[%s] [%s]' % (self.result.runtime, self.result.result)
+        msgb = '[%s] [%s]\n' % (self.result.runtime, self.result.result)
         pad = ' ' * (80 - (len(msga) + len(msgb)))
+        result_line = msga + pad + msgb
 
-        # If -q is specified, only print a line for tests that didn't pass.
-        # This means passing tests need to be logged as DEBUG, or the one
-        # line summary will only be printed in the logfile for failures.
+        # The result line is always written to the log file. If -q was
+        # specified only failures are written to the console, otherwise
+        # the result line is written to the console.
+        write_log(bytearray(result_line, encoding='utf-8'), LOG_FILE)
         if not options.quiet:
-            logger.info('%s%s%s' % (msga, pad, msgb))
-        elif self.result.result is not 'PASS':
-            logger.info('%s%s%s' % (msga, pad, msgb))
-        else:
-            logger.debug('%s%s%s' % (msga, pad, msgb))
+            write_log(result_line, LOG_OUT)
+        elif options.quiet and self.result.result is not 'PASS':
+            write_log(result_line, LOG_OUT)
 
         lines = sorted(self.result.stdout + self.result.stderr,
-                       cmp=lambda x, y: cmp(x[0], y[0]))
+                       key=lambda x: x[0])
 
+        # Write timestamped output (stdout and stderr) to the logfile
         for dt, line in lines:
-            logger.debug('%s %s' % (dt.strftime("%H:%M:%S.%f ")[:11], line))
+            timestamp = bytearray(dt.strftime("%H:%M:%S.%f ")[:11],
+                                  encoding='utf-8')
+            write_log(b'%s %s\n' % (timestamp, line), LOG_FILE)
 
+        # Write the separate stdout/stderr/merged files, if the data exists
         if len(self.result.stdout):
-            with open(os.path.join(self.outputdir, 'stdout'), 'w') as out:
+            with open(os.path.join(self.outputdir, 'stdout'), 'wb') as out:
                 for _, line in self.result.stdout:
-                    os.write(out.fileno(), '%s\n' % line)
+                    os.write(out.fileno(), b'%s\n' % line)
         if len(self.result.stderr):
-            with open(os.path.join(self.outputdir, 'stderr'), 'w') as err:
+            with open(os.path.join(self.outputdir, 'stderr'), 'wb') as err:
                 for _, line in self.result.stderr:
-                    os.write(err.fileno(), '%s\n' % line)
+                    os.write(err.fileno(), b'%s\n' % line)
         if len(self.result.stdout) and len(self.result.stderr):
-            with open(os.path.join(self.outputdir, 'merged'), 'w') as merged:
+            with open(os.path.join(self.outputdir, 'merged'), 'wb') as merged:
                 for _, line in lines:
-                    os.write(merged.fileno(), '%s\n' % line)
+                    os.write(merged.fileno(), b'%s\n' % line)
 
 
 class Test(Cmd):
@@ -318,7 +328,7 @@ class Test(Cmd):
                (self.pathname, self.outputdir, self.timeout, self.pre,
                 pre_user, self.post, post_user, self.user, self.tags)
 
-    def verify(self, logger):
+    def verify(self):
         """
         Check the pre/post scripts, user and Test. Omit the Test from this
         run if there are any problems.
@@ -328,19 +338,19 @@ class Test(Cmd):
 
         for f in [f for f in files if len(f)]:
             if not verify_file(f):
-                logger.info("Warning: Test '%s' not added to this run because"
-                            " it failed verification." % f)
+                write_log("Warning: Test '%s' not added to this run because"
+                          " it failed verification.\n" % f, LOG_ERR)
                 return False
 
         for user in [user for user in users if len(user)]:
-            if not verify_user(user, logger):
-                logger.info("Not adding Test '%s' to this run." %
-                            self.pathname)
+            if not verify_user(user):
+                write_log("Not adding Test '%s' to this run.\n" %
+                          self.pathname, LOG_ERR)
                 return False
 
         return True
 
-    def run(self, logger, options):
+    def run(self, options):
         """
         Create Cmd instances for the pre/post scripts. If the pre script
         doesn't pass, skip this Test. Run the post script regardless.
@@ -358,18 +368,18 @@ class Test(Cmd):
         if len(pretest.pathname):
             pretest.run(options)
             cont = pretest.result.result is 'PASS'
-            pretest.log(logger, options)
+            pretest.log(options)
 
         if cont:
             test.run(options)
         else:
             test.skip()
 
