Tag zfs-2.1.5

META file and changelog updated.

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

META

@@ -1,10 +1,10 @@
 Meta:          1
 Name:          zfs
 Branch:        1.0
-Version:       2.1.4
+Version:       2.1.5
 Release:       1
 Release-Tags:  relext
 License:       CDDL
 Author:        OpenZFS
-Linux-Maximum: 5.17
+Linux-Maximum: 5.18
 Linux-Minimum: 3.10

Makefile.am

@@ -103,7 +103,7 @@ endif
 endif
 
 PHONY += codecheck
-codecheck: cstyle shellcheck checkbashisms flake8 mancheck testscheck vcscheck
+codecheck: cstyle shellcheck checkbashisms flake8 mancheck testscheck vcscheck zstdcheck
 
 PHONY += checkstyle
 checkstyle: codecheck commitcheck
@@ -120,6 +120,7 @@ cstyle:
 		-o -type f -name '*.[hc]' \
 		! -name 'zfs_config.*' ! -name '*.mod.c' \
 		! -name 'opt_global.h' ! -name '*_if*.h' \
+		! -name 'zstd_compat_wrapper.h' \
 		! -path './module/zstd/lib/*' \
 		-exec ${top_srcdir}/scripts/cstyle.pl -cpP {} \+
 
@@ -173,6 +174,10 @@ vcscheck:
 		awk '{c++; print} END {if(c>0) exit 1}' ; \
 	fi
 
+PHONY += zstdcheck
+zstdcheck:
+	@$(MAKE) -C module/zstd checksymbols
+
 PHONY += lint
 lint: cppcheck paxcheck
 

cmd/zdb/zdb.c

@@ -110,6 +110,7 @@ extern int zfs_recover;
 extern unsigned long zfs_arc_meta_min, zfs_arc_meta_limit;
 extern int zfs_vdev_async_read_max_active;
 extern boolean_t spa_load_verify_dryrun;
+extern boolean_t spa_mode_readable_spacemaps;
 extern int zfs_reconstruct_indirect_combinations_max;
 extern int zfs_btree_verify_intensity;
 
@@ -3124,13 +3125,18 @@ dump_znode_symlink(sa_handle_t *hdl)
 {
 	int sa_symlink_size = 0;
 	char linktarget[MAXPATHLEN];
-	linktarget[0] = '\0';
 	int error;
 
 	error = sa_size(hdl, sa_attr_table[ZPL_SYMLINK], &sa_symlink_size);
 	if (error || sa_symlink_size == 0) {
 		return;
 	}
+	if (sa_symlink_size >= sizeof (linktarget)) {
+		(void) printf("symlink size %d is too large\n",
+		    sa_symlink_size);
+		return;
+	}
+	linktarget[sa_symlink_size] = '\0';
 	if (sa_lookup(hdl, sa_attr_table[ZPL_SYMLINK],
 	    &linktarget, sa_symlink_size) == 0)
 		(void) printf("\ttarget	%s\n", linktarget);
@@ -8469,6 +8475,11 @@ main(int argc, char **argv)
 	 */
 	spa_load_verify_dryrun = B_TRUE;
 
+	/*
+	 * ZDB should have ability to read spacemaps.
+	 */
+	spa_mode_readable_spacemaps = B_TRUE;
+
 	kernel_init(SPA_MODE_READ);
 
 	if (dump_all)

cmd/zed/agents/zfs_diagnosis.c

@@ -35,6 +35,7 @@
 #include <sys/fs/zfs.h>
 #include <sys/fm/protocol.h>
 #include <sys/fm/fs/zfs.h>
+#include <sys/zio.h>
 
 #include "zfs_agents.h"
 #include "fmd_api.h"
@@ -773,6 +774,8 @@ zfs_fm_recv(fmd_hdl_t *hdl, fmd_event_t *ep, nvlist_t *nvl, const char *class)
 	    ZFS_MAKE_EREPORT(FM_EREPORT_ZFS_PROBE_FAILURE))) {
 		char *failmode = NULL;
 		boolean_t checkremove = B_FALSE;
+		uint32_t pri = 0;
+		int32_t flags = 0;
 
 		/*
 		 * If this is a checksum or I/O error, then toss it into the
@@ -795,6 +798,23 @@ zfs_fm_recv(fmd_hdl_t *hdl, fmd_event_t *ep, nvlist_t *nvl, const char *class)
 				checkremove = B_TRUE;
 		} else if (fmd_nvl_class_match(hdl, nvl,
 		    ZFS_MAKE_EREPORT(FM_EREPORT_ZFS_CHECKSUM))) {
+			/*
+			 * We ignore ereports for checksum errors generated by
+			 * scrub/resilver I/O to avoid potentially further
+			 * degrading the pool while it's being repaired.
+			 */
+			if (((nvlist_lookup_uint32(nvl,
+			    FM_EREPORT_PAYLOAD_ZFS_ZIO_PRIORITY, &pri) == 0) &&
+			    (pri == ZIO_PRIORITY_SCRUB ||
+			    pri == ZIO_PRIORITY_REBUILD)) ||
+			    ((nvlist_lookup_int32(nvl,
+			    FM_EREPORT_PAYLOAD_ZFS_ZIO_FLAGS, &flags) == 0) &&
+			    (flags & (ZIO_FLAG_SCRUB | ZIO_FLAG_RESILVER)))) {
+				fmd_hdl_debug(hdl, "ignoring '%s' for "
+				    "scrub/resilver I/O", class);
+				return;
+			}
+
 			if (zcp->zc_data.zc_serd_checksum[0] == '\0') {
 				zfs_serd_name(zcp->zc_data.zc_serd_checksum,
 				    pool_guid, vdev_guid, "checksum");

cmd/zed/zed.d/zed-functions.sh

@@ -224,6 +224,8 @@ zed_notify()
 # ZED_EMAIL_OPTS.  This undergoes the following keyword substitutions:
 # - @ADDRESS@ is replaced with the space-delimited recipient email address(es)
 # - @SUBJECT@ is replaced with the notification subject
+#   If @SUBJECT@ was omited here, a "Subject: ..." header will be added to notification
+#
 #
 # Arguments
 #   subject: notification subject
@@ -241,7 +243,7 @@ zed_notify()
 #
 zed_notify_email()
 {
-    local subject="$1"
+    local subject="${1:-"ZED notification"}"
     local pathname="${2:-"/dev/null"}"
 
     : "${ZED_EMAIL_PROG:="mail"}"
@@ -262,12 +264,23 @@ zed_notify_email()
         return 1
     fi
 
-    ZED_EMAIL_OPTS="$(echo "${ZED_EMAIL_OPTS}" \
+    # construct cmdline options
+    ZED_EMAIL_OPTS_PARSED="$(echo "${ZED_EMAIL_OPTS}" \
         | sed   -e "s/@ADDRESS@/${ZED_EMAIL_ADDR}/g" \
                 -e "s/@SUBJECT@/${subject}/g")"
 
-    # shellcheck disable=SC2086
-    eval ${ZED_EMAIL_PROG} ${ZED_EMAIL_OPTS} < "${pathname}" >/dev/null 2>&1
+    # pipe message to email prog
+    # shellcheck disable=SC2086,SC2248
+    {
+        # no subject passed as option?
+        if [ "${ZED_EMAIL_OPTS%@SUBJECT@*}" = "${ZED_EMAIL_OPTS}" ] ; then
+            # inject subject header
+            printf "Subject: %s\n" "${subject}"
+        fi
+        # output message
+        cat "${pathname}"
+    } |
+    eval ${ZED_EMAIL_PROG} ${ZED_EMAIL_OPTS_PARSED} >/dev/null 2>&1
     rv=$?
     if [ "${rv}" -ne 0 ]; then
         zed_log_err "${ZED_EMAIL_PROG##*/} exit=${rv}"

cmd/zed/zed.d/zed.rc

@@ -30,6 +30,7 @@ ZED_EMAIL_ADDR="root"
 # The string @SUBJECT@ will be replaced with the notification subject;
 #   this should be protected with quotes to prevent word-splitting.
 # Email will only be sent if ZED_EMAIL_ADDR is defined.
+# If @SUBJECT@ was omited here, a "Subject: ..." header will be added to notification
 #
 #ZED_EMAIL_OPTS="-s '@SUBJECT@' @ADDRESS@"
 

cmd/zfs/zfs_main.c

@@ -6593,7 +6593,7 @@ zfs_do_holds(int argc, char **argv)
 		/*
 		 *  1. collect holds data, set format options
 		 */
-		ret = zfs_for_each(argc, argv, flags, types, NULL, NULL, limit,
+		ret = zfs_for_each(1, argv + i, flags, types, NULL, NULL, limit,
 		    holds_callback, &cb);
 		if (ret != 0)
 			++errors;
@@ -7672,7 +7672,7 @@ zfs_do_diff(int argc, char **argv)
 	int c;
 	struct sigaction sa;
 
-	while ((c = getopt(argc, argv, "FHt")) != -1) {
+	while ((c = getopt(argc, argv, "FHth")) != -1) {
 		switch (c) {
 		case 'F':
 			flags |= ZFS_DIFF_CLASSIFY;
@@ -7683,6 +7683,9 @@ zfs_do_diff(int argc, char **argv)
 		case 't':
 			flags |= ZFS_DIFF_TIMESTAMP;
 			break;
+		case 'h':
+			flags |= ZFS_DIFF_NO_MANGLE;
+			break;
 		default:
 			(void) fprintf(stderr,
 			    gettext("invalid option '%c'\n"), optopt);

cmd/zvol_wait/zvol_wait

@@ -28,15 +28,17 @@ filter_out_deleted_zvols() {
 list_zvols() {
 	read -r default_volmode < /sys/module/zfs/parameters/zvol_volmode
 	zfs list -t volume -H -o \
-	    name,volmode,receive_resume_token,redact_snaps |
-	    while IFS="	" read -r name volmode token redacted; do # IFS=\t here!
+	    name,volmode,receive_resume_token,redact_snaps,keystatus |
+	    while IFS="	" read -r name volmode token redacted keystatus; do # IFS=\t here!
 
-		# /dev links are not created for zvols with volmode = "none"
-		# or for redacted zvols.
+		# /dev links are not created for zvols with volmode = "none",
+		# redacted zvols, or encrypted zvols for which the key has not
+		# been loaded.
 		[ "$volmode" = "none" ] && continue
 		[ "$volmode" = "default" ] && [ "$default_volmode" = "3" ] &&
 		    continue
 		[ "$redacted" = "-" ] || continue
+		[ "$keystatus" = "unavailable" ] && continue
 
 		# We also ignore partially received zvols if it is
 		# not an incremental receive, as those won't even have a block

config/ax_python_devel.m4

@@ -224,7 +224,7 @@ EOD`
 				ac_python_version=$PYTHON_VERSION
 			else
 				ac_python_version=`$PYTHON -c "import sys; \
-					print (sys.version[[:3]])"`
+					print ('.'.join(sys.version.split('.')[[:2]]))"`
 			fi
 		fi
 

config/kernel-add-disk.m4

@@ -3,16 +3,15 @@ dnl # 5.16 API change
 dnl # add_disk grew a must-check return code
 dnl #
 AC_DEFUN([ZFS_AC_KERNEL_SRC_ADD_DISK], [
-
 	ZFS_LINUX_TEST_SRC([add_disk_ret], [
-		#include <linux/genhd.h>
+		#include <linux/blkdev.h>
 	], [
 		struct gendisk *disk = NULL;
 		int err = add_disk(disk);
 		err = err;
 	])
-
 ])
+
 AC_DEFUN([ZFS_AC_KERNEL_ADD_DISK], [
 	AC_MSG_CHECKING([whether add_disk() returns int])
 	ZFS_LINUX_TEST_RESULT([add_disk_ret],

config/kernel-bio.m4

@@ -464,7 +464,7 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_BLK_CGROUP_HEADER], [
 ])
 
 AC_DEFUN([ZFS_AC_KERNEL_BLK_CGROUP_HEADER], [
-	AC_MSG_CHECKING([for existence of linux/blk-cgroup.h])
+	AC_MSG_CHECKING([whether linux/blk-cgroup.h exists])
 	ZFS_LINUX_TEST_RESULT([blk_cgroup_header],[
 		AC_MSG_RESULT(yes)
 		AC_DEFINE(HAVE_LINUX_BLK_CGROUP_HEADER, 1,
@@ -474,6 +474,41 @@ AC_DEFUN([ZFS_AC_KERNEL_BLK_CGROUP_HEADER], [
 	])
 ])
 
+dnl #
+dnl # Linux 5.18 API
+dnl #
+dnl # In 07888c665b405b1cd3577ddebfeb74f4717a84c4 ("block: pass a block_device and opf to bio_alloc")
+dnl #   bio_alloc(gfp_t gfp_mask, unsigned short nr_iovecs)
+dnl # became
+dnl #   bio_alloc(struct block_device *bdev, unsigned short nr_vecs, unsigned int opf, gfp_t gfp_mask)
+dnl # however
+dnl # > NULL/0 can be passed, both for the
+dnl # > passthrough case on a raw request_queue and to temporarily avoid
+dnl # > refactoring some nasty code.
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_SRC_BIO_ALLOC_4ARG], [
+	ZFS_LINUX_TEST_SRC([bio_alloc_4arg], [
+		#include <linux/bio.h>
+	],[
+		gfp_t gfp_mask = 0;
+		unsigned short nr_iovecs = 0;
+		struct block_device *bdev = NULL;
+		unsigned int opf = 0;
+
+		struct bio *__attribute__((unused)) allocated = bio_alloc(bdev, nr_iovecs, opf, gfp_mask);
+	])
+])
+
+AC_DEFUN([ZFS_AC_KERNEL_BIO_ALLOC_4ARG], [
+	AC_MSG_CHECKING([whether bio_alloc() wants 4 args])
+	ZFS_LINUX_TEST_RESULT([bio_alloc_4arg],[
+		AC_MSG_RESULT(yes)
+		AC_DEFINE([HAVE_BIO_ALLOC_4ARG], 1, [bio_alloc() takes 4 arguments])
+	],[
+		AC_MSG_RESULT(no)
+	])
+])
+
 AC_DEFUN([ZFS_AC_KERNEL_SRC_BIO], [
 	ZFS_AC_KERNEL_SRC_REQ
 	ZFS_AC_KERNEL_SRC_BIO_OPS
@@ -488,6 +523,7 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_BIO], [
 	ZFS_AC_KERNEL_SRC_BDEV_SUBMIT_BIO_RETURNS_VOID
 	ZFS_AC_KERNEL_SRC_BIO_SET_DEV_MACRO
 	ZFS_AC_KERNEL_SRC_BLK_CGROUP_HEADER
+	ZFS_AC_KERNEL_SRC_BIO_ALLOC_4ARG
 ])
 
 AC_DEFUN([ZFS_AC_KERNEL_BIO], [
@@ -512,4 +548,5 @@ AC_DEFUN([ZFS_AC_KERNEL_BIO], [
 	ZFS_AC_KERNEL_BIO_BDEV_DISK
 	ZFS_AC_KERNEL_BDEV_SUBMIT_BIO_RETURNS_VOID
 	ZFS_AC_KERNEL_BLK_CGROUP_HEADER
+	ZFS_AC_KERNEL_BIO_ALLOC_4ARG
 ])

config/kernel-blk-queue.m4

@@ -74,6 +74,8 @@ AC_DEFUN([ZFS_AC_KERNEL_BLK_QUEUE_UPDATE_READAHEAD], [
 		AC_DEFINE(HAVE_BLK_QUEUE_UPDATE_READAHEAD, 1,
 		    [blk_queue_update_readahead() exists])
 	],[
+		AC_MSG_RESULT(no)
+
 		AC_MSG_CHECKING([whether disk_update_readahead() exists])
 		ZFS_LINUX_TEST_RESULT([disk_update_readahead], [
 			AC_MSG_RESULT(yes)
@@ -86,69 +88,111 @@ AC_DEFUN([ZFS_AC_KERNEL_BLK_QUEUE_UPDATE_READAHEAD], [
 ])
 
 dnl #
-dnl # 2.6.32 API,
-dnl #   blk_queue_discard()
+dnl # 5.19: bdev_max_discard_sectors() available
+dnl # 2.6.32: blk_queue_discard() available
 dnl #
 AC_DEFUN([ZFS_AC_KERNEL_SRC_BLK_QUEUE_DISCARD], [
+	ZFS_LINUX_TEST_SRC([bdev_max_discard_sectors], [
+		#include <linux/blkdev.h>
+	],[
+		struct block_device *bdev __attribute__ ((unused)) = NULL;
+		unsigned int error __attribute__ ((unused));
+
+		error = bdev_max_discard_sectors(bdev);
+	])
+
 	ZFS_LINUX_TEST_SRC([blk_queue_discard], [
 		#include <linux/blkdev.h>
 	],[
-		struct request_queue *q __attribute__ ((unused)) = NULL;
+		struct request_queue r;
+		struct request_queue *q = &r;
 		int value __attribute__ ((unused));
+		memset(q, 0, sizeof(r));
 		value = blk_queue_discard(q);
 	])
 ])
 
 AC_DEFUN([ZFS_AC_KERNEL_BLK_QUEUE_DISCARD], [
-	AC_MSG_CHECKING([whether blk_queue_discard() is available])
-	ZFS_LINUX_TEST_RESULT([blk_queue_discard], [
+	AC_MSG_CHECKING([whether bdev_max_discard_sectors() is available])
+	ZFS_LINUX_TEST_RESULT([bdev_max_discard_sectors], [
 		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_BDEV_MAX_DISCARD_SECTORS, 1,
+		    [bdev_max_discard_sectors() is available])
 	],[
-		ZFS_LINUX_TEST_ERROR([blk_queue_discard])
+		AC_MSG_RESULT(no)
+
+		AC_MSG_CHECKING([whether blk_queue_discard() is available])
+		ZFS_LINUX_TEST_RESULT([blk_queue_discard], [
+			AC_MSG_RESULT(yes)
+			AC_DEFINE(HAVE_BLK_QUEUE_DISCARD, 1,
+			    [blk_queue_discard() is available])
+		],[
+			ZFS_LINUX_TEST_ERROR([blk_queue_discard])
+		])
 	])
 ])
 
 dnl #
-dnl # 4.8 API,
-dnl #   blk_queue_secure_erase()
-dnl #
-dnl # 2.6.36 - 4.7 API,
-dnl #   blk_queue_secdiscard()
+dnl # 5.19: bdev_max_secure_erase_sectors() available
+dnl # 4.8: blk_queue_secure_erase() available
+dnl # 2.6.36: blk_queue_secdiscard() available
 dnl #
 AC_DEFUN([ZFS_AC_KERNEL_SRC_BLK_QUEUE_SECURE_ERASE], [
+	ZFS_LINUX_TEST_SRC([bdev_max_secure_erase_sectors], [
+		#include <linux/blkdev.h>
+	],[
+		struct block_device *bdev __attribute__ ((unused)) = NULL;
+		unsigned int error __attribute__ ((unused));
+
+		error = bdev_max_secure_erase_sectors(bdev);
+	])
+
 	ZFS_LINUX_TEST_SRC([blk_queue_secure_erase], [
 		#include <linux/blkdev.h>
 	],[
-		struct request_queue *q __attribute__ ((unused)) = NULL;
+		struct request_queue r;
+		struct request_queue *q = &r;
 		int value __attribute__ ((unused));
+		memset(q, 0, sizeof(r));
 		value = blk_queue_secure_erase(q);
 	])
 