-        test.log(logger, options)
+        test.log(options)
 
         if len(posttest.pathname):
             posttest.run(options)
-            posttest.log(logger, options)
+            posttest.log(options)
 
 
 class TestGroup(Test):
@@ -393,7 +403,7 @@ class TestGroup(Test):
                (self.pathname, self.outputdir, self.tests, self.timeout,
                 self.pre, pre_user, self.post, post_user, self.user, self.tags)
 
-    def verify(self, logger):
+    def verify(self):
         """
         Check the pre/post scripts, user and tests in this TestGroup. Omit
         the TestGroup entirely, or simply delete the relevant tests in the
@@ -411,34 +421,34 @@ class TestGroup(Test):
 
         for f in [f for f in auxfiles if len(f)]:
             if self.pathname != os.path.dirname(f):
-                logger.info("Warning: TestGroup '%s' not added to this run. "
+                write_log("Warning: TestGroup '%s' not added to this run. "
                           "Auxiliary script '%s' exists in a different "
-                            "directory." % (self.pathname, f))
+                          "directory.\n" % (self.pathname, f), LOG_ERR)
                 return False
 
             if not verify_file(f):
-                logger.info("Warning: TestGroup '%s' not added to this run. "
-                            "Auxiliary script '%s' failed verification." %
-                            (self.pathname, f))
+                write_log("Warning: TestGroup '%s' not added to this run. "
+                          "Auxiliary script '%s' failed verification.\n" %
+                          (self.pathname, f), LOG_ERR)
                 return False
 
         for user in [user for user in users if len(user)]:
-            if not verify_user(user, logger):
-                logger.info("Not adding TestGroup '%s' to this run." %
-                            self.pathname)
+            if not verify_user(user):
+                write_log("Not adding TestGroup '%s' to this run.\n" %
+                          self.pathname, LOG_ERR)
                 return False
 
         # If one of the tests is invalid, delete it, log it, and drive on.
         for test in self.tests:
             if not verify_file(os.path.join(self.pathname, test)):
                 del self.tests[self.tests.index(test)]
-                logger.info("Warning: Test '%s' removed from TestGroup '%s' "
-                            "because it failed verification." %
-                            (test, self.pathname))
+                write_log("Warning: Test '%s' removed from TestGroup '%s' "
+                          "because it failed verification.\n" %
+                          (test, self.pathname), LOG_ERR)
 
         return len(self.tests) is not 0
 
-    def run(self, logger, options):
+    def run(self, options):
         """
         Create Cmd instances for the pre/post scripts. If the pre script
         doesn't pass, skip all the tests in this TestGroup. Run the post
@@ -459,7 +469,7 @@ class TestGroup(Test):
         if len(pretest.pathname):
             pretest.run(options)
             cont = pretest.result.result is 'PASS'
-            pretest.log(logger, options)
+            pretest.log(options)
 
         for fname in self.tests:
             test = Cmd(os.path.join(self.pathname, fname),
@@ -470,11 +480,11 @@ class TestGroup(Test):
             else:
                 test.skip()
 
-            test.log(logger, options)
+            test.log(options)
 
         if len(posttest.pathname):
             posttest.run(options)
-            posttest.log(logger, options)
+            posttest.log(options)
 
 
 class TestRun(object):
@@ -486,7 +496,7 @@ class TestRun(object):
         self.starttime = time()
         self.timestamp = datetime.now().strftime('%Y%m%dT%H%M%S')
         self.outputdir = os.path.join(options.outputdir, self.timestamp)
-        self.logger = self.setup_logging(options)
+        self.setup_logging(options)
         self.defaults = [
             ('outputdir', BASEDIR),
             ('quiet', False),
@@ -519,7 +529,7 @@ class TestRun(object):
         for prop in Test.props:
             setattr(test, prop, getattr(options, prop))
 