 	ZFS_LINUX_TEST_SRC([blk_queue_secdiscard], [
 		#include <linux/blkdev.h>
 	],[
-		struct request_queue *q __attribute__ ((unused)) = NULL;
+		struct request_queue r;
+		struct request_queue *q = &r;
 		int value __attribute__ ((unused));
+		memset(q, 0, sizeof(r));
 		value = blk_queue_secdiscard(q);
 	])
 ])
 
 AC_DEFUN([ZFS_AC_KERNEL_BLK_QUEUE_SECURE_ERASE], [
-	AC_MSG_CHECKING([whether blk_queue_secure_erase() is available])
-	ZFS_LINUX_TEST_RESULT([blk_queue_secure_erase], [
+	AC_MSG_CHECKING([whether bdev_max_secure_erase_sectors() is available])
+	ZFS_LINUX_TEST_RESULT([bdev_max_secure_erase_sectors], [
 		AC_MSG_RESULT(yes)
-		AC_DEFINE(HAVE_BLK_QUEUE_SECURE_ERASE, 1,
-		    [blk_queue_secure_erase() is available])
+		AC_DEFINE(HAVE_BDEV_MAX_SECURE_ERASE_SECTORS, 1,
+		    [bdev_max_secure_erase_sectors() is available])
 	],[
 		AC_MSG_RESULT(no)
 
-		AC_MSG_CHECKING([whether blk_queue_secdiscard() is available])
-		ZFS_LINUX_TEST_RESULT([blk_queue_secdiscard], [
+		AC_MSG_CHECKING([whether blk_queue_secure_erase() is available])
+		ZFS_LINUX_TEST_RESULT([blk_queue_secure_erase], [
 			AC_MSG_RESULT(yes)
-			AC_DEFINE(HAVE_BLK_QUEUE_SECDISCARD, 1,
-			    [blk_queue_secdiscard() is available])
+			AC_DEFINE(HAVE_BLK_QUEUE_SECURE_ERASE, 1,
+			    [blk_queue_secure_erase() is available])
 		],[
-			ZFS_LINUX_TEST_ERROR([blk_queue_secure_erase])
+			AC_MSG_RESULT(no)
+
+			AC_MSG_CHECKING([whether blk_queue_secdiscard() is available])
+			ZFS_LINUX_TEST_RESULT([blk_queue_secdiscard], [
+				AC_MSG_RESULT(yes)
+			AC_DEFINE(HAVE_BLK_QUEUE_SECDISCARD, 1,
+				    [blk_queue_secdiscard() is available])
+			],[
+				ZFS_LINUX_TEST_ERROR([blk_queue_secure_erase])
+			])
 		])
 	])
 ])

config/kernel-blkdev.m4

@@ -294,6 +294,57 @@ AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_BDEV_WHOLE], [
 	])
 ])
 
+dnl #
+dnl # 5.19 API: blkdev_issue_secure_erase()
+dnl # 3.10 API: blkdev_issue_discard(..., BLKDEV_DISCARD_SECURE)
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_ISSUE_SECURE_ERASE], [
+	ZFS_LINUX_TEST_SRC([blkdev_issue_secure_erase], [
+		#include <linux/blkdev.h>
+	],[
+		struct block_device *bdev = NULL;
+		sector_t sector = 0;
+		sector_t nr_sects = 0;
+		int error __attribute__ ((unused));
+
+		error = blkdev_issue_secure_erase(bdev,
+		    sector, nr_sects, GFP_KERNEL);
+	])
+
+	ZFS_LINUX_TEST_SRC([blkdev_issue_discard_flags], [
+		#include <linux/blkdev.h>
+	],[
+		struct block_device *bdev = NULL;
+		sector_t sector = 0;
+		sector_t nr_sects = 0;
+		unsigned long flags = 0;
+		int error __attribute__ ((unused));
+
+		error = blkdev_issue_discard(bdev,
+		    sector, nr_sects, GFP_KERNEL, flags);
+	])
+])
+
+AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_ISSUE_SECURE_ERASE], [
+	AC_MSG_CHECKING([whether blkdev_issue_secure_erase() is available])
+	ZFS_LINUX_TEST_RESULT([blkdev_issue_secure_erase], [
+		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_BLKDEV_ISSUE_SECURE_ERASE, 1,
+		    [blkdev_issue_secure_erase() is available])
+	],[
+		AC_MSG_RESULT(no)
+
+		AC_MSG_CHECKING([whether blkdev_issue_discard() is available])
+		ZFS_LINUX_TEST_RESULT([blkdev_issue_discard_flags], [
+			AC_MSG_RESULT(yes)
+			AC_DEFINE(HAVE_BLKDEV_ISSUE_DISCARD, 1,
+			    [blkdev_issue_discard() is available])
+		],[
+			ZFS_LINUX_TEST_ERROR([blkdev_issue_discard()])
+		])
+	])
+])
+
 dnl #
 dnl # 5.13 API change
 dnl # blkdev_get_by_path() no longer handles ERESTARTSYS
@@ -326,6 +377,7 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV], [
 	ZFS_AC_KERNEL_SRC_BLKDEV_CHECK_DISK_CHANGE
 	ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_CHECK_MEDIA_CHANGE
 	ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_WHOLE
+	ZFS_AC_KERNEL_SRC_BLKDEV_ISSUE_SECURE_ERASE
 ])
 
 AC_DEFUN([ZFS_AC_KERNEL_BLKDEV], [
@@ -340,4 +392,5 @@ AC_DEFUN([ZFS_AC_KERNEL_BLKDEV], [
 	ZFS_AC_KERNEL_BLKDEV_BDEV_CHECK_MEDIA_CHANGE
 	ZFS_AC_KERNEL_BLKDEV_BDEV_WHOLE
 	ZFS_AC_KERNEL_BLKDEV_GET_ERESTARTSYS
+	ZFS_AC_KERNEL_BLKDEV_ISSUE_SECURE_ERASE
 ])

config/kernel-config-defined.m4

@@ -19,19 +19,48 @@ AC_DEFUN([ZFS_AC_KERNEL_CONFIG_DEFINED], [
 		])
 	])
 
+	ZFS_AC_KERNEL_SRC_CONFIG_MODULES
+	ZFS_AC_KERNEL_SRC_CONFIG_BLOCK
 	ZFS_AC_KERNEL_SRC_CONFIG_DEBUG_LOCK_ALLOC
 	ZFS_AC_KERNEL_SRC_CONFIG_TRIM_UNUSED_KSYMS
-	ZFS_AC_KERNEL_SRC_CONFIG_ZLIB_INFLATE
 	ZFS_AC_KERNEL_SRC_CONFIG_ZLIB_DEFLATE
+	ZFS_AC_KERNEL_SRC_CONFIG_ZLIB_INFLATE
 
 	AC_MSG_CHECKING([for kernel config option compatibility])
 	ZFS_LINUX_TEST_COMPILE_ALL([config])
 	AC_MSG_RESULT([done])
 
+	ZFS_AC_KERNEL_CONFIG_MODULES
+	ZFS_AC_KERNEL_CONFIG_BLOCK
 	ZFS_AC_KERNEL_CONFIG_DEBUG_LOCK_ALLOC
 	ZFS_AC_KERNEL_CONFIG_TRIM_UNUSED_KSYMS
-	ZFS_AC_KERNEL_CONFIG_ZLIB_INFLATE
 	ZFS_AC_KERNEL_CONFIG_ZLIB_DEFLATE
+	ZFS_AC_KERNEL_CONFIG_ZLIB_INFLATE
+])
+
+dnl #
+dnl # Check CONFIG_BLOCK
+dnl #
+dnl # Verify the kernel has CONFIG_BLOCK support enabled.
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_SRC_CONFIG_BLOCK], [
+	ZFS_LINUX_TEST_SRC([config_block], [
+		#if !defined(CONFIG_BLOCK)
+		#error CONFIG_BLOCK not defined
+		#endif
+	],[])
+])
+
+AC_DEFUN([ZFS_AC_KERNEL_CONFIG_BLOCK], [
+	AC_MSG_CHECKING([whether CONFIG_BLOCK is defined])
+	ZFS_LINUX_TEST_RESULT([config_block], [
+		AC_MSG_RESULT([yes])
+	],[
+		AC_MSG_RESULT([no])
+		AC_MSG_ERROR([
+	*** This kernel does not include the required block device support.
+	*** Rebuild the kernel with CONFIG_BLOCK=y set.])
+	])
 ])
 
 dnl #
@@ -72,6 +101,61 @@ AC_DEFUN([ZFS_AC_KERNEL_CONFIG_DEBUG_LOCK_ALLOC], [
 	])
 ])
 
+dnl #
+dnl # Check CONFIG_MODULES
+dnl #
+dnl # Verify the kernel has CONFIG_MODULES support enabled.
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_SRC_CONFIG_MODULES], [
+	ZFS_LINUX_TEST_SRC([config_modules], [
+		#if !defined(CONFIG_MODULES)
+		#error CONFIG_MODULES not defined
+		#endif
+	],[])
+])
+
+AC_DEFUN([ZFS_AC_KERNEL_CONFIG_MODULES], [
+	AC_MSG_CHECKING([whether CONFIG_MODULES is defined])
+	AS_IF([test "x$enable_linux_builtin" != xyes], [
+		ZFS_LINUX_TEST_RESULT([config_modules], [
+			AC_MSG_RESULT([yes])
+		],[
+			AC_MSG_RESULT([no])
+			AC_MSG_ERROR([
+		*** This kernel does not include the required loadable module
+		*** support!
+		***
+		*** To build OpenZFS as a loadable Linux kernel module
+		*** enable loadable module support by setting
+		*** `CONFIG_MODULES=y` in the kernel configuration and run
+		*** `make modules_prepare` in the Linux source tree.
+		***
+		*** If you don't intend to enable loadable kernel module
+		*** support, please compile OpenZFS as a Linux kernel built-in.
+		***
+		*** Prepare the Linux source tree by running `make prepare`,
+		*** use the OpenZFS `--enable-linux-builtin` configure option,
+		*** copy the OpenZFS sources into the Linux source tree using
+		*** `./copy-builtin <linux source directory>`,
+		*** set `CONFIG_ZFS=y` in the kernel configuration and compile
+		*** kernel as usual.
+			])
+		])
+	], [
+		ZFS_LINUX_TRY_COMPILE([], [], [
+			AC_MSG_RESULT([not needed])
+		],[
+			AC_MSG_RESULT([error])
+			AC_MSG_ERROR([
+		*** This kernel is unable to compile object files.
+		***
+		*** Please make sure you prepared the Linux source tree
+		*** by running `make prepare` there.
+			])
+		])
+	])
+])
+
 dnl #
 dnl # Check CONFIG_TRIM_UNUSED_KSYMS
 dnl #

config/kernel-copy-from-user-inatomic.m4

@@ -0,0 +1,29 @@
+dnl #
+dnl # On certain architectures `__copy_from_user_inatomic`
+dnl # is a GPL exported variable and cannot be used by OpenZFS.
+dnl #
+
+dnl #
+dnl # Checking if `__copy_from_user_inatomic` is available.
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_SRC___COPY_FROM_USER_INATOMIC], [
+	ZFS_LINUX_TEST_SRC([__copy_from_user_inatomic], [
+		#include <linux/uaccess.h>
+	], [
+		int result __attribute__ ((unused)) = __copy_from_user_inatomic(NULL, NULL, 0);
+	], [], [ZFS_META_LICENSE])
+])
+
+AC_DEFUN([ZFS_AC_KERNEL___COPY_FROM_USER_INATOMIC], [
+	AC_MSG_CHECKING([whether __copy_from_user_inatomic is available])
+	ZFS_LINUX_TEST_RESULT([__copy_from_user_inatomic_license], [
+		AC_MSG_RESULT(yes)
+	], [
+		AC_MSG_RESULT(no)
+		AC_MSG_ERROR([
+	*** The `__copy_from_user_inatomic()` Linux kernel function is
+	*** incompatible with the CDDL license and will prevent the module
+	*** linking stage from succeeding.  OpenZFS cannot be compiled.
+		])
+	])
+])

config/kernel-fpu.m4

@@ -2,6 +2,9 @@ dnl #
 dnl # Handle differences in kernel FPU code.
 dnl #
 dnl # Kernel
+dnl # 5.19:	The asm/fpu/internal.h header was removed, it has been
+dnl #		effectively empty since the 5.16 kernel.
+dnl #
 dnl # 5.16:	XCR code put into asm/fpu/xcr.h
 dnl #		HAVE_KERNEL_FPU_XCR_HEADER
 dnl #
@@ -33,21 +36,31 @@ AC_DEFUN([ZFS_AC_KERNEL_FPU_HEADER], [
 	],[
 		AC_DEFINE(HAVE_KERNEL_FPU_API_HEADER, 1,
 		    [kernel has asm/fpu/api.h])
-		AC_MSG_RESULT(asm/fpu/api.h)
-		AC_MSG_CHECKING([whether fpu/xcr header is available])
+		fpu_headers="asm/fpu/api.h"
+
 		ZFS_LINUX_TRY_COMPILE([
 			#include <linux/module.h>
 			#include <asm/fpu/xcr.h>
 		],[
 		],[
 			AC_DEFINE(HAVE_KERNEL_FPU_XCR_HEADER, 1,
-				[kernel has asm/fpu/xcr.h])
-			AC_MSG_RESULT(asm/fpu/xcr.h)
-		],[
-			AC_MSG_RESULT(no asm/fpu/xcr.h)
+			    [kernel has asm/fpu/xcr.h])
+			fpu_headers="$fpu_headers asm/fpu/xcr.h"
 		])
+
+		ZFS_LINUX_TRY_COMPILE([
+			#include <linux/module.h>
+			#include <asm/fpu/internal.h>
+		],[
+		],[
+			AC_DEFINE(HAVE_KERNEL_FPU_INTERNAL_HEADER, 1,
+			    [kernel has asm/fpu/internal.h])
+			fpu_headers="$fpu_headers asm/fpu/internal.h"
+		])
+
+		AC_MSG_RESULT([$fpu_headers])
 	],[
-		AC_MSG_RESULT(i387.h & xcr.h)
+		AC_MSG_RESULT([i387.h & xcr.h])
 	])
 ])
 
@@ -93,7 +106,9 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_FPU], [
 		#include <linux/types.h>
 		#ifdef HAVE_KERNEL_FPU_API_HEADER
 		#include <asm/fpu/api.h>
+		#ifdef HAVE_KERNEL_FPU_INTERNAL_HEADER
 		#include <asm/fpu/internal.h>
+		#endif
 		#else
 		#include <asm/i387.h>
 		#include <asm/xcr.h>
@@ -130,7 +145,9 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_FPU], [
 		#include <linux/types.h>
 		#ifdef HAVE_KERNEL_FPU_API_HEADER
 		#include <asm/fpu/api.h>
+		#ifdef HAVE_KERNEL_FPU_INTERNAL_HEADER
 		#include <asm/fpu/internal.h>
+		#endif
 		#else
 		#include <asm/i387.h>
 		#include <asm/xcr.h>

config/kernel-generic_io_acct.m4

@@ -2,6 +2,19 @@ dnl #
 dnl # Check for generic io accounting interface.
 dnl #
 AC_DEFUN([ZFS_AC_KERNEL_SRC_GENERIC_IO_ACCT], [
+	ZFS_LINUX_TEST_SRC([bdev_io_acct], [
+		#include <linux/blkdev.h>
+	], [
+		struct block_device *bdev = NULL;
+		struct bio *bio = NULL;
+		unsigned long passed_time = 0;
+		unsigned long start_time;
+
+		start_time = bdev_start_io_acct(bdev, bio_sectors(bio),
+		    bio_op(bio), passed_time);
+		bdev_end_io_acct(bdev, bio_op(bio), start_time);
+	])
+
 	ZFS_LINUX_TEST_SRC([disk_io_acct], [
 		#include <linux/blkdev.h>
 	], [
@@ -50,61 +63,75 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_GENERIC_IO_ACCT], [
 
 AC_DEFUN([ZFS_AC_KERNEL_GENERIC_IO_ACCT], [
 	dnl #
-	dnl # 5.12 API,
+	dnl # 5.19 API,
 	dnl #
-	dnl # bio_start_io_acct() and bio_end_io_acct() became GPL-exported
-	dnl # so use disk_start_io_acct() and disk_end_io_acct() instead
+	dnl # disk_start_io_acct() and disk_end_io_acct() have been replaced by
+	dnl # bdev_start_io_acct() and bdev_end_io_acct().
 	dnl #
-	AC_MSG_CHECKING([whether generic disk_*_io_acct() are available])
-	ZFS_LINUX_TEST_RESULT([disk_io_acct], [
+	AC_MSG_CHECKING([whether generic bdev_*_io_acct() are available])
+	ZFS_LINUX_TEST_RESULT([bdev_io_acct], [
 		AC_MSG_RESULT(yes)
-		AC_DEFINE(HAVE_DISK_IO_ACCT, 1, [disk_*_io_acct() available])
+		AC_DEFINE(HAVE_BDEV_IO_ACCT, 1, [bdev_*_io_acct() available])
 	], [
 		AC_MSG_RESULT(no)
 
 		dnl #
-		dnl # 5.7 API,
+		dnl # 5.12 API,
 		dnl #
-		dnl # Added bio_start_io_acct() and bio_end_io_acct() helpers.
+		dnl # bio_start_io_acct() and bio_end_io_acct() became GPL-exported
+		dnl # so use disk_start_io_acct() and disk_end_io_acct() instead
 		dnl #
-		AC_MSG_CHECKING([whether generic bio_*_io_acct() are available])
-		ZFS_LINUX_TEST_RESULT([bio_io_acct], [
+		AC_MSG_CHECKING([whether generic disk_*_io_acct() are available])
+		ZFS_LINUX_TEST_RESULT([disk_io_acct], [
 			AC_MSG_RESULT(yes)
-			AC_DEFINE(HAVE_BIO_IO_ACCT, 1, [bio_*_io_acct() available])
+			AC_DEFINE(HAVE_DISK_IO_ACCT, 1, [disk_*_io_acct() available])
 		], [
 			AC_MSG_RESULT(no)
 
 			dnl #
-			dnl # 4.14 API,
+			dnl # 5.7 API,
 			dnl #
-			dnl # generic_start_io_acct/generic_end_io_acct now require
-			dnl # request_queue to be provided. No functional changes,
-			dnl # but preparation for inflight accounting.
+			dnl # Added bio_start_io_acct() and bio_end_io_acct() helpers.
 			dnl #
-			AC_MSG_CHECKING([whether generic_*_io_acct wants 4 args])
-			ZFS_LINUX_TEST_RESULT_SYMBOL([generic_acct_4args],
-			    [generic_start_io_acct], [block/bio.c], [
+			AC_MSG_CHECKING([whether generic bio_*_io_acct() are available])
+			ZFS_LINUX_TEST_RESULT([bio_io_acct], [
 				AC_MSG_RESULT(yes)
-				AC_DEFINE(HAVE_GENERIC_IO_ACCT_4ARG, 1,
-				    [generic_*_io_acct() 4 arg available])
+				AC_DEFINE(HAVE_BIO_IO_ACCT, 1, [bio_*_io_acct() available])
 			], [
 				AC_MSG_RESULT(no)
 
 				dnl #
-				dnl # 3.19 API addition
+				dnl # 4.14 API,
 				dnl #
-				dnl # torvalds/linux@394ffa50 allows us to increment
-				dnl # iostat counters without generic_make_request().
+				dnl # generic_start_io_acct/generic_end_io_acct now require
+				dnl # request_queue to be provided. No functional changes,
+				dnl # but preparation for inflight accounting.
 				dnl #
-				AC_MSG_CHECKING(
-				    [whether generic_*_io_acct wants 3 args])
-				ZFS_LINUX_TEST_RESULT_SYMBOL([generic_acct_3args],
+				AC_MSG_CHECKING([whether generic_*_io_acct wants 4 args])
+				ZFS_LINUX_TEST_RESULT_SYMBOL([generic_acct_4args],
 				    [generic_start_io_acct], [block/bio.c], [
 					AC_MSG_RESULT(yes)
-					AC_DEFINE(HAVE_GENERIC_IO_ACCT_3ARG, 1,
-					    [generic_*_io_acct() 3 arg available])
+					AC_DEFINE(HAVE_GENERIC_IO_ACCT_4ARG, 1,
+					    [generic_*_io_acct() 4 arg available])
 				], [
 					AC_MSG_RESULT(no)
+
+					dnl #
+					dnl # 3.19 API addition
+					dnl #
+					dnl # torvalds/linux@394ffa50 allows us to increment
+					dnl # iostat counters without generic_make_request().
+					dnl #
+					AC_MSG_CHECKING(
+					    [whether generic_*_io_acct wants 3 args])
+					ZFS_LINUX_TEST_RESULT_SYMBOL([generic_acct_3args],
+					    [generic_start_io_acct], [block/bio.c], [
+						AC_MSG_RESULT(yes)
+						AC_DEFINE(HAVE_GENERIC_IO_ACCT_3ARG, 1,
+						    [generic_*_io_acct() 3 arg available])
+					], [
+						AC_MSG_RESULT(no)
+					])
 				])
 			])
 		])

config/kernel-genhd-flags.m4

@@ -0,0 +1,58 @@
+dnl #
+dnl # 5.17 API change,
+dnl #
+dnl # GENHD_FL_EXT_DEVT flag removed
+dnl # GENHD_FL_NO_PART_SCAN renamed GENHD_FL_NO_PART
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_SRC_GENHD_FLAGS], [
+
+	ZFS_LINUX_TEST_SRC([genhd_fl_ext_devt], [
+		#include <linux/blkdev.h>
+	], [
+		int flags __attribute__ ((unused)) = GENHD_FL_EXT_DEVT;
+	])
+
+	ZFS_LINUX_TEST_SRC([genhd_fl_no_part], [
+		#include <linux/blkdev.h>
+	], [
+		int flags __attribute__ ((unused)) = GENHD_FL_NO_PART;
+	])
+
+	ZFS_LINUX_TEST_SRC([genhd_fl_no_part_scan], [
+		#include <linux/blkdev.h>
+	], [
+		int flags __attribute__ ((unused)) = GENHD_FL_NO_PART_SCAN;
+	])
+])
+
+AC_DEFUN([ZFS_AC_KERNEL_GENHD_FLAGS], [
+
+	AC_MSG_CHECKING([whether GENHD_FL_EXT_DEVT flag is available])
+	ZFS_LINUX_TEST_RESULT([genhd_fl_ext_devt], [
+		AC_MSG_RESULT(yes)
+		AC_DEFINE(ZFS_GENHD_FL_EXT_DEVT, GENHD_FL_EXT_DEVT,
+		    [GENHD_FL_EXT_DEVT flag is available])
+	], [
+		AC_MSG_RESULT(no)
+		AC_DEFINE(ZFS_GENHD_FL_EXT_DEVT, 0,
+		    [GENHD_FL_EXT_DEVT flag is not available])
+	])
+
+	AC_MSG_CHECKING([whether GENHD_FL_NO_PART flag is available])
+	ZFS_LINUX_TEST_RESULT([genhd_fl_no_part], [
+		AC_MSG_RESULT(yes)
+		AC_DEFINE(ZFS_GENHD_FL_NO_PART, GENHD_FL_NO_PART,
+		    [GENHD_FL_NO_PART flag is available])
+	], [
+		AC_MSG_RESULT(no)
+
+		AC_MSG_CHECKING([whether GENHD_FL_NO_PART_SCAN flag is available])
+		ZFS_LINUX_TEST_RESULT([genhd_fl_no_part_scan], [
+			AC_MSG_RESULT(yes)
+			AC_DEFINE(ZFS_GENHD_FL_NO_PART, GENHD_FL_NO_PART_SCAN,
+			    [GENHD_FL_NO_PART_SCAN flag is available])
+		], [
+			ZFS_LINUX_TEST_ERROR([GENHD_FL_NO_PART|GENHD_FL_NO_PART_SCAN])
+		])
+	])
+])

config/kernel-group-info.m4

@@ -6,8 +6,8 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_GROUP_INFO_GID], [
 	ZFS_LINUX_TEST_SRC([group_info_gid], [
 		#include <linux/cred.h>
 	],[
-		struct group_info *gi = groups_alloc(1);
-		gi->gid[0] = KGIDT_INIT(0);
+		struct group_info gi __attribute__ ((unused)) = {};
+		gi.gid[0] = KGIDT_INIT(0);
 	])
 ])
 

config/kernel-mkdir.m4

@@ -53,6 +53,8 @@ AC_DEFUN([ZFS_AC_KERNEL_MKDIR], [
 		AC_DEFINE(HAVE_IOPS_MKDIR_USERNS, 1,
 		    [iops->mkdir() takes struct user_namespace*])
 	],[
+		AC_MSG_RESULT(no)
+
 		AC_MSG_CHECKING([whether iops->mkdir() takes umode_t])
 		ZFS_LINUX_TEST_RESULT([inode_operations_mkdir], [
 			AC_MSG_RESULT(yes)

config/kernel-pagemap-folio_wait_bit.m4

@@ -15,7 +15,7 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_PAGEMAP_FOLIO_WAIT_BIT], [
 ])
 
 AC_DEFUN([ZFS_AC_KERNEL_PAGEMAP_FOLIO_WAIT_BIT], [
-	AC_MSG_CHECKING([folio_wait_bit() exists])
+	AC_MSG_CHECKING([whether folio_wait_bit() exists])
 	ZFS_LINUX_TEST_RESULT([pagemap_has_folio_wait_bit], [
 		AC_MSG_RESULT([yes])
 		AC_DEFINE(HAVE_PAGEMAP_FOLIO_WAIT_BIT, 1,

config/kernel-readpages.m4

@@ -0,0 +1,25 @@
+dnl #
+dnl # Linux 5.18 removes address_space_operations ->readpages in favour of
+dnl # ->readahead
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_SRC_VFS_READPAGES], [
+	ZFS_LINUX_TEST_SRC([vfs_has_readpages], [
+		#include <linux/fs.h>
+
+		static const struct address_space_operations
+		    aops __attribute__ ((unused)) = {
+			.readpages = NULL,
+		};
+	],[])
+])
+
+AC_DEFUN([ZFS_AC_KERNEL_VFS_READPAGES], [
+	AC_MSG_CHECKING([whether aops->readpages exists])
+	ZFS_LINUX_TEST_RESULT([vfs_has_readpages], [
+		AC_MSG_RESULT([yes])
+		AC_DEFINE(HAVE_VFS_READPAGES, 1,
+			[address_space_operations->readpages exists])
+	],[
+		AC_MSG_RESULT([no])
+	])
+])

config/kernel-revalidate-disk-size.m4

@@ -8,14 +8,14 @@ dnl #
 AC_DEFUN([ZFS_AC_KERNEL_SRC_REVALIDATE_DISK], [
 
 	ZFS_LINUX_TEST_SRC([revalidate_disk_size], [
-		#include <linux/genhd.h>
+		#include <linux/blkdev.h>
 	], [
 		struct gendisk *disk = NULL;
 		(void) revalidate_disk_size(disk, false);
 	])
 
 	ZFS_LINUX_TEST_SRC([revalidate_disk], [
-		#include <linux/genhd.h>
+		#include <linux/blkdev.h>
 	], [
 		struct gendisk *disk = NULL;
 		(void) revalidate_disk(disk);

config/kernel-shrink.m4

@@ -84,7 +84,7 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_SHRINKER_CALLBACK], [
 AC_DEFUN([ZFS_AC_KERNEL_SHRINKER_CALLBACK],[
 	dnl #
 	dnl # 3.0 - 3.11 API change
-	dnl # ->shrink(struct shrinker *, struct shrink_control *sc)
+	dnl # cs->shrink(struct shrinker *, struct shrink_control *sc)
 	dnl #
 	AC_MSG_CHECKING([whether new 2-argument shrinker exists])
 	ZFS_LINUX_TEST_RESULT([shrinker_cb_shrink_control], [
@@ -96,14 +96,14 @@ AC_DEFUN([ZFS_AC_KERNEL_SHRINKER_CALLBACK],[
 
 		dnl #
 		dnl # 3.12 API change,
-		dnl # ->shrink() is logically split in to
-		dnl # ->count_objects() and ->scan_objects()
+		dnl # cs->shrink() is logically split in to
+		dnl # cs->count_objects() and cs->scan_objects()
 		dnl #
-		AC_MSG_CHECKING([whether ->count_objects callback exists])
+		AC_MSG_CHECKING([whether cs->count_objects callback exists])
 		ZFS_LINUX_TEST_RESULT([shrinker_cb_shrink_control_split], [
 			AC_MSG_RESULT(yes)
 			AC_DEFINE(HAVE_SPLIT_SHRINKER_CALLBACK, 1,
-			    [->count_objects exists])
+			    [cs->count_objects exists])
 		],[
 			ZFS_LINUX_TEST_ERROR([shrinker])
 		])

config/kernel-sysfs.m4

@@ -0,0 +1,37 @@
+dnl #
+dnl # Linux 5.2/5.18 API
+dnl #
+dnl # In cdb4f26a63c391317e335e6e683a614358e70aeb ("kobject: kobj_type: remove default_attrs")
+dnl # 	struct kobj_type.default_attrs
+dnl # was finally removed in favour of
+dnl # 	struct kobj_type.default_groups
+dnl #
+dnl # This was added in aa30f47cf666111f6bbfd15f290a27e8a7b9d854 ("kobject: Add support for default attribute groups to kobj_type"),
+dnl # if both are present (5.2-5.17), we prefer default_groups; they're otherwise equivalent
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_SRC_SYSFS_DEFAULT_GROUPS], [
+	ZFS_LINUX_TEST_SRC([sysfs_default_groups], [
+		#include <linux/kobject.h>
+	],[
+		struct kobj_type __attribute__ ((unused)) kt = {
+			.default_groups = (const struct attribute_group **)NULL };
+	])
+])
+
+AC_DEFUN([ZFS_AC_KERNEL_SYSFS_DEFAULT_GROUPS], [
+	AC_MSG_CHECKING([whether struct kobj_type.default_groups exists])
+	ZFS_LINUX_TEST_RESULT([sysfs_default_groups],[
+		AC_MSG_RESULT(yes)
+		AC_DEFINE([HAVE_SYSFS_DEFAULT_GROUPS], 1, [struct kobj_type has default_groups])
+	],[
+		AC_MSG_RESULT(no)
+	])
+])
+
+AC_DEFUN([ZFS_AC_KERNEL_SRC_SYSFS], [
+	ZFS_AC_KERNEL_SRC_SYSFS_DEFAULT_GROUPS
+])
+
+AC_DEFUN([ZFS_AC_KERNEL_SYSFS], [
+	ZFS_AC_KERNEL_SYSFS_DEFAULT_GROUPS
+])

config/kernel-vfs-filemap_dirty_folio.m4

@@ -0,0 +1,30 @@
+dnl #
+dnl # Linux 5.18 uses filemap_dirty_folio in lieu of
+dnl # ___set_page_dirty_nobuffers
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_SRC_VFS_FILEMAP_DIRTY_FOLIO], [
+	ZFS_LINUX_TEST_SRC([vfs_has_filemap_dirty_folio], [
+		#include <linux/pagemap.h>
+		#include <linux/writeback.h>
+
+		static const struct address_space_operations
+		    aops __attribute__ ((unused)) = {
+			.dirty_folio	= filemap_dirty_folio,
+		};
+	],[])
+])
+
+AC_DEFUN([ZFS_AC_KERNEL_VFS_FILEMAP_DIRTY_FOLIO], [
+	dnl #
+	dnl # Linux 5.18 uses filemap_dirty_folio in lieu of
+	dnl # ___set_page_dirty_nobuffers
+	dnl #
+	AC_MSG_CHECKING([whether filemap_dirty_folio exists])
+	ZFS_LINUX_TEST_RESULT([vfs_has_filemap_dirty_folio], [
+		AC_MSG_RESULT([yes])
+		AC_DEFINE(HAVE_VFS_FILEMAP_DIRTY_FOLIO, 1,
+			[filemap_dirty_folio exists])
+	],[
+		AC_MSG_RESULT([no])
+	])
+])

config/kernel-vfs-iov_iter.m4

@@ -134,6 +134,8 @@ AC_DEFUN([ZFS_AC_KERNEL_VFS_IOV_ITER], [
 		AC_DEFINE(HAVE_IOV_ITER_FAULT_IN_READABLE, 1,
 		    [iov_iter_fault_in_readable() is available])
 	],[
+		AC_MSG_RESULT(no)
+
 		AC_MSG_CHECKING([whether fault_in_iov_iter_readable() is available])
 		ZFS_LINUX_TEST_RESULT([fault_in_iov_iter_readable], [
 			AC_MSG_RESULT(yes)

config/kernel-vfs-read_folio.m4

@@ -0,0 +1,32 @@
+dnl #
+dnl # Linux 5.19 uses read_folio in lieu of readpage
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_SRC_VFS_READ_FOLIO], [
+	ZFS_LINUX_TEST_SRC([vfs_has_read_folio], [
+		#include <linux/fs.h>
+
+		static int
+		test_read_folio(struct file *file, struct folio *folio) {
+			(void) file; (void) folio;
+			return (0);
+		}
+
+		static const struct address_space_operations
+		    aops __attribute__ ((unused)) = {
+			.read_folio	= test_read_folio,
+		};
+	],[])
+])
+
+AC_DEFUN([ZFS_AC_KERNEL_VFS_READ_FOLIO], [
+	dnl #
+	dnl # Linux 5.19 uses read_folio in lieu of readpage
+	dnl #
+	AC_MSG_CHECKING([whether read_folio exists])
+	ZFS_LINUX_TEST_RESULT([vfs_has_read_folio], [
+		AC_MSG_RESULT([yes])
+		AC_DEFINE(HAVE_VFS_READ_FOLIO, 1, [read_folio exists])
+	],[
+		AC_MSG_RESULT([no])
+	])
+])

config/kernel-vfs-set_page_dirty.m4

@@ -23,7 +23,7 @@ AC_DEFUN([ZFS_AC_KERNEL_VFS_SET_PAGE_DIRTY_NOBUFFERS], [
 	dnl # Linux 5.14 change requires set_page_dirty() to be assigned
 	dnl # in address_space_operations()
 	dnl #
-	AC_MSG_CHECKING([__set_page_dirty_nobuffers exists])
+	AC_MSG_CHECKING([whether __set_page_dirty_nobuffers exists])
 	ZFS_LINUX_TEST_RESULT([vfs_has_set_page_dirty_nobuffers], [
 		AC_MSG_RESULT([yes])
 		AC_DEFINE(HAVE_VFS_SET_PAGE_DIRTY_NOBUFFERS, 1,

config/kernel-zero_page.m4

@@ -0,0 +1,27 @@
+dnl #
+dnl # ZERO_PAGE() is an alias for emtpy_zero_page. On certain architectures
+dnl # this is a GPL exported variable.
+dnl #
+
+dnl #
+dnl # Checking if ZERO_PAGE is exported GPL-only
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_SRC_ZERO_PAGE], [
+	ZFS_LINUX_TEST_SRC([zero_page], [
+		#include <asm/pgtable.h>
+	], [
+		struct page *p __attribute__ ((unused));
+		p = ZERO_PAGE(0);
+	], [], [ZFS_META_LICENSE])
+])
+
+AC_DEFUN([ZFS_AC_KERNEL_ZERO_PAGE], [
+	AC_MSG_CHECKING([whether ZERO_PAGE() is GPL-only])
+	ZFS_LINUX_TEST_RESULT([zero_page_license], [
+		AC_MSG_RESULT(no)
+	], [
+		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_ZERO_PAGE_GPL_ONLY, 1,
+		    [ZERO_PAGE() is GPL-only])
+	])
+])

config/kernel.m4

@@ -8,8 +8,8 @@ AC_DEFUN([ZFS_AC_CONFIG_KERNEL], [
 		ZFS_AC_QAT
 
 		dnl # Sanity checks for module building and CONFIG_* defines
-		ZFS_AC_KERNEL_TEST_MODULE
 		ZFS_AC_KERNEL_CONFIG_DEFINED
+		ZFS_AC_MODULE_SYMVERS
 
 		dnl # Sequential ZFS_LINUX_TRY_COMPILE tests
 		ZFS_AC_KERNEL_FPU_HEADER
@@ -61,6 +61,7 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_SRC], [
 	ZFS_AC_KERNEL_SRC_BIO
 	ZFS_AC_KERNEL_SRC_BLKDEV
 	ZFS_AC_KERNEL_SRC_BLK_QUEUE
+	ZFS_AC_KERNEL_SRC_GENHD_FLAGS
 	ZFS_AC_KERNEL_SRC_REVALIDATE_DISK
 	ZFS_AC_KERNEL_SRC_GET_DISK_RO
 	ZFS_AC_KERNEL_SRC_GENERIC_READLINK_GLOBAL
@@ -99,10 +100,14 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_SRC], [
 	ZFS_AC_KERNEL_SRC_SET_NLINK
 	ZFS_AC_KERNEL_SRC_SGET
 	ZFS_AC_KERNEL_SRC_LSEEK_EXECUTE
+	ZFS_AC_KERNEL_SRC_VFS_FILEMAP_DIRTY_FOLIO
+	ZFS_AC_KERNEL_SRC_VFS_READ_FOLIO
 	ZFS_AC_KERNEL_SRC_VFS_GETATTR
 	ZFS_AC_KERNEL_SRC_VFS_FSYNC_2ARGS
 	ZFS_AC_KERNEL_SRC_VFS_ITERATE
 	ZFS_AC_KERNEL_SRC_VFS_DIRECT_IO
+	ZFS_AC_KERNEL_SRC_VFS_READPAGES
+	ZFS_AC_KERNEL_SRC_VFS_SET_PAGE_DIRTY_NOBUFFERS
 	ZFS_AC_KERNEL_SRC_VFS_RW_ITERATE
 	ZFS_AC_KERNEL_SRC_VFS_GENERIC_WRITE_CHECKS
 	ZFS_AC_KERNEL_SRC_VFS_IOV_ITER
@@ -131,12 +136,14 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_SRC], [
 	ZFS_AC_KERNEL_SRC_BIO_MAX_SEGS
 	ZFS_AC_KERNEL_SRC_SIGNAL_STOP
 	ZFS_AC_KERNEL_SRC_SIGINFO
+	ZFS_AC_KERNEL_SRC_SYSFS
 	ZFS_AC_KERNEL_SRC_SET_SPECIAL_STATE
-	ZFS_AC_KERNEL_SRC_VFS_SET_PAGE_DIRTY_NOBUFFERS
 	ZFS_AC_KERNEL_SRC_STANDALONE_LINUX_STDARG
 	ZFS_AC_KERNEL_SRC_PAGEMAP_FOLIO_WAIT_BIT
 	ZFS_AC_KERNEL_SRC_ADD_DISK
 	ZFS_AC_KERNEL_SRC_KTHREAD
+	ZFS_AC_KERNEL_SRC_ZERO_PAGE
+	ZFS_AC_KERNEL_SRC___COPY_FROM_USER_INATOMIC
 