-        if test.verify(self.logger):
+        if test.verify():
             self.tests[pathname] = test
 
     def addtestgroup(self, dirname, filenames, options):
@@ -541,9 +551,9 @@ class TestRun(object):
             self.testgroups[dirname] = testgroup
             self.testgroups[dirname].tests = sorted(filenames)
 
-            testgroup.verify(self.logger)
+            testgroup.verify()
 
-    def read(self, logger, options):
+    def read(self, options):
         """
         Read in the specified runfile, and apply the TestRun properties
         listed in the 'DEFAULT' section to our TestRun. Then read each
@@ -552,7 +562,7 @@ class TestRun(object):
         in the 'DEFAULT' section. If the Test or TestGroup passes
         verification, add it to the TestRun.
         """
-        config = ConfigParser.RawConfigParser()
+        config = configparser.RawConfigParser()
         if not len(config.read(options.runfile)):
             fail("Coulnd't read config file %s" % options.runfile)
 
@@ -584,7 +594,7 @@ class TestRun(object):
                 # Repopulate tests using eval to convert the string to a list
                 testgroup.tests = eval(config.get(section, 'tests'))
 
-                if testgroup.verify(logger):
+                if testgroup.verify():
                     self.testgroups[section] = testgroup
             else:
                 test = Test(section)
@@ -593,7 +603,7 @@ class TestRun(object):
                         if config.has_option(sect, prop):
                             setattr(test, prop, config.get(sect, prop))
 
-                if test.verify(logger):
+                if test.verify():
                     self.tests[section] = test
 
     def write(self, options):
@@ -608,7 +618,7 @@ class TestRun(object):
 
         defaults = dict([(prop, getattr(options, prop)) for prop, _ in
                          self.defaults])
-        config = ConfigParser.RawConfigParser(defaults)
+        config = configparser.RawConfigParser(defaults)
 
         for test in sorted(self.tests.keys()):
             config.add_section(test)
@@ -637,14 +647,15 @@ class TestRun(object):
         """
         done = False
         components = 0
-        tmp_dict = dict(self.tests.items() + self.testgroups.items())
+        tmp_dict = dict(list(self.tests.items()) +
+                        list(self.testgroups.items()))
         total = len(tmp_dict)
         base = self.outputdir
 
         while not done:
             paths = []
             components -= 1
-            for testfile in tmp_dict.keys():
+            for testfile in list(tmp_dict.keys()):
                 uniq = '/'.join(testfile.split('/')[components:]).lstrip('/')
                 if uniq not in paths:
                     paths.append(uniq)
@@ -655,42 +666,23 @@ class TestRun(object):
 
     def setup_logging(self, options):
         """
-        Two loggers are set up here. The first is for the logfile which
-        will contain one line summarizing the test, including the test
-        name, result, and running time. This logger will also capture the
-        timestamped combined stdout and stderr of each run. The second
-        logger is optional console output, which will contain only the one
-        line summary. The loggers are initialized at two different levels
-        to facilitate segregating the output.
+        This funtion creates the output directory and gets a file object
+        for the logfile. This function must be called before write_log()
+        can be used.
         """
         if options.dryrun is True:
             return
 
-        testlogger = logging.getLogger(__name__)
-        testlogger.setLevel(logging.DEBUG)
-
+        global LOG_FILE_OBJ
         if options.cmd is not 'wrconfig':
             try:
                 old = os.umask(0)
-                os.makedirs(self.outputdir, mode=0777)
+                os.makedirs(self.outputdir, mode=0o777)
                 os.umask(old)
-            except OSError, e:
-                fail('%s' % e)
                 filename = os.path.join(self.outputdir, 'log')
-
-            logfile = logging.FileHandler(filename)
-            logfile.setLevel(logging.DEBUG)
-            logfilefmt = logging.Formatter('%(message)s')
-            logfile.setFormatter(logfilefmt)
-            testlogger.addHandler(logfile)
-
-        cons = logging.StreamHandler()
-        cons.setLevel(logging.INFO)
-        consfmt = logging.Formatter('%(message)s')
-        cons.setFormatter(consfmt)
-        testlogger.addHandler(cons)
-
-        return testlogger
+                LOG_FILE_OBJ = open(filename, buffering=0, mode='wb')
+            except OSError as e:
+                fail('%s' % e)
 