 	AC_MSG_CHECKING([for available kernel interfaces])
 	ZFS_LINUX_TEST_COMPILE_ALL([kabi])
@@ -171,6 +178,7 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_RESULT], [
 	ZFS_AC_KERNEL_BIO
 	ZFS_AC_KERNEL_BLKDEV
 	ZFS_AC_KERNEL_BLK_QUEUE
+	ZFS_AC_KERNEL_GENHD_FLAGS
 	ZFS_AC_KERNEL_REVALIDATE_DISK
 	ZFS_AC_KERNEL_GET_DISK_RO
 	ZFS_AC_KERNEL_GENERIC_READLINK_GLOBAL
@@ -209,10 +217,14 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_RESULT], [
 	ZFS_AC_KERNEL_SET_NLINK
 	ZFS_AC_KERNEL_SGET
 	ZFS_AC_KERNEL_LSEEK_EXECUTE
+	ZFS_AC_KERNEL_VFS_FILEMAP_DIRTY_FOLIO
+	ZFS_AC_KERNEL_VFS_READ_FOLIO
 	ZFS_AC_KERNEL_VFS_GETATTR
 	ZFS_AC_KERNEL_VFS_FSYNC_2ARGS
 	ZFS_AC_KERNEL_VFS_ITERATE
 	ZFS_AC_KERNEL_VFS_DIRECT_IO
+	ZFS_AC_KERNEL_VFS_READPAGES
+	ZFS_AC_KERNEL_VFS_SET_PAGE_DIRTY_NOBUFFERS
 	ZFS_AC_KERNEL_VFS_RW_ITERATE
 	ZFS_AC_KERNEL_VFS_GENERIC_WRITE_CHECKS
 	ZFS_AC_KERNEL_VFS_IOV_ITER
@@ -241,12 +253,14 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_RESULT], [
 	ZFS_AC_KERNEL_BIO_MAX_SEGS
 	ZFS_AC_KERNEL_SIGNAL_STOP
 	ZFS_AC_KERNEL_SIGINFO
+	ZFS_AC_KERNEL_SYSFS
 	ZFS_AC_KERNEL_SET_SPECIAL_STATE
-	ZFS_AC_KERNEL_VFS_SET_PAGE_DIRTY_NOBUFFERS
 	ZFS_AC_KERNEL_STANDALONE_LINUX_STDARG
 	ZFS_AC_KERNEL_PAGEMAP_FOLIO_WAIT_BIT
 	ZFS_AC_KERNEL_ADD_DISK
 	ZFS_AC_KERNEL_KTHREAD
+	ZFS_AC_KERNEL_ZERO_PAGE
+	ZFS_AC_KERNEL___COPY_FROM_USER_INATOMIC
 ])
 
 dnl #
@@ -435,8 +449,6 @@ AC_DEFUN([ZFS_AC_KERNEL], [
 	AC_SUBST(LINUX)
 	AC_SUBST(LINUX_OBJ)
 	AC_SUBST(LINUX_VERSION)
-
-	ZFS_AC_MODULE_SYMVERS
 ])
 
 dnl #
@@ -531,27 +543,6 @@ AC_DEFUN([ZFS_AC_QAT], [
 	])
 ])
 
-dnl #
-dnl # Basic toolchain sanity check.
-dnl #
-AC_DEFUN([ZFS_AC_KERNEL_TEST_MODULE], [
-	AC_MSG_CHECKING([whether modules can be built])
-	ZFS_LINUX_TRY_COMPILE([], [], [
-		AC_MSG_RESULT([yes])
-	],[
-		AC_MSG_RESULT([no])
-		if test "x$enable_linux_builtin" != xyes; then
-			AC_MSG_ERROR([
-	*** Unable to build an empty module.
-			])
-		else
-			AC_MSG_ERROR([
-	*** Unable to build an empty module.
-	*** Please run 'make scripts' inside the kernel source tree.])
-		fi
-	])
-])
-
 dnl #
 dnl # ZFS_LINUX_CONFTEST_H
 dnl #
@@ -654,8 +645,10 @@ AC_DEFUN([ZFS_LINUX_COMPILE], [
 		build kernel modules with LLVM/CLANG toolchain])
 	AC_TRY_COMMAND([
 	    KBUILD_MODPOST_NOFINAL="$5" KBUILD_MODPOST_WARN="$6"
-	    make modules -k -j$TEST_JOBS ${KERNEL_CC:+CC=$KERNEL_CC} ${KERNEL_LD:+LD=$KERNEL_LD} ${KERNEL_LLVM:+LLVM=$KERNEL_LLVM} -C $LINUX_OBJ $ARCH_UM
-	    M=$PWD/$1 >$1/build.log 2>&1])
+	    make modules -k -j$TEST_JOBS ${KERNEL_CC:+CC=$KERNEL_CC}
+	    ${KERNEL_LD:+LD=$KERNEL_LD} ${KERNEL_LLVM:+LLVM=$KERNEL_LLVM}
+	    CONFIG_MODULES=y CFLAGS_MODULE=-DCONFIG_MODULES
+	    -C $LINUX_OBJ $ARCH_UM M=$PWD/$1 >$1/build.log 2>&1])
 	AS_IF([AC_TRY_COMMAND([$2])], [$3], [$4])
 ])
 

config/zfs-build.m4

@@ -323,6 +323,10 @@ AC_DEFUN([ZFS_AC_RPM], [
 	RPM_DEFINE_COMMON=${RPM_DEFINE_COMMON}' --define "$(DEBUG_KMEM_TRACKING_ZFS) 1"'
 	RPM_DEFINE_COMMON=${RPM_DEFINE_COMMON}' --define "$(ASAN_ZFS) 1"'
 
+	AS_IF([test "x$enable_debuginfo" = xyes], [
+		RPM_DEFINE_COMMON=${RPM_DEFINE_COMMON}' --define "__strip /bin/true"'
+	])
+
 	RPM_DEFINE_UTIL=' --define "_initconfdir $(initconfdir)"'
 
 	dnl # Make the next three RPM_DEFINE_UTIL additions conditional, since

contrib/dracut/90zfs/export-zfs.sh.in

@@ -1,14 +1,12 @@
 #!/bin/sh
 
-. /lib/dracut-zfs-lib.sh
-
 _do_zpool_export() {
 	ret=0
 	errs=""
 	final="${1}"
 
 	info "ZFS: Exporting ZFS storage pools..."
-	errs=$(export_all -F 2>&1)
+	errs=$(zpool export -aF 2>&1)
 	ret=$?
 	[ -z "${errs}" ] || echo "${errs}" | vwarn
 	if [ "x${ret}" != "x0" ]; then

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

@@ -6,8 +6,8 @@ check() {
 	[ "${1}" = "-d" ] && return 0
 
 	# Verify the zfs tool chain
-	for tool in "@sbindir@/zgenhostid" "@sbindir@/zpool" "@sbindir@/zfs" "@mounthelperdir@/mount.zfs" ; do
-		test -x "$tool" || return 1
+	for tool in "zgenhostid" "zpool" "zfs" "mount.zfs"; do
+		command -v "${tool}" >/dev/null || return 1
 	done
 
 	return 0
@@ -19,125 +19,85 @@ depends() {
 }
 
 installkernel() {
-	instmods zfs
-	instmods zcommon
-	instmods znvpair
-	instmods zavl
-	instmods zunicode
-	instmods zlua
-	instmods icp
-	instmods spl
-	instmods zlib_deflate
-	instmods zlib_inflate
+	instmods -c zfs
 }
 
 install() {
-	inst_rules @udevruledir@/90-zfs.rules
-	inst_rules @udevruledir@/69-vdev.rules
-	inst_rules @udevruledir@/60-zvol.rules
-	dracut_install hostid
-	dracut_install grep
-	dracut_install @sbindir@/zgenhostid
-	dracut_install @sbindir@/zfs
-	dracut_install @sbindir@/zpool
-	# Workaround for https://github.com/openzfs/zfs/issues/4749 by
-	# ensuring libgcc_s.so(.1) is included
-	if ldd @sbindir@/zpool | grep -qF 'libgcc_s.so'; then
-		# Dracut will have already tracked and included it
-		:;
-	elif command -v gcc-config >/dev/null 2>&1; then
-		# On systems with gcc-config (Gentoo, Funtoo, etc.):
-		# Use the current profile to resolve the appropriate path
-		s="$(gcc-config -c)"
-		dracut_install "/usr/lib/gcc/${s%-*}/${s##*-}/libgcc_s.so"*
-	elif [ "$(echo /usr/lib/libgcc_s.so*)" != "/usr/lib/libgcc_s.so*" ]; then
-		# Try a simple path first
-		dracut_install /usr/lib/libgcc_s.so*
-	elif [ "$(echo /lib*/libgcc_s.so*)" != "/lib*/libgcc_s.so*" ]; then
-		# SUSE
-		dracut_install /lib*/libgcc_s.so*
-	else
-		# Fallback: Guess the path and include all matches
-		dracut_install /usr/lib*/gcc/**/libgcc_s.so*
+	inst_rules 90-zfs.rules 69-vdev.rules 60-zvol.rules
+
+	inst_multiple \
+		zgenhostid \
+		zfs \
+		zpool \
+		mount.zfs \
+		hostid \
+		grep \
+		awk \
+		tr \
+		cut \
+		head ||
+		{ dfatal "Failed to install essential binaries"; exit 1; }
+
+	# Adapted from https://github.com/zbm-dev/zfsbootmenu
+	if ! ldd "$(command -v zpool)" | grep -qF 'libgcc_s.so'; then
+		# On systems with gcc-config (Gentoo, Funtoo, etc.), use it to find libgcc_s
+		if command -v gcc-config >/dev/null; then
+			inst_simple "/usr/lib/gcc/$(s=$(gcc-config -c); echo "${s%-*}/${s##*-}")/libgcc_s.so.1" ||
+				{ dfatal "Unable to install libgcc_s.so"; exit 1; }
+			# Otherwise, use dracut's library installation function to find the right one
+		elif ! inst_libdir_file "libgcc_s.so*"; then
+			# If all else fails, just try looking for some gcc arch directory
+			inst_simple /usr/lib/gcc/*/*/libgcc_s.so* ||
+				{ dfatal "Unable to install libgcc_s.so"; exit 1; }
+		fi
 	fi
-	# shellcheck disable=SC2050
-	if [ @LIBFETCH_DYNAMIC@ -gt 0 ]; then
-		for d in $libdirs; do
-			[ -e "$d/@LIBFETCH_SONAME@" ] && dracut_install "$d/@LIBFETCH_SONAME@"
-		done
-	fi
-	dracut_install @mounthelperdir@/mount.zfs
-	dracut_install @udevdir@/vdev_id
-	dracut_install awk
-	dracut_install cut
-	dracut_install tr
-	dracut_install head
-	dracut_install @udevdir@/zvol_id
+
 	inst_hook cmdline 95 "${moddir}/parse-zfs.sh"
-	if [ -n "$systemdutildir" ] ; then
-		inst_script "${moddir}/zfs-generator.sh" "$systemdutildir"/system-generators/dracut-zfs-generator
+	if [ -n "${systemdutildir}" ]; then
+		inst_script "${moddir}/zfs-generator.sh" "${systemdutildir}/system-generators/dracut-zfs-generator"
 	fi
 	inst_hook pre-mount 90 "${moddir}/zfs-load-key.sh"
 	inst_hook mount 98 "${moddir}/mount-zfs.sh"
 	inst_hook cleanup 99 "${moddir}/zfs-needshutdown.sh"
 	inst_hook shutdown 20 "${moddir}/export-zfs.sh"
 
-	inst_simple "${moddir}/zfs-lib.sh" "/lib/dracut-zfs-lib.sh"
-	if [ -e @sysconfdir@/zfs/zpool.cache ]; then
-		inst @sysconfdir@/zfs/zpool.cache
-		type mark_hostonly >/dev/null 2>&1 && mark_hostonly @sysconfdir@/zfs/zpool.cache
-	fi
+	inst_script "${moddir}/zfs-lib.sh" "/lib/dracut-zfs-lib.sh"
 
-	if [ -e @sysconfdir@/zfs/vdev_id.conf ]; then
-		inst @sysconfdir@/zfs/vdev_id.conf
-		type mark_hostonly >/dev/null 2>&1 && mark_hostonly @sysconfdir@/zfs/vdev_id.conf
-	fi
+	# -H ensures they are marked host-only
+	# -o ensures there is no error upon absence of these files
+	inst_multiple -o -H \
+		"@sysconfdir@/zfs/zpool.cache" \
+		"@sysconfdir@/zfs/vdev_id.conf"
 
 	# Synchronize initramfs and system hostid
-	if [ -f @sysconfdir@/hostid ]; then
-		inst @sysconfdir@/hostid
-		type mark_hostonly >/dev/null 2>&1 && mark_hostonly @sysconfdir@/hostid
-	elif HOSTID="$(hostid 2>/dev/null)" && [ "${HOSTID}" != "00000000" ]; then
-		zgenhostid -o "${initdir}@sysconfdir@/hostid" "${HOSTID}"
-		type mark_hostonly >/dev/null 2>&1 && mark_hostonly @sysconfdir@/hostid
+	if ! inst_simple -H @sysconfdir@/hostid; then
+		if HOSTID="$(hostid 2>/dev/null)" && [ "${HOSTID}" != "00000000" ]; then
+			zgenhostid -o "${initdir}@sysconfdir@/hostid" "${HOSTID}"
+			mark_hostonly @sysconfdir@/hostid
+		fi
 	fi
 
 	if dracut_module_included "systemd"; then
-		mkdir -p "${initdir}/$systemdsystemunitdir/zfs-import.target.wants"
-		for _service in "zfs-import-scan.service" "zfs-import-cache.service" ; do
-			dracut_install "@systemdunitdir@/$_service"
-			if ! [ -L "${initdir}/$systemdsystemunitdir/zfs-import.target.wants/$_service" ]; then
-				ln -sf ../$_service "${initdir}/$systemdsystemunitdir/zfs-import.target.wants/$_service"
-				type mark_hostonly >/dev/null 2>&1 && mark_hostonly "@systemdunitdir@/$_service"
-			fi
+		inst_simple "${systemdsystemunitdir}/zfs-import.target"
+		systemctl -q --root "${initdir}" add-wants initrd.target zfs-import.target
+
+		inst_simple "${moddir}/zfs-env-bootfs.service" "${systemdsystemunitdir}/zfs-env-bootfs.service"
+		systemctl -q --root "${initdir}" add-wants zfs-import.target zfs-env-bootfs.service
+
+		for _service in \
+			"zfs-import-scan.service" \
+			"zfs-import-cache.service"; do
+			inst_simple "${systemdsystemunitdir}/${_service}"
+			systemctl -q --root "${initdir}" add-wants zfs-import.target "${_service}"
 		done
 
-		inst "${moddir}"/zfs-env-bootfs.service "${systemdsystemunitdir}"/zfs-env-bootfs.service
-		ln -s ../zfs-env-bootfs.service "${initdir}/${systemdsystemunitdir}/zfs-import.target.wants"/zfs-env-bootfs.service
-		type mark_hostonly >/dev/null 2>&1 && mark_hostonly @systemdunitdir@/zfs-env-bootfs.service
-
-		dracut_install systemd-ask-password
-		dracut_install systemd-tty-ask-password-agent
-
-		mkdir -p "${initdir}/$systemdsystemunitdir/initrd.target.wants"
-		dracut_install @systemdunitdir@/zfs-import.target
-		if ! [ -L "${initdir}/$systemdsystemunitdir/initrd.target.wants"/zfs-import.target ]; then
-			ln -s ../zfs-import.target "${initdir}/$systemdsystemunitdir/initrd.target.wants"/zfs-import.target
-			type mark_hostonly >/dev/null 2>&1 && mark_hostonly @systemdunitdir@/zfs-import.target
-		fi
-
-		for _service in zfs-snapshot-bootfs.service zfs-rollback-bootfs.service ; do
-			inst "${moddir}/$_service" "${systemdsystemunitdir}/$_service"
-			if ! [ -L "${initdir}/$systemdsystemunitdir/initrd.target.wants/$_service" ]; then
-				ln -s "../$_service" "${initdir}/$systemdsystemunitdir/initrd.target.wants/$_service"
-			fi
+		for _service in \
+			"zfs-snapshot-bootfs.service" \
+			"zfs-rollback-bootfs.service"; do
+			inst_simple "${moddir}/${_service}" "${systemdsystemunitdir}/${_service}"
+			systemctl -q --root "${initdir}" add-wants initrd.target "${_service}"
 		done
 
-		# There isn't a pkg-config variable for this,
-		# and dracut doesn't automatically resolve anything this'd be next to
-		local systemdsystemenvironmentgeneratordir
-		systemdsystemenvironmentgeneratordir="$(pkg-config --variable=prefix systemd || echo "/usr")/lib/systemd/system-environment-generators"
-		mkdir -p "${initdir}/${systemdsystemenvironmentgeneratordir}"
-		inst "${moddir}"/import-opts-generator.sh "${systemdsystemenvironmentgeneratordir}"/zfs-import-opts.sh
+		inst_simple "${moddir}/import-opts-generator.sh" "${systemdutildir}/system-environment-generators/zfs-import-opts.sh"
 	fi
 }

contrib/dracut/90zfs/mount-zfs.sh.in

@@ -3,48 +3,73 @@
 
 . /lib/dracut-zfs-lib.sh
 
-ZFS_DATASET=""
-ZFS_POOL=""
-
-case "${root}" in
-	zfs:*) ;;
-	*) return ;;
-esac
+decode_root_args || return 0
 
 GENERATOR_FILE=/run/systemd/generator/sysroot.mount
 GENERATOR_EXTENSION=/run/systemd/generator/sysroot.mount.d/zfs-enhancement.conf
 
-if [ -e "$GENERATOR_FILE" ] && [ -e "$GENERATOR_EXTENSION" ] ; then
-	# If the ZFS sysroot.mount flag exists, the initial RAM disk configured
-	# it to mount ZFS on root.  In that case, we bail early.  This flag
-	# file gets created by the zfs-generator program upon successful run.
-	info "ZFS: There is a sysroot.mount and zfs-generator has extended it."
-	info "ZFS: Delegating root mount to sysroot.mount."
-	# Let us tell the initrd to run on shutdown.
-	# We have a shutdown hook to run
-	# because we imported the pool.
+if [ -e "$GENERATOR_FILE" ] && [ -e "$GENERATOR_EXTENSION" ]; then
+	# We're under systemd and dracut-zfs-generator ran to completion.
+	info "ZFS: Delegating root mount to sysroot.mount at al."
+
 	# We now prevent Dracut from running this thing again.
-	for zfsmounthook in "$hookdir"/mount/*zfs* ; do
-		if [ -f "$zfsmounthook" ] ; then
-			rm -f "$zfsmounthook"
-		fi
-	done
+	rm -f "$hookdir"/mount/*zfs*
 	return
 fi
+
 info "ZFS: No sysroot.mount exists or zfs-generator did not extend it."
 info "ZFS: Mounting root with the traditional mount-zfs.sh instead."
 
+# ask_for_password tries prompt cmd
+#
+# Wraps around plymouth ask-for-password and adds fallback to tty password ask
+# if plymouth is not present.
+ask_for_password() {
+    tries="$1"
+    prompt="$2"
+    cmd="$3"
+
+    {
+        flock -s 9
+
+        # Prompt for password with plymouth, if installed and running.
+        if plymouth --ping 2>/dev/null; then
+            plymouth ask-for-password \
+                --prompt "$prompt" --number-of-tries="$tries" | \
+                eval "$cmd"
+            ret=$?
+        else
+            i=1
+            while [ "$i" -le "$tries" ]; do
+                printf "%s [%i/%i]:" "$prompt" "$i" "$tries" >&2
+                eval "$cmd" && ret=0 && break
+                ret=$?
+                i=$((i+1))
+                printf '\n' >&2
+            done
+            unset i
+        fi
+    } 9>/.console_lock
+
+    [ "$ret" -ne 0 ] && echo "Wrong password" >&2
+    return "$ret"
+}
+
+
 # Delay until all required block devices are present.
 modprobe zfs 2>/dev/null
 udevadm settle
 
+ZFS_DATASET=
+ZFS_POOL=
+
 if [ "${root}" = "zfs:AUTO" ] ; then
-	if ! ZFS_DATASET="$(find_bootfs)" ; then
+	if ! ZFS_DATASET="$(zpool get -Ho value bootfs | grep -m1 -vFx -)"; then
 		# shellcheck disable=SC2086
 		zpool import -N -a ${ZPOOL_IMPORT_OPTS}
-		if ! ZFS_DATASET="$(find_bootfs)" ; then
+		if ! ZFS_DATASET="$(zpool get -Ho value bootfs | grep -m1 -vFx -)"; then
 			warn "ZFS: No bootfs attribute found in importable pools."
-			export_all -F
+			zpool export -aF
 
 			rootok=0
 			return 1
@@ -53,34 +78,43 @@ if [ "${root}" = "zfs:AUTO" ] ; then
 	info "ZFS: Using ${ZFS_DATASET} as root."
 fi
 
-ZFS_DATASET="${ZFS_DATASET:-${root#zfs:}}"
+ZFS_DATASET="${ZFS_DATASET:-${root}}"
 ZFS_POOL="${ZFS_DATASET%%/*}"
 
-if import_pool "${ZFS_POOL}" ; then
-	# Load keys if we can or if we need to
-	if [ "$(zpool list -H -o feature@encryption "${ZFS_POOL}")" = 'active' ]; then
-		# if the root dataset has encryption enabled
-		ENCRYPTIONROOT="$(zfs get -H -o value encryptionroot "${ZFS_DATASET}")"
-		if ! [ "${ENCRYPTIONROOT}" = "-" ]; then
-			KEYSTATUS="$(zfs get -H -o value keystatus "${ENCRYPTIONROOT}")"
-			# if the key needs to be loaded
-			if [ "$KEYSTATUS" = "unavailable" ]; then
-				# decrypt them
-				ask_for_password \
-					--tries 5 \
-					--prompt "Encrypted ZFS password for ${ENCRYPTIONROOT}: " \
-					--cmd "zfs load-key '${ENCRYPTIONROOT}'"
-			fi
+
+if ! zpool get -Ho name "${ZFS_POOL}" > /dev/null 2>&1; then
+    info "ZFS: Importing pool ${ZFS_POOL}..."
+    # shellcheck disable=SC2086
+    if ! zpool import -N ${ZPOOL_IMPORT_OPTS} "${ZFS_POOL}"; then
+        warn "ZFS: Unable to import pool ${ZFS_POOL}"
+        rootok=0
+        return 1
+    fi
+fi
+
+# Load keys if we can or if we need to
+# TODO: for_relevant_root_children like in zfs-load-key.sh.in
+if [ "$(zpool get -Ho value feature@encryption "${ZFS_POOL}")" = 'active' ]; then
+	# if the root dataset has encryption enabled
+	ENCRYPTIONROOT="$(zfs get -Ho value encryptionroot "${ZFS_DATASET}")"
+	if ! [ "${ENCRYPTIONROOT}" = "-" ]; then
+		KEYSTATUS="$(zfs get -Ho value keystatus "${ENCRYPTIONROOT}")"
+		# if the key needs to be loaded
+		if [ "$KEYSTATUS" = "unavailable" ]; then
+			# decrypt them
+			ask_for_password \
+				5 \
+				"Encrypted ZFS password for ${ENCRYPTIONROOT}: " \
+				"zfs load-key '${ENCRYPTIONROOT}'"
 		fi
 	fi
-	# Let us tell the initrd to run on shutdown.
-	# We have a shutdown hook to run
-	# because we imported the pool.
-	info "ZFS: Mounting dataset ${ZFS_DATASET}..."
-	if mount_dataset "${ZFS_DATASET}" ; then
-		ROOTFS_MOUNTED=yes
-		return 0
-	fi
 fi
 
-rootok=0
+# Let us tell the initrd to run on shutdown.
+# We have a shutdown hook to run
+# because we imported the pool.
+info "ZFS: Mounting dataset ${ZFS_DATASET}..."
+if ! mount_dataset "${ZFS_DATASET}"; then
+  rootok=0
+  return 1
+fi

contrib/dracut/90zfs/parse-zfs.sh.in

@@ -1,7 +1,8 @@
 #!/bin/sh
 # shellcheck disable=SC2034,SC2154
 
-. /lib/dracut-lib.sh
+# shellcheck source=zfs-lib.sh.in
+. /lib/dracut-zfs-lib.sh
 
 # Let the command line override our host id.
 spl_hostid=$(getarg spl_hostid=)
@@ -15,49 +16,20 @@ else
 	warn "ZFS: Pools may not import correctly."
 fi
 
-wait_for_zfs=0
-case "${root}" in
-	""|zfs|zfs:)
-		# We'll take root unset, root=zfs, or root=zfs:
-		# No root set, so we want to read the bootfs attribute.  We
-		# can't do that until udev settles so we'll set dummy values
-		# and hope for the best later on.
-		root="zfs:AUTO"
-		rootok=1
-		wait_for_zfs=1
+if decode_root_args; then
+	if [ "$root" = "zfs:AUTO" ]; then
+		info "ZFS: Boot dataset autodetected from bootfs=."
+	else
+		info "ZFS: Boot dataset is ${root}."
+	fi
 
-		info "ZFS: Enabling autodetection of bootfs after udev settles."
-		;;
-
-	ZFS=*|zfs:*|FILESYSTEM=*)
-		# root is explicit ZFS root.  Parse it now.  We can handle
-		# a root=... param in any of the following formats:
-		# root=ZFS=rpool/ROOT
-		# root=zfs:rpool/ROOT
-		# root=zfs:FILESYSTEM=rpool/ROOT
-		# root=FILESYSTEM=rpool/ROOT
-		# root=ZFS=pool+with+space/ROOT+WITH+SPACE (translates to root=ZFS=pool with space/ROOT WITH SPACE)
-
-		# Strip down to just the pool/fs
-		root="${root#zfs:}"
-		root="${root#FILESYSTEM=}"
-		root="zfs:${root#ZFS=}"
-		# switch + with spaces because kernel cmdline does not allow us to quote parameters
-		root=$(echo "$root" | tr '+' ' ')
-		rootok=1
-		wait_for_zfs=1
-
-		info "ZFS: Set ${root} as bootfs."
-		;;
-esac
-
-# Make sure Dracut is happy that we have a root and will wait for ZFS
-# modules to settle before mounting.
-if [ ${wait_for_zfs} -eq 1 ]; then
-	ln -s /dev/null /dev/root 2>/dev/null
-	initqueuedir="${hookdir}/initqueue/finished"
-	test -d "${initqueuedir}" || {
-		initqueuedir="${hookdir}/initqueue-finished"
-	}
-	echo '[ -e /dev/zfs ]' > "${initqueuedir}/zfs.sh"
+	rootok=1
+	# Make sure Dracut is happy that we have a root and will wait for ZFS
+	# modules to settle before mounting.
+	if [ -n "${wait_for_zfs}" ]; then
+		ln -s null /dev/root
+		echo '[ -e /dev/zfs ]' > "${hookdir}/initqueue/finished/zfs.sh"
+	fi
+else
+	info "ZFS: no ZFS-on-root."
 fi

contrib/dracut/90zfs/zfs-env-bootfs.service.in

@@ -8,7 +8,7 @@ Before=zfs-import.target
 
 [Service]
 Type=oneshot
-ExecStart=/bin/sh -c "exec systemctl set-environment BOOTFS=$(@sbindir@/zpool list -H -o bootfs | grep -m1 -v '^-$')"
+ExecStart=/bin/sh -c "exec systemctl set-environment BOOTFS=$(@sbindir@/zpool list -H -o bootfs | grep -m1 -vFx -)"
 
 [Install]
 WantedBy=zfs-import.target

contrib/dracut/90zfs/zfs-generator.sh.in

@@ -1,5 +1,5 @@
 #!/bin/sh
-# shellcheck disable=SC2016,SC1004
+# shellcheck disable=SC2016,SC1004,SC2154
 
 grep -wq debug /proc/cmdline && debug=1
 [ -n "$debug" ] && echo "zfs-generator: starting" >> /dev/kmsg
@@ -10,37 +10,17 @@ GENERATOR_DIR="$1"
     exit 1
 }
 
-[ -f /lib/dracut-lib.sh ] && dracutlib=/lib/dracut-lib.sh
-[ -f /usr/lib/dracut/modules.d/99base/dracut-lib.sh ] && dracutlib=/usr/lib/dracut/modules.d/99base/dracut-lib.sh
-command -v getarg >/dev/null 2>&1 || {
-    [ -n "$debug" ] && echo "zfs-generator: loading Dracut library from $dracutlib" >> /dev/kmsg
-    . "$dracutlib"
-}
-
+# shellcheck source=zfs-lib.sh.in
 . /lib/dracut-zfs-lib.sh
+decode_root_args || exit 0
 
-[ -z "$root" ]       && root=$(getarg root=)
-[ -z "$rootfstype" ] && rootfstype=$(getarg rootfstype=)
-[ -z "$rootflags" ]  && rootflags=$(getarg rootflags=)
-
-# If root is not ZFS= or zfs: or rootfstype is not zfs
-# then we are not supposed to handle it.
-[ "${root##zfs:}" = "${root}" ] &&
-	[ "${root##ZFS=}" = "${root}" ] &&
-	[ "$rootfstype" != "zfs" ] &&
-	exit 0
-
+[ -z "${rootflags}" ] && rootflags=$(getarg rootflags=)
 case ",${rootflags}," in
 	*,zfsutil,*) ;;
 	,,)	rootflags=zfsutil ;;
 	*)	rootflags="zfsutil,${rootflags}" ;;
 esac
 
-if [ "${root}" != "zfs:AUTO" ]; then
-  root="${root##zfs:}"
-  root="${root##ZFS=}"
-fi
-
 [ -n "$debug" ] && echo "zfs-generator: writing extension for sysroot.mount to $GENERATOR_DIR/sysroot.mount.d/zfs-enhancement.conf" >> /dev/kmsg
 
 
@@ -89,7 +69,7 @@ else
   _zfs_generator_cb() {
       dset="${1}"
       mpnt="${2}"
-      unit="sysroot$(echo "$mpnt" | tr '/' '-').mount"
+      unit="$(systemd-escape --suffix=mount -p "/sysroot${mpnt}")"
 
       {
           echo "[Unit]"

contrib/dracut/90zfs/zfs-lib.sh.in

@@ -1,74 +1,16 @@
 #!/bin/sh
+# shellcheck disable=SC2034
 
-command -v getarg >/dev/null || . /lib/dracut-lib.sh
-command -v getargbool >/dev/null || {
-    # Compatibility with older Dracut versions.
-    # With apologies to the Dracut developers.
-    getargbool() {
-        _default="$1"; shift
-        ! _b=$(getarg "$@") && [ -z "$_b" ] && _b="$_default"
-        if [ -n "$_b" ]; then
-            [ "$_b" = "0" ] && return 1
-            [ "$_b" = "no" ] && return 1
-            [ "$_b" = "off" ] && return 1
-        fi
-        return 0
-    }
-}
+command -v getarg >/dev/null || . /lib/dracut-lib.sh || . /usr/lib/dracut/modules.d/99base/dracut-lib.sh
 
-OLDIFS="${IFS}"
-NEWLINE="
-"
 TAB="	"
 
-ZPOOL_IMPORT_OPTS=""
-if getargbool 0 zfs_force -y zfs.force -y zfsforce ; then
+ZPOOL_IMPORT_OPTS=
+if getargbool 0 zfs_force -y zfs.force -y zfsforce; then
     warn "ZFS: Will force-import pools if necessary."
-    ZPOOL_IMPORT_OPTS="${ZPOOL_IMPORT_OPTS} -f"
+    ZPOOL_IMPORT_OPTS=-f
 fi
 
-# find_bootfs
-#   returns the first dataset with the bootfs attribute.
-find_bootfs() {
-    IFS="${NEWLINE}"
-    for dataset in $(zpool list -H -o bootfs); do
-        case "${dataset}" in
-            "" | "-")
-                continue
-                ;;
-            "no pools available")
-                IFS="${OLDIFS}"
-                return 1
-                ;;
-            *)
-                IFS="${OLDIFS}"
-                echo "${dataset}"
-                return 0
-                ;;
-        esac
-    done
-
-    IFS="${OLDIFS}"
-    return 1
-}
-
-# import_pool POOL
-#   imports the given zfs pool if it isn't imported already.
-import_pool() {
-    pool="${1}"
-
-    if ! zpool list -H "${pool}" > /dev/null 2>&1; then
-        info "ZFS: Importing pool ${pool}..."
-        # shellcheck disable=SC2086
-        if ! zpool import -N ${ZPOOL_IMPORT_OPTS} "${pool}" ; then
-            warn "ZFS: Unable to import pool ${pool}"
-            return 1
-        fi
-    fi
-
-    return 0
-}
-
 _mount_dataset_cb() {
     mount -o zfsutil -t zfs "${1}" "${NEWROOT}${2}"
 }
@@ -121,87 +63,57 @@ for_relevant_root_children() {
         )
 }
 
-# export_all OPTS
-#   exports all imported zfs pools.
-export_all() {
-    ret=0
-
-    IFS="${NEWLINE}"
-    for pool in $(zpool list -H -o name) ; do
-        if zpool list -H "${pool}" > /dev/null 2>&1; then
-            zpool export "${pool}" "$@" || ret=$?
-        fi
-    done
-    IFS="${OLDIFS}"
-
-    return ${ret}
-}
-
-# ask_for_password
+# Parse root=, rootfstype=, return them decoded and normalised to zfs:AUTO for auto, plain dset for explicit
 #
-# Wraps around plymouth ask-for-password and adds fallback to tty password ask
-# if plymouth is not present.
+# True if ZFS-on-root, false if we shouldn't
 #
-# --cmd command
-#   Command to execute. Required.
-# --prompt prompt
-#   Password prompt. Note that function already adds ':' at the end.
-#   Recommended.
-# --tries n
-#   How many times repeat command on its failure.  Default is 3.
-# --ply-[cmd|prompt|tries]
-#   Command/prompt/tries specific for plymouth password ask only.
-# --tty-[cmd|prompt|tries]
-#   Command/prompt/tries specific for tty password ask only.
-# --tty-echo-off
-#   Turn off input echo before tty command is executed and turn on after.
-#   It's useful when password is read from stdin.
-ask_for_password() {
-    ply_tries=3
-    tty_tries=3
-    while [ "$#" -gt 0 ]; do
-        case "$1" in
-            --cmd) ply_cmd="$2"; tty_cmd="$2"; shift;;
-            --ply-cmd) ply_cmd="$2"; shift;;
-            --tty-cmd) tty_cmd="$2"; shift;;
-            --prompt) ply_prompt="$2"; tty_prompt="$2"; shift;;
-            --ply-prompt) ply_prompt="$2"; shift;;
-            --tty-prompt) tty_prompt="$2"; shift;;
-            --tries) ply_tries="$2"; tty_tries="$2"; shift;;
-            --ply-tries) ply_tries="$2"; shift;;
-            --tty-tries) tty_tries="$2"; shift;;
-            --tty-echo-off) tty_echo_off=yes;;
+# Supported values:
+#   root=
+#   root=zfs
+#   root=zfs:
+#   root=zfs:AUTO
+#
+#   root=ZFS=data/set
+#   root=zfs:data/set
+#   root=zfs:ZFS=data/set (as a side-effect; allowed but undocumented)
+#
+#   rootfstype=zfs AND root=data/set <=> root=data/set
+#   rootfstype=zfs AND root=         <=> root=zfs:AUTO
+#
+# '+'es in explicit dataset decoded to ' 's.
+decode_root_args() {
+    if [ -n "$rootfstype" ]; then
+        [ "$rootfstype" = zfs ]
+        return
+    fi
+
+    root=$(getarg root=)
+    rootfstype=$(getarg rootfstype=)
+
+    # shellcheck disable=SC2249
+    case "$root" in
+        ""|zfs|zfs:|zfs:AUTO)
+            root=zfs:AUTO
+            rootfstype=zfs
+            return 0
+            ;;
+
+        ZFS=*|zfs:*)
+            root="${root#zfs:}"
+            root="${root#ZFS=}"
+            root=$(echo "$root" | tr '+' ' ')
+            rootfstype=zfs
+            return 0
+            ;;
+    esac
+
+    if [ "$rootfstype" = "zfs" ]; then
+        case "$root" in
+            "") root=zfs:AUTO ;;
+            *)  root=$(echo "$root" | tr '+' ' ') ;;
         esac
-        shift
-    done
+        return 0
+    fi
 