     def run(self, options):
         """
@@ -707,31 +699,31 @@ class TestRun(object):
         if not os.path.exists(logsymlink):
             os.symlink(self.outputdir, logsymlink)
         else:
-            print 'Could not make a symlink to directory %s' % (
-                self.outputdir)
+            write_log('Could not make a symlink to directory %s\n' %
+                      self.outputdir, LOG_ERR)
         iteration = 0
         while iteration < options.iterations:
             for test in sorted(self.tests.keys()):
-                self.tests[test].run(self.logger, options)
+                self.tests[test].run(options)
             for testgroup in sorted(self.testgroups.keys()):
-                self.testgroups[testgroup].run(self.logger, options)
+                self.testgroups[testgroup].run(options)
             iteration += 1
 
     def summary(self):
         if Result.total is 0:
             return 2
 
-        print '\nResults Summary'
-        for key in Result.runresults.keys():
+        print('\nResults Summary')
+        for key in list(Result.runresults.keys()):
             if Result.runresults[key] is not 0:
-                print '%s\t% 4d' % (key, Result.runresults[key])
+                print('%s\t% 4d' % (key, Result.runresults[key]))
 
         m, s = divmod(time() - self.starttime, 60)
         h, m = divmod(m, 60)
-        print '\nRunning Time:\t%02d:%02d:%02d' % (h, m, s)
-        print 'Percent passed:\t%.1f%%' % ((float(Result.runresults['PASS']) /
-                                            float(Result.total)) * 100)
-        print 'Log directory:\t%s' % self.outputdir
+        print('\nRunning Time:\t%02d:%02d:%02d' % (h, m, s))
+        print('Percent passed:\t%.1f%%' % ((float(Result.runresults['PASS']) /
+                                            float(Result.total)) * 100))
+        print('Log directory:\t%s' % self.outputdir)
 
         if Result.runresults['FAIL'] > 0:
             return 1
@@ -742,6 +734,23 @@ class TestRun(object):
         return 0
 
 
+def write_log(msg, target):
+    """
+    Write the provided message to standard out, standard error or
+    the logfile. If specifying LOG_FILE, then `msg` must be a bytes
+    like object. This way we can still handle output from tests that
+    may be in unexpected encodings.
+    """
+    if target == LOG_OUT:
+        os.write(sys.stdout.fileno(), bytearray(msg, encoding='utf-8'))
+    elif target == LOG_ERR:
+        os.write(sys.stderr.fileno(), bytearray(msg, encoding='utf-8'))
+    elif target == LOG_FILE:
+        os.write(LOG_FILE_OBJ.fileno(), msg)
+    else:
+        fail('log_msg called with unknown target "%s"' % target)
+
+
 def verify_file(pathname):
     """
     Verify that the supplied pathname is an executable regular file.
@@ -757,7 +766,7 @@ def verify_file(pathname):
     return False
 
 
-def verify_user(user, logger):
+def verify_user(user):
     """
     Verify that the specified user exists on this system, and can execute
     sudo without being prompted for a password.
@@ -770,13 +779,15 @@ def verify_user(user, logger):
     try:
         getpwnam(user)
     except KeyError:
-        logger.info("Warning: user '%s' does not exist.", user)
+        write_log("Warning: user '%s' does not exist.\n" % user,
+                  LOG_ERR)
         return False
 
     p = Popen(testcmd)
     p.wait()
     if p.returncode is not 0:
-        logger.info("Warning: user '%s' cannot use passwordless sudo.", user)
+        write_log("Warning: user '%s' cannot use passwordless sudo.\n" % user,
+                  LOG_ERR)
         return False
     else:
         Cmd.verified_users.append(user)
@@ -804,7 +815,7 @@ def find_tests(testrun, options):
 
 
 def fail(retstr, ret=1):
-    print '%s: %s' % (argv[0], retstr)
+    print('%s: %s' % (sys.argv[0], retstr))
     exit(ret)
 