-    { flock -s 9;
-        # Prompt for password with plymouth, if installed and running.
-        if plymouth --ping 2>/dev/null; then
-            plymouth ask-for-password \
-                --prompt "$ply_prompt" --number-of-tries="$ply_tries" | \
-                eval "$ply_cmd"
-            ret=$?
-        else
-            if [ "$tty_echo_off" = yes ]; then
-                stty_orig="$(stty -g)"
-                stty -echo
-            fi
-
-            i=1
-            while [ "$i" -le "$tty_tries" ]; do
-                [ -n "$tty_prompt" ] && \
-                    printf "%s [%i/%i]:" "$tty_prompt" "$i" "$tty_tries" >&2
-                eval "$tty_cmd" && ret=0 && break
-                ret=$?
-                i=$((i+1))
-                [ -n "$tty_prompt" ] && printf '\n' >&2
-            done
-            unset i
-            [ "$tty_echo_off" = yes ] && stty "$stty_orig"
-        fi
-    } 9>/.console_lock
-
-    [ $ret -ne 0 ] && echo "Wrong password" >&2
-    return $ret
+    return 1
 }

contrib/dracut/90zfs/zfs-load-key.sh.in

@@ -4,70 +4,61 @@
 # only run this on systemd systems, we handle the decrypt in mount-zfs.sh in the mount hook otherwise
 [ -e /bin/systemctl ] || [ -e /usr/bin/systemctl ] || return 0
 
-# This script only gets executed on systemd systems, see mount-zfs.sh for non-systemd systems
+# shellcheck source=zfs-lib.sh.in
+. /lib/dracut-zfs-lib.sh
 
-# import the libs now that we know the pool imported
-[ -f /lib/dracut-lib.sh ] && dracutlib=/lib/dracut-lib.sh
-[ -f /usr/lib/dracut/modules.d/99base/dracut-lib.sh ] && dracutlib=/usr/lib/dracut/modules.d/99base/dracut-lib.sh
-# shellcheck source=./lib-zfs.sh.in
-. "$dracutlib"
-
-# load the kernel command line vars
-[ -z "$root" ] && root="$(getarg root=)"
-# If root is not ZFS= or zfs: or rootfstype is not zfs then we are not supposed to handle it.
-[ "${root##zfs:}" = "${root}" ] && [ "${root##ZFS=}" = "${root}" ] && [ "$rootfstype" != "zfs" ] && exit 0
+decode_root_args || return 0
 
 # There is a race between the zpool import and the pre-mount hooks, so we wait for a pool to be imported
-while [ "$(zpool list -H)" = "" ]; do
-    systemctl is-failed --quiet zfs-import-cache.service zfs-import-scan.service && exit 1
+while ! systemctl is-active --quiet zfs-import.target; do
+    systemctl is-failed --quiet zfs-import-cache.service zfs-import-scan.service && return 1
     sleep 0.1s
 done
 
-# run this after import as zfs-import-cache/scan service is confirmed good
-# we do not overwrite the ${root} variable, but create a new one, BOOTFS, to hold the dataset
-if [ "${root}" = "zfs:AUTO" ] ; then
-    BOOTFS="$(zpool list -H -o bootfs | awk '$1 != "-" {print; exit}')"
-else
-    BOOTFS="${root##zfs:}"
-    BOOTFS="${BOOTFS##ZFS=}"
+BOOTFS="$root"
+if [ "$BOOTFS" = "zfs:AUTO" ]; then
+    BOOTFS="$(zpool get -Ho value bootfs | grep -m1 -vFx -)"
 fi
 
-# if pool encryption is active and the zfs command understands '-o encryption'
-if [ "$(zpool list -H -o feature@encryption "${BOOTFS%%/*}")" = 'active' ]; then
-    # if the root dataset has encryption enabled
-    ENCRYPTIONROOT="$(zfs get -H -o value encryptionroot "${BOOTFS}")"
-    if ! [ "${ENCRYPTIONROOT}" = "-" ]; then
-        KEYSTATUS="$(zfs get -H -o value keystatus "${ENCRYPTIONROOT}")"
-        # continue only if the key needs to be loaded
-        [ "$KEYSTATUS" = "unavailable" ] || exit 0
+[ "$(zpool get -Ho value feature@encryption "${BOOTFS%%/*}")" = 'active' ] || return 0
 
-        KEYLOCATION="$(zfs get -H -o value keylocation "${ENCRYPTIONROOT}")"
-        case "${KEYLOCATION%%://*}" in
-            prompt)
-                for _ in 1 2 3; do
-                    systemd-ask-password --no-tty "Encrypted ZFS password for ${BOOTFS}" | zfs load-key "${ENCRYPTIONROOT}" && break
+_load_key_cb() {
+    dataset="$1"
+
+    ENCRYPTIONROOT="$(zfs get -Ho value encryptionroot "${dataset}")"
+    [ "${ENCRYPTIONROOT}" = "-" ] && return 0
+
+    [ "$(zfs get -Ho value keystatus "${ENCRYPTIONROOT}")" = "unavailable" ] || return 0
+
+    KEYLOCATION="$(zfs get -Ho value keylocation "${ENCRYPTIONROOT}")"
+    case "${KEYLOCATION%%://*}" in
+        prompt)
+            for _ in 1 2 3; do
+                systemd-ask-password --no-tty "Encrypted ZFS password for ${dataset}" | zfs load-key "${ENCRYPTIONROOT}" && break
+            done
+            ;;
+        http*)
+            systemctl start network-online.target
+            zfs load-key "${ENCRYPTIONROOT}"
+            ;;
+        file)
+            KEYFILE="${KEYLOCATION#file://}"
+            [ -r "${KEYFILE}" ] || udevadm settle
+            [ -r "${KEYFILE}" ] || {
+                info "ZFS: Waiting for key ${KEYFILE} for ${ENCRYPTIONROOT}..."
+                for _ in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20; do
+                    sleep 0.5s
+                    [ -r "${KEYFILE}" ] && break
                 done
-                ;;
-            http*)
-                systemctl start network-online.target
-                zfs load-key "${ENCRYPTIONROOT}"
-                ;;
-            file)
-                KEYFILE="${KEYLOCATION#file://}"
-                [ -r "${KEYFILE}" ] || udevadm settle
-                [ -r "${KEYFILE}" ] || {
-                    info "Waiting for key ${KEYFILE} for ${ENCRYPTIONROOT}..."
-                    for _ in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20; do
-                        sleep 0.5s
-                        [ -r "${KEYFILE}" ] && break
-                    done
-                }
-                [ -r "${KEYFILE}" ] || warn "Key ${KEYFILE} for ${ENCRYPTIONROOT} hasn't appeared. Trying anyway."
-                zfs load-key "${ENCRYPTIONROOT}"
-                ;;
-            *)
-                zfs load-key "${ENCRYPTIONROOT}"
-                ;;
-        esac
-    fi
-fi
+            }
+            [ -r "${KEYFILE}" ] || warn "ZFS: Key ${KEYFILE} for ${ENCRYPTIONROOT} hasn't appeared. Trying anyway."
+            zfs load-key "${ENCRYPTIONROOT}"
+            ;;
+        *)
+            zfs load-key "${ENCRYPTIONROOT}"
+            ;;
+    esac
+}
+
+_load_key_cb "$BOOTFS"
+for_relevant_root_children "$BOOTFS" _load_key_cb

contrib/dracut/90zfs/zfs-needshutdown.sh.in

@@ -2,7 +2,7 @@
 
 command -v getarg >/dev/null 2>&1 || . /lib/dracut-lib.sh
 
-if zpool list 2>&1 | grep -q 'no pools available' ; then
+if [ -z "$(zpool get -Ho value name)" ]; then
     info "ZFS: No active pools, no need to export anything."
 else
     info "ZFS: There is an active pool, will export it."

contrib/dracut/90zfs/zfs-rollback-bootfs.service.in

@@ -1,14 +1,12 @@
 [Unit]
 Description=Rollback bootfs just before it is mounted
 Requisite=zfs-import.target
-After=zfs-import.target zfs-snapshot-bootfs.service
+After=zfs-import.target dracut-pre-mount.service zfs-snapshot-bootfs.service
 Before=dracut-mount.service
 DefaultDependencies=no
 ConditionKernelCommandLine=bootfs.rollback
 
 [Service]
-# ${BOOTFS} should have been set by zfs-env-bootfs.service
 Type=oneshot
-ExecStartPre=/bin/sh -c 'test -n "${BOOTFS}"'
-ExecStart=/bin/sh -c '. /lib/dracut-lib.sh; SNAPNAME="$(getarg bootfs.rollback)"; exec @sbindir@/zfs rollback -Rf "${BOOTFS}@${SNAPNAME:-%v}"'
+ExecStart=/bin/sh -c '. /lib/dracut-zfs-lib.sh; decode_root_args || exit; [ "$root" = "zfs:AUTO" ] && root="$BOOTFS" SNAPNAME="$(getarg bootfs.rollback)"; exec @sbindir@/zfs rollback -Rf "$root@${SNAPNAME:-%v}"'
 RemainAfterExit=yes

contrib/dracut/90zfs/zfs-snapshot-bootfs.service.in

@@ -1,14 +1,12 @@
 [Unit]
 Description=Snapshot bootfs just before it is mounted
 Requisite=zfs-import.target
-After=zfs-import.target
+After=zfs-import.target dracut-pre-mount.service
 Before=dracut-mount.service
 DefaultDependencies=no
 ConditionKernelCommandLine=bootfs.snapshot
 
 [Service]
-# ${BOOTFS} should have been set by zfs-env-bootfs.service
 Type=oneshot
-ExecStartPre=/bin/sh -c 'test -n "${BOOTFS}"'
-ExecStart=-/bin/sh -c '. /lib/dracut-lib.sh; SNAPNAME="$(getarg bootfs.snapshot)"; exec @sbindir@/zfs snapshot "${BOOTFS}@${SNAPNAME:-%v}"'
+ExecStart=/bin/sh -c '. /lib/dracut-zfs-lib.sh; decode_root_args || exit; [ "$root" = "zfs:AUTO" ] && root="$BOOTFS" SNAPNAME="$(getarg bootfs.snapshot)"; exec @sbindir@/zfs snapshot "$root@${SNAPNAME:-%v}"'
 RemainAfterExit=yes

contrib/dracut/README.dracut.markdown

@@ -1,225 +1,50 @@
-How to setup a zfs root filesystem using dracut
------------------------------------------------
+## Basic setup
+1. Install `zfs-dracut`
+2. Set `mountpoint=/` for your root dataset (for compatibility, `legacy` also works, but is not recommended for new installations):
+    ```sh
+    zfs set mountpoint=/ pool/dataset
+    ```
+3. Either (a) set `bootfs=` on the pool to the dataset:
+    ```sh
+    zpool set bootfs=pool/dataset pool
+    ```
+4. Or (b) append `root=zfs:pool/dataset` to your kernel cmdline.
+5. Re-generate your initrd and update it in your boot bundle
 
-1) Install the zfs-dracut package.  This package adds a zfs dracut module
-to the /usr/share/dracut/modules.d/ directory which allows dracut to
-create an initramfs which is zfs aware.
+Encrypted datasets have keys loaded automatically or prompted for.
 
-2) Set the bootfs property for the bootable dataset in the pool.  Then set
-the dataset mountpoint property to '/'.
+If the root dataset contains children with `mountpoint=`s of `/etc`, `/bin`, `/lib*`, or `/usr`, they're mounted too.
 
-    $ zpool set bootfs=pool/dataset pool
-    $ zfs set mountpoint=/ pool/dataset
+For complete documentation, see `dracut.zfs(7)`.
 
-Alternately, legacy mountpoints can be used by setting the 'root=' option
-on the kernel line of your grub.conf/menu.lst configuration file.  Then
-set the dataset mountpoint property to 'legacy'.
+## cmdline
+1. `root=`                    | Root dataset is…                                         |
+   ---------------------------|----------------------------------------------------------|
+   *(empty)*                  | the first `bootfs=` after `zpool import -aN`             |
+   `zfs:AUTO`, `zfs:`, `zfs`  | *(as above, but overriding other autoselection methods)* |
+   `ZFS=pool/dataset`         | `pool/dataset`                                           |
+   `zfs:pool/dataset`         | *(as above)*                                             |
 
-    $ grub.conf/menu.lst: kernel ... root=ZFS=pool/dataset
-    $ zfs set mountpoint=legacy pool/dataset
+   All `+`es are replaced with spaces (i.e. to boot from `root pool/data set`, pass `root=zfs:root+pool/data+set`).
 
-3) To set zfs module options put them in /etc/modprobe.d/zfs.conf file.
-The complete list of zfs module options is available by running the
-_modinfo zfs_ command.  Commonly set options include: zfs_arc_min,
-zfs_arc_max, zfs_prefetch_disable, and zfs_vdev_max_pending.
+   The dataset can be at any depth, including being the pool's root dataset (i.e. `root=zfs:pool`).
 
-4) Finally, create your new initramfs by running dracut.
+   `rootfstype=zfs` is equivalent to `root=zfs:AUTO`, `rootfstype=zfs root=pool/dataset` is equivalent to `root=zfs:pool/dataset`.
 
-    $ dracut --force /path/to/initramfs kernel_version
+2. `spl_hostid`: passed to `zgenhostid -f`, useful to override the `/etc/hostid` file baked into the initrd.
 
-Kernel Command Line
--------------------
+3. `bootfs.snapshot`, `bootfs.snapshot=snapshot-name`: enables `zfs-snapshot-bootfs.service`,
+   which creates a snapshot `$root_dataset@$(uname -r)` (or, in the second form, `$root_dataset@snapshot-name`)
+   after pool import but before the rootfs is mounted.
+   Failure to create the snapshot is noted, but booting continues.
 
-The initramfs' behavior is influenced by the following kernel command line
-parameters passed in from the boot loader:
+4. `bootfs.rollback`, `bootfs.rollback=snapshot-name`: enables `zfs-snapshot-bootfs.service`,
+   which `-Rf` rolls back to `$root_dataset@$(uname -r)` (or, in the second form, `$root_dataset@snapshot-name`)
+   after pool import but before the rootfs is mounted.
+   Failure to roll back will fall down to the rescue shell.
+   This has obvious potential for data loss: make sure your persistent data is not below the rootfs and you don't care about any intermediate snapshots.
 
-* `root=...`: If not set, importable pools are searched for a bootfs
-attribute.  If an explicitly set root is desired, you may use
-`root=ZFS:pool/dataset`
+5. If both `bootfs.snapshot` and `bootfs.rollback` are set, `bootfs.rollback` is ordered *after* `bootfs.snapshot`.
 