 
@@ -894,7 +905,7 @@ def main():
     if options.cmd is 'runtests':
         find_tests(testrun, options)
     elif options.cmd is 'rdconfig':
-        testrun.read(testrun.logger, options)
+        testrun.read(options)
     elif options.cmd is 'wrconfig':
         find_tests(testrun, options)
         testrun.write(options)

tests/zfs-tests/cmd/mmapwrite/mmapwrite.c

@@ -31,74 +31,132 @@
 #include <string.h>
 #include <sys/mman.h>
 #include <pthread.h>
+#include <errno.h>
+#include <err.h>
 
 /*
  * --------------------------------------------------------------------
- * Bug Id: 5032643
+ * Bug Issue Id: #7512
+ * The bug time sequence:
+ * 1. context #1, zfs_write assign a txg "n".
+ * 2. In the same process, context #2, mmap page fault (which means the mm_sem
+ *    is hold) occurred, zfs_dirty_inode open a txg failed, and wait previous
+ *    txg "n" completed.
+ * 3. context #1 call uiomove to write, however page fault is occurred in
+ *    uiomove, which means it need mm_sem, but mm_sem is hold by
+ *    context #2, so it stuck and can't complete, then txg "n" will not
+ *    complete.
  *
- * Simply writing to a file and mmaping that file at the same time can
- * result in deadlock.  Nothing perverse like writing from the file's
- * own mapping is required.
+ * So context #1 and context #2 trap into the "dead lock".
  * --------------------------------------------------------------------
  */
 
-static void *
-mapper(void *fdp)
-{
-	void *addr;
-	int fd = *(int *)fdp;
+#define	NORMAL_WRITE_TH_NUM	2
 
-	if ((addr =
-	    mmap(0, 8192, PROT_READ, MAP_SHARED, fd, 0)) == MAP_FAILED) {
-		perror("mmap");
-		exit(1);
+static void *
+normal_writer(void *filename)
+{
+	char *file_path = filename;
+	int fd = -1;
+	ssize_t write_num = 0;
+	int page_size = getpagesize();
+
+	fd = open(file_path, O_RDWR | O_CREAT, 0777);
+	if (fd == -1) {
+		err(1, "failed to open %s", file_path);
 	}
-	for (;;) {
-		if (mmap(addr, 8192, PROT_READ,
-		    MAP_SHARED|MAP_FIXED, fd, 0) == MAP_FAILED) {
-			perror("mmap");
-			exit(1);
+
+	char *buf = malloc(1);
+	while (1) {
+		write_num = write(fd, buf, 1);
+		if (write_num == 0) {
+			err(1, "write failed!");
+			break;
+		}
+		lseek(fd, page_size, SEEK_CUR);
+	}
+
+	if (buf) {
+		free(buf);
+	}
+}
+
+static void *
+map_writer(void *filename)
+{
+	int fd = -1;
+	int ret = 0;
+	char *buf = NULL;
+	int page_size = getpagesize();
+	int op_errno = 0;
+	char *file_path = filename;
+
+	while (1) {
+		ret = access(file_path, F_OK);
+		if (ret) {
+			op_errno = errno;
+			if (op_errno == ENOENT) {
+				fd = open(file_path, O_RDWR | O_CREAT, 0777);
+				if (fd == -1) {
+					err(1, "open file failed");
+				}
+
+				ret = ftruncate(fd, page_size);
+				if (ret == -1) {
+					err(1, "truncate file failed");
+				}
+			} else {
+				err(1, "access file failed!");
+			}
+		} else {
+			fd = open(file_path, O_RDWR, 0777);
+			if (fd == -1) {
+				err(1, "open file failed");
+			}
+		}
+
+		if ((buf = mmap(NULL, page_size, PROT_READ | PROT_WRITE,
+		    MAP_SHARED, fd, 0)) == MAP_FAILED) {
+			err(1, "map file failed");
+		}
+
+		if (fd != -1)
+			close(fd);
+
+		char s[10] = {0, };
+		memcpy(buf, s, 10);
+		ret = munmap(buf, page_size);
+		if (ret != 0) {
+			err(1, "unmap file failed");
 		}
 	}
-	/* NOTREACHED */
-	return ((void *)1);
 }
 
 int
 main(int argc, char **argv)
 {
-	int fd;
-	char buf[1024];
-	pthread_t tid;
+	pthread_t map_write_tid;
+	pthread_t normal_write_tid[NORMAL_WRITE_TH_NUM];
+	int i = 0;
 
-	memset(buf, 'a', sizeof (buf));
-
-	if (argc != 2) {
-		(void) printf("usage: %s <file name>\n", argv[0]);
+	if (argc != 3) {
+		(void) printf("usage: %s <normal write file name>"
+		    "<map write file name>\n", argv[0]);
 		exit(1);
 	}
 
-	if ((fd = open(argv[1], O_RDWR|O_CREAT|O_TRUNC, 0666)) == -1) {
-		perror("open");
-		exit(1);
-	}
-
-	(void) pthread_setconcurrency(2);
-	if (pthread_create(&tid, NULL, mapper, &fd) != 0) {
-		perror("pthread_create");
-		close(fd);
-		exit(1);
-	}
-	for (;;) {
-		if (write(fd, buf, sizeof (buf)) == -1) {
-			perror("write");
-			close(fd);
-			exit(1);
+	for (i = 0; i < NORMAL_WRITE_TH_NUM; i++) {
+		if (pthread_create(&normal_write_tid[i], NULL, normal_writer,
+		    argv[1])) {
+			err(1, "pthread_create normal_writer failed.");
 		}
 	}
 
-	close(fd);
+	if (pthread_create(&map_write_tid, NULL, map_writer, argv[2])) {
+		err(1, "pthread_create map_writer failed.");
+	}
 
 	/* NOTREACHED */
+	pthread_join(map_write_tid, NULL);
 	return (0);
 }

tests/zfs-tests/tests/functional/mmap/mmap_write_001_pos.ksh

@@ -53,12 +53,14 @@ if ! is_mp; then
 fi
 
 log_must chmod 777 $TESTDIR
-mmapwrite $TESTDIR/test-write-file &
+mmapwrite $TESTDIR/normal_write_file $TESTDIR/map_write_file &
 PID_MMAPWRITE=$!
-log_note "mmapwrite $TESTDIR/test-write-file pid: $PID_MMAPWRITE"
+log_note "mmapwrite $TESTDIR/normal_write_file $TESTDIR/map_write_file"\
+	 "pid: $PID_MMAPWRITE"
 log_must sleep 30
 
 log_must kill -9 $PID_MMAPWRITE
-log_must ls -l $TESTDIR/test-write-file
+log_must ls -l $TESTDIR/normal_write_file
+log_must ls -l $TESTDIR/map_write_file
 
 log_pass "write(2) a mmap(2)'ing file succeeded."

udev/rules.d/69-vdev.rules.in

@@ -8,3 +8,7 @@ ENV{DEVTYPE}=="partition", IMPORT{program}="@udevdir@/vdev_id -d %k"
 KERNEL=="*[!0-9]", ENV{SUBSYSTEM}=="block", ENV{ID_VDEV}=="?*", SYMLINK+="$env{ID_VDEV_PATH}"
 KERNEL=="*[0-9]", ENV{SUBSYSTEM}=="block", ENV{DEVTYPE}=="partition", ENV{ID_VDEV}=="?*", SYMLINK+="$env{ID_VDEV_PATH}-part%n"
 KERNEL=="dm-[0-9]*", ENV{SUBSYSTEM}=="block", ENV{ID_VDEV}=="?*", SYMLINK+="$env{ID_VDEV_PATH}"
+
+# Enclosure device symlink rules
+ENV{SUBSYSTEM}=="scsi_generic", IMPORT{program}="@udevdir@/vdev_id -e"
+ENV{SUBSYSTEM}=="scsi_generic", ENV{ID_ENCLOSURE}=="?*", SYMLINK+="$env{ID_ENCLOSURE_PATH}"

udev/rules.d/90-zfs.rules.in

@@ -7,6 +7,6 @@ ENV{ID_FS_TYPE}=="zfs_member", RUN+="/sbin/modprobe zfs"
 KERNEL=="null", SYMLINK+="root"
 SYMLINK=="null", SYMLINK+="root"
 
-SUBSYSTEM=="misc", KERNEL=="zfs", MODE="0666"
+KERNEL=="zfs", MODE="0666", OPTIONS+="static_node=zfs"
 
 LABEL="zfs_end"