-* `zfs_force=0`: If set to 1, the initramfs will run `zpool import -f` when
-attempting to import pools if the required pool isn't automatically imported
-by the zfs module.  This can save you a trip to a bootcd if hostid has
-changed, but is dangerous and can lead to zpool corruption, particularly in
-cases where storage is on a shared fabric such as iSCSI where multiple hosts
-can access storage devices concurrently.  _Please understand the implications
-of force-importing a pool before enabling this option!_
-
-* `spl_hostid`: By default, the hostid used by the SPL module is read from
-/etc/hostid inside the initramfs.  This file is placed there from the host
-system when the initramfs is built which effectively ties the ramdisk to the
-host which builds it.  If a different hostid is desired, one may be set in
-this attribute and will override any file present in the ramdisk.  The
-format should be hex exactly as found in the `/etc/hostid` file, IE
-`spl_hostid=0x00bab10c`.
-
-Note that changing the hostid between boots will most likely lead to an
-un-importable pool since the last importing hostid won't match.  In order
-to recover from this, you may use the `zfs_force` option or boot from a
-different filesystem and `zpool import -f` then `zpool export` the pool
-before rebooting with the new hostid.
-
-* `bootfs.snapshot`: If listed, enables the zfs-snapshot-bootfs service on a Dracut system. The zfs-snapshot-bootfs service simply runs `zfs snapshot $BOOTFS@%v` after the pool has been imported but before the bootfs is mounted. `$BOOTFS` is substituted with the value of the bootfs setting on the pool. `%v` is substituted with the version string of the kernel currently being booted (e.g. 5.6.6-200.fc31.x86\_64). Failure to create the snapshot (e.g. because one with the same name already exists) will be logged, but will not otherwise interrupt the boot process.
-
-    It is safe to leave the bootfs.snapshot flag set persistently on your kernel command line so that a new snapshot of your bootfs will be created on every kernel update. If you leave bootfs.snapshot set persistently on your kernel command line, you may find the below script helpful for automatically removing old snapshots of the bootfs along with their associated kernel.
-
-        #!/usr/bin/sh
-
-        if [[ "$1" == "remove" ]] && grep -q "\bbootfs.snapshot\b" /proc/cmdline; then
-           zfs destroy $(findmnt -n -o source /)@$2 &> /dev/null
-        fi
-
-        exit 0
-
-    To use the above script place it in a plain text file named /etc/kernel/install.d/99-zfs-cleanup.install and mark it executable with the following command:
-
-        $ chmod +x /etc/kernel/install.d/99-zfs-cleanup.install
-
-    On Red Hat based systems, you can change the value of `installonly_limit` in /etc/dnf/dnf.conf to adjust the number of kernels and their associated snapshots that are kept.
-
-* `bootfs.snapshot=<snapname>`: Is identical to the bootfs.snapshot parameter explained above except that the value substituted for \<snapname\> will be used when creating the snapshot instead of the version string of the kernel currently being booted. 
-
-* `bootfs.rollback`: If listed, enables the zfs-rollback-bootfs service on a Dracut system. The zfs-rollback-bootfs service simply runs `zfs rollback -Rf $BOOTFS@%v` after the pool has been imported but before the bootfs is mounted. If the rollback operation fails, the boot process will be interrupted with a Dracut rescue shell. __Use this parameter with caution. Intermediate snapshots of the bootfs will be destroyed!__ TIP: Keep your user data (e.g. /home) on separate file systems (it can be in the same pool though).
-
-* `bootfs.rollback=<snapname>`: Is identical to the bootfs.rollback parameter explained above except that the value substituted for \<snapname\> will be used when rolling back the bootfs instead of the version string of the kernel currently being booted. If you use this form, choose a snapshot that is new enough to contain the needed kernel modules under /lib/modules or use a kernel that has all the needed modules built-in.
-
-How it Works
-============
-
-The Dracut module consists of the following files (less Makefile's):
-
-* `module-setup.sh`: Script run by the initramfs builder to create the
-ramdisk.  Contains instructions on which files are required by the modules
-and z* programs.  Also triggers inclusion of `/etc/hostid` and the zpool
-cache.  This file is not included in the initramfs.
-
-* `90-zfs.rules`: udev rules which trigger loading of the ZFS modules at boot.
-
-* `zfs-lib.sh`: Utility functions used by the other files.
-
-* `parse-zfs.sh`: Run early in the initramfs boot process to parse kernel
-command line and determine if ZFS is the active root filesystem.
-
-* `mount-zfs.sh`: Run later in initramfs boot process after udev has settled
-to mount the root dataset.
-
-* `export-zfs.sh`: Run on shutdown after dracut has restored the initramfs
-and pivoted to it, allowing for a clean unmount and export of the ZFS root.
-
-`zfs-lib.sh`
-------------
-
-This file provides a few handy functions for working with ZFS. Those
-functions are used by the `mount-zfs.sh` and `export-zfs.sh` files.
-However, they could be used by any other file as well, as long as the file
-sources `/lib/dracut-zfs-lib.sh`.
-
-`module-setup.sh`
------------------
-
-This file is run by the Dracut script within the live system, not at boot
-time.  It's not included in the final initramfs.  Functions in this script
-describe which files are needed by ZFS at boot time.
-
-Currently all the various z* and spl modules are included, a dependency is
-asserted on udev-rules, and the various zfs, zpool, etc. helpers are included.
-Dracut provides library functions which automatically gather the shared libs
-necessary to run each of these binaries, so statically built binaries are
-not required.
-
-The zpool and zvol udev rules files are copied from where they are
-installed by the ZFS build.  __PACKAGERS TAKE NOTE__: If you move
-`/etc/udev/rules/60-z*.rules`, you'll need to update this file to match.
-
-Currently this file also includes `/etc/hostid` and `/etc/zfs/zpool.cache`
-which means the generated ramdisk is specific to the host system which built
-it.  If a generic initramfs is required, it may be preferable to omit these
-files and specify the `spl_hostid` from the boot loader instead.
-
-`parse-zfs.sh`
---------------
-
-Run during the cmdline phase of the initramfs boot process, this script
-performs some basic sanity checks on kernel command line parameters to
-determine if booting from ZFS is likely to be what is desired.  Dracut
-requires this script to adjust the `root` variable if required and to set
-`rootok=1` if a mountable root filesystem is available.  Unfortunately this
-script must run before udev is settled and kernel modules are known to be
-loaded, so accessing the zpool and zfs commands is unsafe.
-
-If the root=ZFS... parameter is set on the command line, then it's at least
-certain that ZFS is what is desired, though this script is unable to
-determine if ZFS is in fact available.  This script will alter the `root`
-parameter to replace several historical forms of specifying the pool and
-dataset name with the canonical form of `zfs:pool/dataset`.
-
-If no root= parameter is set, the best this script can do is guess that
-ZFS is desired.  At present, no other known filesystems will work with no
-root= parameter, though this might possibly interfere with using the
-compiled-in default root in the kernel image.  It's considered unlikely
-that would ever be the case when an initramfs is in use, so this script
-sets `root=zfs:AUTO` and hopes for the best.
-
-Once the root=... (or lack thereof) parameter is parsed, a dummy symlink
-is created from `/dev/root` -> `/dev/null` to satisfy parts of the Dracut
-process which check for presence of a single root device node.
-
-Finally, an initqueue/finished hook is registered which causes the initqueue
-phase of Dracut to wait for `/dev/zfs` to become available before attempting
-to mount anything.
-
-`mount-zfs.sh`
---------------
-
-This script is run after udev has settled and all tasks in the initqueue
-have succeeded.  This ensures that `/dev/zfs` is available and that the
-various ZFS modules are successfully loaded.  As it is now safe to call
-zpool and friends, we can proceed to find the bootfs attribute if necessary.
-
-If the root parameter was explicitly set on the command line, no parsing is
-necessary.  The list of imported pools is checked to see if the desired pool
-is already imported.  If it's not, and attempt is made to import the pool
-explicitly, though no force is attempted.  Finally the specified dataset
-is mounted on `$NEWROOT`, first using the `-o zfsutil` option to handle
-non-legacy mounts, then if that fails, without zfsutil to handle legacy
-mount points.
-
-If no root parameter was specified, this script attempts to find a pool with
-its bootfs attribute set.  First, already-imported pools are scanned and if
-an appropriate pool is found, no additional pools are imported.  If no pool
-with bootfs is found, any additional pools in the system are imported with
-`zpool import -N -a`, and the scan for bootfs is tried again.  If no bootfs
-is found with all pools imported, all pools are re-exported, and boot fails.
-Assuming a bootfs is found, an attempt is made to mount it to `$NEWROOT`,
-first with, then without the zfsutil option as above.
-
-Ordinarily pools are imported _without_ the force option which may cause
-boot to fail if the hostid has changed or a pool has been physically moved
-between servers.  The `zfs_force` kernel parameter is provided which when
-set to `1` causes `zpool import` to be run with the `-f` flag.  Forcing pool
-import can lead to serious data corruption and loss of pools, so this option
-should be used with extreme caution.  Note that even with this flag set, if
-the required zpool was auto-imported by the kernel module, no additional
-`zpool import` commands are run, so nothing is forced.
-
-`export-zfs.sh`
----------------
-
-Normally the zpool containing the root dataset cannot be exported on
-shutdown as it is still in use by the init process. To work around this,
-Dracut is able to restore the initramfs on shutdown and pivot to it.
-All remaining process are then running from a ramdisk, allowing for a
-clean unmount and export of the ZFS root. The theory of operation is
-described in detail in the [Dracut manual](https://www.kernel.org/pub/linux/utils/boot/dracut/dracut.html#_dracut_on_shutdown).
-
-This script will try to export all remaining zpools after Dracut has
-pivoted to the initramfs. If an initial regular export is not successful,
-Dracut will call this script once more with the `final` option,
-in which case a forceful export is attempted.
-
-Other Dracut modules include similar shutdown scripts and Dracut
-invokes these scripts round-robin until they succeed. In particular,
-the `90dm` module installs a script which tries to close and remove
-all device mapper targets. Thus, if there are ZVOLs containing
-dm-crypt volumes or if the zpool itself is backed by a dm-crypt
-volume, the shutdown scripts will try to untangle this.
+6. `zfs_force`, `zfs.force`, `zfsforce`: add `-f` to all `zpool import` invocations.
+   May be useful. Use with caution.

etc/default/Makefile.am

@@ -1,9 +1,10 @@
 include $(top_srcdir)/config/Substfiles.am
 include $(top_srcdir)/config/Shellcheck.am
 
-initconf_SCRIPTS = zfs
+initconf_DATA = zfs
 
-SUBSTFILES += $(initconf_SCRIPTS)
+SUBSTFILES += $(initconf_DATA)
 
+SHELLCHECKSCRIPTS = $(initconf_DATA)
 SHELLCHECK_SHELL = sh
 SHELLCHECK_IGNORE = ,SC2034

etc/systemd/system-generators/zfs-mount-generator.c

@@ -27,9 +27,6 @@
 #include <sys/types.h>
 #include <sys/time.h>
 #include <sys/stat.h>
-#include <sys/wait.h>
-#include <sys/mman.h>
-#include <semaphore.h>
 #include <stdbool.h>
 #include <unistd.h>
 #include <fcntl.h>
@@ -44,25 +41,16 @@
 #include <errno.h>
 #include <libzfs.h>
 
+/*
+ * For debugging only.
+ *
+ * Free statics with trivial life-times,
+ * but saved line filenames are replaced with a static string.
+ */
+#define	FREE_STATICS false
+
+#define	nitems(arr) (sizeof (arr) / sizeof (*arr))
 #define	STRCMP ((int(*)(const void *, const void *))&strcmp)
-#define	PID_T_CMP ((int(*)(const void *, const void *))&pid_t_cmp)
-
-static int
-pid_t_cmp(const pid_t *lhs, const pid_t *rhs)
-{
-	/*
-	 * This is always valid, quoth sys_types.h(7posix):
-	 * > blksize_t, pid_t, and ssize_t shall be signed integer types.
-	 */
-	return (*lhs - *rhs);
-}
-
-#define	EXIT_ENOMEM() \
-	do { \
-		fprintf(stderr, PROGNAME "[%d]: " \
-		    "not enough memory (L%d)!\n", getpid(), __LINE__); \
-		_exit(1); \
-	} while (0)
 
 
 #define	PROGNAME "zfs-mount-generator"
@@ -80,20 +68,11 @@ pid_t_cmp(const pid_t *lhs, const pid_t *rhs)
 #define	URI_REGEX_S "^\\([A-Za-z][A-Za-z0-9+.\\-]*\\):\\/\\/\\(.*\\)$"
 static regex_t uri_regex;
 
-static char *argv0;
-
 static const char *destdir = "/tmp";
 static int destdir_fd = -1;
 
 static void *known_pools = NULL; /* tsearch() of C strings */
-static struct {
-	sem_t noauto_not_on_sem;
-
-	sem_t noauto_names_sem;
-	size_t noauto_names_len;
-	size_t noauto_names_max;
-	char noauto_names[][NAME_MAX];
-} *noauto_files;
+static void *noauto_files = NULL; /* tsearch() of C strings */
 
 
 static char *
@@ -103,8 +82,12 @@ systemd_escape(const char *input, const char *prepend, const char *append)
 	size_t applen = strlen(append);
 	size_t prelen = strlen(prepend);
 	char *ret = malloc(4 * len + prelen + applen + 1);
-	if (!ret)
-		EXIT_ENOMEM();
+	if (!ret) {
+		fprintf(stderr, PROGNAME "[%d]: "
+		    "out of memory to escape \"%s%s%s\"!\n",
+		    getpid(), prepend, input, append);
+		return (NULL);
+	}
 
 	memcpy(ret, prepend, prelen);
 	char *out = ret + prelen;
@@ -166,8 +149,12 @@ systemd_escape_path(char *input, const char *prepend, const char *append)
 {
 	if (strcmp(input, "/") == 0) {
 		char *ret;
-		if (asprintf(&ret, "%s-%s", prepend, append) == -1)
-			EXIT_ENOMEM();
+		if (asprintf(&ret, "%s-%s", prepend, append) == -1) {
+			fprintf(stderr, PROGNAME "[%d]: "
+			    "out of memory to escape \"%s%s%s\"!\n",
+			    getpid(), prepend, input, append);
+			ret = NULL;
+		}
 		return (ret);
 	} else {
 		/*
@@ -209,6 +196,10 @@ fopenat(int dirfd, const char *pathname, int flags,
 static int
 line_worker(char *line, const char *cachefile)
 {
+	int ret = 0;
+	void *tofree_all[8];
+	void **tofree = tofree_all;
+
 	char *toktmp;
 	/* BEGIN CSTYLED */
 	const char *dataset                     = strtok_r(line, "\t", &toktmp);
@@ -240,11 +231,9 @@ line_worker(char *line, const char *cachefile)
 	if (p_nbmand == NULL) {
 		fprintf(stderr, PROGNAME "[%d]: %s: not enough tokens!\n",
 		    getpid(), dataset);
-		return (1);
+		goto err;
 	}
 
-	strncpy(argv0, dataset, strlen(argv0));
-
 	/* Minimal pre-requisites to mount a ZFS dataset */
 	const char *after = "zfs-import.target";
 	const char *wants = "zfs-import.target";
@@ -280,28 +269,31 @@ line_worker(char *line, const char *cachefile)
 
 
 	if (strcmp(p_encroot, "-") != 0) {
-		char *keyloadunit =
+		char *keyloadunit = *(tofree++) =
 		    systemd_escape(p_encroot, "zfs-load-key@", ".service");
+		if (keyloadunit == NULL)
+			goto err;
 
 		if (strcmp(dataset, p_encroot) == 0) {
 			const char *keymountdep = NULL;
 			bool is_prompt = false;
+			bool need_network = false;
 
 			regmatch_t uri_matches[3];
 			if (regexec(&uri_regex, p_keyloc,
-			    sizeof (uri_matches) / sizeof (*uri_matches),
-			    uri_matches, 0) == 0) {
+			    nitems(uri_matches), uri_matches, 0) == 0) {
+				p_keyloc[uri_matches[1].rm_eo] = '\0';
 				p_keyloc[uri_matches[2].rm_eo] = '\0';
+				const char *scheme =
+				    &p_keyloc[uri_matches[1].rm_so];
 				const char *path =
 				    &p_keyloc[uri_matches[2].rm_so];
 
-				/*
-				 * Assumes all URI keylocations need
-				 * the mount for their path;
-				 * http://, for example, wouldn't
-				 * (but it'd need network-online.target et al.)
-				 */
-				keymountdep = path;
+				if (strcmp(scheme, "https") == 0 ||
+				    strcmp(scheme, "http") == 0)
+					need_network = true;
+				else
+					keymountdep = path;
 			} else {
 				if (strcmp(p_keyloc, "prompt") != 0)
 					fprintf(stderr, PROGNAME "[%d]: %s: "
@@ -321,7 +313,7 @@ line_worker(char *line, const char *cachefile)
 				    "couldn't open %s under %s: %s\n",
 				    getpid(), dataset, keyloadunit, destdir,
 				    strerror(errno));
-				return (1);
+				goto err;
 			}
 
 			fprintf(keyloadunit_f,
@@ -335,20 +327,22 @@ line_worker(char *line, const char *cachefile)
 			    "After=%s\n",
 			    dataset, cachefile, wants, after);
 
+			if (need_network)
+				fprintf(keyloadunit_f,
+				    "Wants=network-online.target\n"
+				    "After=network-online.target\n");
+
 			if (p_systemd_requires)
 				fprintf(keyloadunit_f,
 				    "Requires=%s\n", p_systemd_requires);
 
-			if (p_systemd_requiresmountsfor || keymountdep) {
-				fprintf(keyloadunit_f, "RequiresMountsFor=");
-				if (p_systemd_requiresmountsfor)
-					fprintf(keyloadunit_f,
-					    "%s ", p_systemd_requiresmountsfor);
-				if (keymountdep)
-					fprintf(keyloadunit_f,
-					    "'%s'", keymountdep);
-				fprintf(keyloadunit_f, "\n");
-			}
+			if (p_systemd_requiresmountsfor)
+				fprintf(keyloadunit_f,
+				    "RequiresMountsFor=%s\n",
+				    p_systemd_requiresmountsfor);
+			if (keymountdep)
+				fprintf(keyloadunit_f,
+				    "RequiresMountsFor='%s'\n", keymountdep);
 
 			/* BEGIN CSTYLED */
 			fprintf(keyloadunit_f,
@@ -393,9 +387,13 @@ line_worker(char *line, const char *cachefile)
 		if (after[0] == '\0')
 			after = keyloadunit;
 		else if (asprintf(&toktmp, "%s %s", after, keyloadunit) != -1)
-			after = toktmp;
-		else
-			EXIT_ENOMEM();
+			after = *(tofree++) = toktmp;
+		else {
+			fprintf(stderr, PROGNAME "[%d]: %s: "
+			    "out of memory to generate after=\"%s %s\"!\n",
+			    getpid(), dataset, after, keyloadunit);
+			goto err;
+		}
 	}
 
 
@@ -404,12 +402,12 @@ line_worker(char *line, const char *cachefile)
 	    strcmp(p_systemd_ignore, "off") == 0) {
 		/* ok */
 	} else if (strcmp(p_systemd_ignore, "on") == 0)
-		return (0);
+		goto end;
 	else {
 		fprintf(stderr, PROGNAME "[%d]: %s: "
 		    "invalid org.openzfs.systemd:ignore=%s\n",
 		    getpid(), dataset, p_systemd_ignore);
-		return (1);
+		goto err;
 	}
 
 	/* Check for canmount */
@@ -418,21 +416,21 @@ line_worker(char *line, const char *cachefile)
 	} else if (strcmp(p_canmount, "noauto") == 0)
 		noauto = true;
 	else if (strcmp(p_canmount, "off") == 0)
-		return (0);
+		goto end;
 	else {
 		fprintf(stderr, PROGNAME "[%d]: %s: invalid canmount=%s\n",
 		    getpid(), dataset, p_canmount);
-		return (1);
+		goto err;
 	}
 
 	/* Check for legacy and blank mountpoints */
 	if (strcmp(p_mountpoint, "legacy") == 0 ||
 	    strcmp(p_mountpoint, "none") == 0)
-		return (0);
+		goto end;
 	else if (p_mountpoint[0] != '/') {
 		fprintf(stderr, PROGNAME "[%d]: %s: invalid mountpoint=%s\n",
 		    getpid(), dataset, p_mountpoint);
-		return (1);
+		goto err;
 	}
 
 	/* Escape the mountpoint per systemd policy */
@@ -442,7 +440,7 @@ line_worker(char *line, const char *cachefile)
 		fprintf(stderr,
 		    PROGNAME "[%d]: %s: abnormal simplified mountpoint: %s\n",
 		    getpid(), dataset, p_mountpoint);
-		return (1);
+		goto err;
 	}
 
 
@@ -552,8 +550,7 @@ line_worker(char *line, const char *cachefile)
 	 * 	files if we're sure they were created by us. (see 5.)
 	 * 2.	We handle files differently based on canmount.
 	 * 	Units with canmount=on always have precedence over noauto.
-	 * 	This is enforced by the noauto_not_on_sem semaphore,
-	 * 	which is only unlocked when the last canmount=on process exits.
+	 * 	This is enforced by processing these units before all others.
 	 * 	It is important to use p_canmount and not noauto here,
 	 * 	since we categorise by canmount while other properties,
 	 * 	e.g. org.openzfs.systemd:wanted-by, also modify noauto.
@@ -561,7 +558,7 @@ line_worker(char *line, const char *cachefile)
 	 * 	Additionally, we use noauto_files to track the unit file names
 	 * 	(which are the systemd-escaped mountpoints) of all (exclusively)
 	 * 	noauto datasets that had a file created.
-	 * 4.	If the file to be created is found in the tracking array,
+	 * 4.	If the file to be created is found in the tracking tree,
 	 * 	we do NOT create it.
 	 * 5.	If a file exists for a noauto dataset,
 	 * 	we check whether the file name is in the array.
@@ -571,29 +568,14 @@ line_worker(char *line, const char *cachefile)
 	 * 	further noauto datasets creating a file for this path again.
 	 */
 
-	{
-		sem_t *our_sem = (strcmp(p_canmount, "on") == 0) ?
-		    &noauto_files->noauto_names_sem :
-		    &noauto_files->noauto_not_on_sem;
-		while (sem_wait(our_sem) == -1 && errno == EINTR)
-			;
-	}
-
 	struct stat stbuf;
 	bool already_exists = fstatat(destdir_fd, mountfile, &stbuf, 0) == 0;
+	bool is_known = tfind(mountfile, &noauto_files, STRCMP) != NULL;
 
-	bool is_known = false;
-	for (size_t i = 0; i < noauto_files->noauto_names_len; ++i) {
-		if (strncmp(
-		    noauto_files->noauto_names[i], mountfile, NAME_MAX) == 0) {
-			is_known = true;
-			break;
-		}
-	}
-
+	*(tofree++) = (void *)mountfile;
 	if (already_exists) {
 		if (is_known) {
-			/* If it's in $noauto_files, we must be noauto too */
+			/* If it's in noauto_files, we must be noauto too */
 
 			/* See 5 */
 			errno = 0;
@@ -614,43 +596,31 @@ line_worker(char *line, const char *cachefile)
 		}
 
 		/* File exists: skip current dataset */
-		if (strcmp(p_canmount, "on") == 0)
-			sem_post(&noauto_files->noauto_names_sem);
-		return (0);
+		goto end;
 	} else {
 		if (is_known) {
 			/* See 4 */
-			if (strcmp(p_canmount, "on") == 0)
-				sem_post(&noauto_files->noauto_names_sem);
-			return (0);
+			goto end;
 		} else if (strcmp(p_canmount, "noauto") == 0) {
-			if (noauto_files->noauto_names_len ==
-			    noauto_files->noauto_names_max)
+			if (tsearch(mountfile, &noauto_files, STRCMP) == NULL)
 				fprintf(stderr, PROGNAME "[%d]: %s: "
-				    "noauto dataset limit (%zu) reached! "
-				    "Not tracking %s. Please report this to "
-				    "https://github.com/openzfs/zfs\n",
-				    getpid(), dataset,
-				    noauto_files->noauto_names_max, mountfile);
-			else {
-				strncpy(noauto_files->noauto_names[
-				    noauto_files->noauto_names_len],
-				    mountfile, NAME_MAX);
-				++noauto_files->noauto_names_len;
-			}
+				    "out of memory for noauto datasets! "
+				    "Not tracking %s.\n",
+				    getpid(), dataset, mountfile);
+			else
+				/* mountfile escaped to noauto_files */
+				*(--tofree) = NULL;
 		}
 	}
 
 
 	FILE *mountfile_f = fopenat(destdir_fd, mountfile,
 	    O_WRONLY | O_CREAT | O_TRUNC | O_CLOEXEC, "w", 0644);
-	if (strcmp(p_canmount, "on") == 0)
-		sem_post(&noauto_files->noauto_names_sem);
 	if (!mountfile_f) {
 		fprintf(stderr,
 		    PROGNAME "[%d]: %s: couldn't open %s under %s: %s\n",
 		    getpid(), dataset, mountfile, destdir, strerror(errno));
-		return (1);
+		goto err;
 	}
 
 	fprintf(mountfile_f,
@@ -699,12 +669,17 @@ line_worker(char *line, const char *cachefile)
 	(void) fclose(mountfile_f);
 
 	if (!requiredby && !wantedby)
-		return (0);
+		goto end;
 
 	/* Finally, create the appropriate dependencies */
 	char *linktgt;
-	if (asprintf(&linktgt, "../%s", mountfile) == -1)
-		EXIT_ENOMEM();
+	if (asprintf(&linktgt, "../%s", mountfile) == -1) {
+		fprintf(stderr, PROGNAME "[%d]: %s: "
+		    "out of memory for dependents of %s!\n",
+		    getpid(), dataset, mountfile);
+		goto err;
+	}
+	*(tofree++) = linktgt;
 
 	char *dependencies[][2] = {
 		{"wants", wantedby},
@@ -719,8 +694,14 @@ line_worker(char *line, const char *cachefile)
 		    reqby;
 		    reqby = strtok_r(NULL, " ", &toktmp)) {
 			char *depdir;
-			if (asprintf(&depdir, "%s.%s", reqby, (*dep)[0]) == -1)
-				EXIT_ENOMEM();
+			if (asprintf(
+			    &depdir, "%s.%s", reqby, (*dep)[0]) == -1) {
+				fprintf(stderr, PROGNAME "[%d]: %s: "
+				    "out of memory for dependent dir name "
+				    "\"%s.%s\"!\n",
+				    getpid(), dataset, reqby, (*dep)[0]);
+				continue;
+			}
 
 			(void) mkdirat(destdir_fd, depdir, 0755);
 			int depdir_fd = openat(destdir_fd, depdir,
@@ -746,7 +727,24 @@ line_worker(char *line, const char *cachefile)
 		}
 	}
 
-	return (0);
+end:
+	if (tofree >= tofree_all + nitems(tofree_all)) {
+		/*
+		 * This won't happen as-is:
+		 * we've got 8 slots and allocate 4 things at most.
+		 */
+		fprintf(stderr,
+		    PROGNAME "[%d]: %s: need to free %zu > %zu!\n",
+		    getpid(), dataset, tofree - tofree_all, nitems(tofree_all));
+		ret = tofree - tofree_all;
+	}
+
+	while (tofree-- != tofree_all)
+		free(*tofree);
+	return (ret);
+err:
+	ret = 1;
+	goto end;
 }
 
 
@@ -780,12 +778,11 @@ main(int argc, char **argv)
 		if (kmfd >= 0) {
 			(void) dup2(kmfd, STDERR_FILENO);
 			(void) close(kmfd);
+
+			setlinebuf(stderr);
 		}
 	}
 
-	uint8_t debug = 0;
-
-	argv0 = argv[0];
 	switch (argc) {
 	case 1:
 		/* Use default */
@@ -844,33 +841,9 @@ main(int argc, char **argv)
 		}
 	}
 
-	{
-		/*
-		 * We could just get a gigabyte here and Not Care,
-		 * but if vm.overcommit_memory=2, then MAP_NORESERVE is ignored
-		 * and we'd try (and likely fail) to rip it out of swap
-		 */
-		noauto_files = mmap(NULL, 4 * 1024 * 1024,
-		    PROT_READ | PROT_WRITE,
-		    MAP_SHARED | MAP_ANONYMOUS | MAP_NORESERVE, -1, 0);
-		if (noauto_files == MAP_FAILED) {
-			fprintf(stderr,
-			    PROGNAME "[%d]: couldn't allocate IPC region: %s\n",
-			    getpid(), strerror(errno));
-			_exit(1);
-		}
-
-		sem_init(&noauto_files->noauto_not_on_sem, true, 0);
-		sem_init(&noauto_files->noauto_names_sem, true, 1);
-		noauto_files->noauto_names_len = 0;
-		/* Works out to 16447ish, *well* enough */
-		noauto_files->noauto_names_max =
-		    (4 * 1024 * 1024 - sizeof (*noauto_files)) / NAME_MAX;
-	}
-
+	bool debug = false;
 	char *line = NULL;
 	size_t linelen = 0;
-	struct timespec time_start = {};
 	{
 		const char *dbgenv = getenv("ZFS_DEBUG");
 		if (dbgenv)
@@ -879,7 +852,7 @@ main(int argc, char **argv)
 			FILE *cmdline = fopen("/proc/cmdline", "re");
 			if (cmdline != NULL) {
 				if (getline(&line, &linelen, cmdline) >= 0)
-					debug = strstr(line, "debug") ? 2 : 0;
+					debug = strstr(line, "debug");
 				(void) fclose(cmdline);
 			}
 		}
@@ -888,19 +861,17 @@ main(int argc, char **argv)
 			dup2(STDERR_FILENO, STDOUT_FILENO);
 	}
 
-	size_t forked_canmount_on = 0;
-	size_t forked_canmount_not_on = 0;
-	size_t canmount_on_pids_len = 128;
-	pid_t *canmount_on_pids =
-	    malloc(canmount_on_pids_len * sizeof (*canmount_on_pids));
-	if (canmount_on_pids == NULL)
-		canmount_on_pids_len = 0;
-
+	struct timespec time_start = {};
 	if (debug)
 		clock_gettime(CLOCK_MONOTONIC_RAW, &time_start);
 
-	ssize_t read;
-	pid_t pid;
+	struct line {
+		char *line;
+		const char *fname;
+		struct line *next;
+	} *lines_canmount_not_on = NULL;
+
+	int ret = 0;
 	struct dirent *cachent;
 	while ((cachent = readdir(fslist_dir)) != NULL) {
 		if (strcmp(cachent->d_name, ".") == 0 ||
@@ -916,129 +887,67 @@ main(int argc, char **argv)
 			continue;
 		}
 
+		const char *filename = FREE_STATICS ? "(elided)" : NULL;
+
+		ssize_t read;
 		while ((read = getline(&line, &linelen, cachefile)) >= 0) {
 			line[read - 1] = '\0'; /* newline */
 
-			switch (pid = fork()) {
-			case -1:
-				fprintf(stderr,
-				    PROGNAME "[%d]: couldn't fork for %s: %s\n",
-				    getpid(), line, strerror(errno));
-				break;
-			case 0: /* child */
-				_exit(line_worker(line, cachent->d_name));
-			default: { /* parent */
-				char *tmp;
-				char *dset = strtok_r(line, "\t", &tmp);
-				strtok_r(NULL, "\t", &tmp);
-				char *canmount = strtok_r(NULL, "\t", &tmp);
-				bool canmount_on =
-				    canmount && strncmp(canmount, "on", 2) == 0;
+			char *canmount = line;
+			canmount += strcspn(canmount, "\t");
+			canmount += strspn(canmount, "\t");
+			canmount += strcspn(canmount, "\t");
+			canmount += strspn(canmount, "\t");
+			bool canmount_on = strncmp(canmount, "on", 2) == 0;
 
-				if (debug >= 2)
-					printf(PROGNAME ": forked %d, "
-					    "canmount_on=%d, dataset=%s\n",
-					    (int)pid, canmount_on, dset);
+			if (canmount_on)
+				ret |= line_worker(line, cachent->d_name);
+			else {
+				if (filename == NULL)
+					filename =
+					    strdup(cachent->d_name) ?: "(?)";
 
-				if (canmount_on &&
-				    forked_canmount_on ==
-				    canmount_on_pids_len) {
-					size_t new_len =
-					    (canmount_on_pids_len ?: 16) * 2;
-					void *new_pidlist =
-					    realloc(canmount_on_pids,
-					    new_len *
-					    sizeof (*canmount_on_pids));
-					if (!new_pidlist) {
-						fprintf(stderr,
-						    PROGNAME "[%d]: "
-						    "out of memory! "
-						    "Mount ordering may be "
-						    "affected.\n", getpid());
-						continue;
-					}
-
-					canmount_on_pids = new_pidlist;
-					canmount_on_pids_len = new_len;
+				struct line *l = calloc(1, sizeof (*l));
+				char *nl = strdup(line);
+				if (l == NULL || nl == NULL) {
+					fprintf(stderr, PROGNAME "[%d]: "
+					    "out of memory for \"%s\" in %s\n",
+					    getpid(), line, cachent->d_name);
+					free(l);
+					free(nl);
+					continue;
 				}
-
-				if (canmount_on) {
-					canmount_on_pids[forked_canmount_on] =
-					    pid;
-					++forked_canmount_on;
-				} else
-					++forked_canmount_not_on;
-				break;
-			}
+				l->line = nl;
+				l->fname = filename;
+				l->next = lines_canmount_not_on;
+				lines_canmount_not_on = l;
 			}
 		}
 
-		(void) fclose(cachefile);
+		fclose(cachefile);
 	}
 	free(line);
 
-	if (forked_canmount_on == 0) {
-		/* No canmount=on processes to finish, so don't deadlock here */
-		for (size_t i = 0; i < forked_canmount_not_on; ++i)
-			sem_post(&noauto_files->noauto_not_on_sem);
-	} else {
-		/* Likely a no-op, since we got these from a narrow fork loop */
-		qsort(canmount_on_pids, forked_canmount_on,
-		    sizeof (*canmount_on_pids), PID_T_CMP);
-	}
+	while (lines_canmount_not_on) {
+		struct line *l = lines_canmount_not_on;
+		lines_canmount_not_on = l->next;
 
-	int status, ret = 0;
-	struct rusage usage;
-	size_t forked_canmount_on_max = forked_canmount_on;
-	while ((pid = wait4(-1, &status, 0, &usage)) != -1) {
-		ret |= WEXITSTATUS(status) | WTERMSIG(status);
-
-		if (forked_canmount_on != 0) {
-			if (bsearch(&pid, canmount_on_pids,
-			    forked_canmount_on_max, sizeof (*canmount_on_pids),
-			    PID_T_CMP))
-				--forked_canmount_on;
-
-			if (forked_canmount_on == 0) {
-				/*
-				 * All canmount=on processes have finished,
-				 * let all the lower-priority ones finish now
-				 */
-				for (size_t i = 0;
-				    i < forked_canmount_not_on; ++i)
-					sem_post(
-					    &noauto_files->noauto_not_on_sem);
-			}
+		ret |= line_worker(l->line, l->fname);
+		if (FREE_STATICS) {
+			free(l->line);
+			free(l);
 		}
-
-		if (debug >= 2)
-			printf(PROGNAME ": %d done, user=%llu.%06us, "
-			    "system=%llu.%06us, maxrss=%ldB, ex=0x%x\n",
-			    (int)pid,
-			    (unsigned long long) usage.ru_utime.tv_sec,
-			    (unsigned int) usage.ru_utime.tv_usec,
-			    (unsigned long long) usage.ru_stime.tv_sec,
-			    (unsigned int) usage.ru_stime.tv_usec,
-			    usage.ru_maxrss * 1024, status);
 	}
 
 	if (debug) {
 		struct timespec time_end = {};
 		clock_gettime(CLOCK_MONOTONIC_RAW, &time_end);
 
+		struct rusage usage;
 		getrusage(RUSAGE_SELF, &usage);
 		printf(
 		    "\n"
-		    PROGNAME ": self    : "
-		    "user=%llu.%06us, system=%llu.%06us, maxrss=%ldB\n",
-		    (unsigned long long) usage.ru_utime.tv_sec,
-		    (unsigned int) usage.ru_utime.tv_usec,
-		    (unsigned long long) usage.ru_stime.tv_sec,
-		    (unsigned int) usage.ru_stime.tv_usec,
-		    usage.ru_maxrss * 1024);
-
-		getrusage(RUSAGE_CHILDREN, &usage);
-		printf(PROGNAME ": children: "
+		    PROGNAME ": "
 		    "user=%llu.%06us, system=%llu.%06us, maxrss=%ldB\n",
 		    (unsigned long long) usage.ru_utime.tv_sec,
 		    (unsigned int) usage.ru_utime.tv_usec,
@@ -1068,7 +977,7 @@ main(int argc, char **argv)
 		    time_init.tv_nsec / 1000000000;
 		time_init.tv_nsec %= 1000000000;
 
-		printf(PROGNAME ": wall    : "
+		printf(PROGNAME ": "
 		    "total=%llu.%09llus = "
 		    "init=%llu.%09llus + real=%llu.%09llus\n",
 		    (unsigned long long) time_init.tv_sec,
@@ -1077,7 +986,15 @@ main(int argc, char **argv)
 		    (unsigned long long) time_start.tv_nsec,
 		    (unsigned long long) time_end.tv_sec,
 		    (unsigned long long) time_end.tv_nsec);
+
+		fflush(stdout);
 	}
 
+	if (FREE_STATICS) {
+		closedir(fslist_dir);
+		tdestroy(noauto_files, free);
+		tdestroy(known_pools, free);
+		regfree(&uri_regex);
+	}
 	_exit(ret);
 }

etc/zfs/Makefile.am

@@ -10,9 +10,10 @@ dist_pkgsysconf_DATA = \
 	vdev_id.conf.multipath.example \
 	vdev_id.conf.scsi.example
 
-pkgsysconf_SCRIPTS = \
+pkgsysconf_DATA = \
 	zfs-functions
 
-SUBSTFILES += $(pkgsysconf_SCRIPTS)
+SUBSTFILES += $(pkgsysconf_DATA)
 
+SHELLCHECKSCRIPTS = $(pkgsysconf_DATA)
 SHELLCHECK_SHELL = dash # local variables

include/libzfs.h

@@ -795,9 +795,10 @@ extern int zfs_receive(libzfs_handle_t *, const char *, nvlist_t *,
     recvflags_t *, int, avl_tree_t *);
 
 typedef enum diff_flags {
-	ZFS_DIFF_PARSEABLE = 0x1,
-	ZFS_DIFF_TIMESTAMP = 0x2,
-	ZFS_DIFF_CLASSIFY = 0x4
+	ZFS_DIFF_PARSEABLE = 1 << 0,
+	ZFS_DIFF_TIMESTAMP = 1 << 1,
+	ZFS_DIFF_CLASSIFY = 1 << 2,
+	ZFS_DIFF_NO_MANGLE = 1 << 3
 } diff_flags_t;
 
 extern int zfs_show_diffs(zfs_handle_t *, int, const char *, const char *,

include/libzfs_impl.h

@@ -234,6 +234,7 @@ typedef struct differ_info {
 	boolean_t scripted;
 	boolean_t classify;
 	boolean_t timestamped;
+	boolean_t no_mangle;
 	uint64_t shares;
 	int zerr;
 	int cleanupfd;

include/os/freebsd/zfs/sys/zfs_znode_impl.h

@@ -123,25 +123,29 @@ extern minor_t zfsdev_minor_alloc(void);
 #define	zn_rlimit_fsize(zp, uio) \
     vn_rlimit_fsize(ZTOV(zp), GET_UIO_STRUCT(uio), zfs_uio_td(uio))
 
+#define	ZFS_ENTER_ERROR(zfsvfs, error) do {			\
+	ZFS_TEARDOWN_ENTER_READ((zfsvfs), FTAG);		\
+	if (__predict_false((zfsvfs)->z_unmounted)) {		\
+		ZFS_TEARDOWN_EXIT_READ(zfsvfs, FTAG);		\
+		return (error);					\
+	}							\
+} while (0)
+
 /* Called on entry to each ZFS vnode and vfs operation  */
-#define	ZFS_ENTER(zfsvfs) \
-	{ \
-		ZFS_TEARDOWN_ENTER_READ((zfsvfs), FTAG); \
-		if (__predict_false((zfsvfs)->z_unmounted)) { \
-			ZFS_TEARDOWN_EXIT_READ(zfsvfs, FTAG); \
-			return (EIO); \
-		} \
-	}
+#define	ZFS_ENTER(zfsvfs)	ZFS_ENTER_ERROR(zfsvfs, EIO)
 
 /* Must be called before exiting the vop */
-#define	ZFS_EXIT(zfsvfs) ZFS_TEARDOWN_EXIT_READ(zfsvfs, FTAG)
+#define	ZFS_EXIT(zfsvfs)	ZFS_TEARDOWN_EXIT_READ(zfsvfs, FTAG)
+
+#define	ZFS_VERIFY_ZP_ERROR(zp, error) do {			\
+	if (__predict_false((zp)->z_sa_hdl == NULL)) {		\
+		ZFS_EXIT((zp)->z_zfsvfs);			\
+		return (error);					\
+	}							\
+} while (0)
 
 /* Verifies the znode is valid */
-#define	ZFS_VERIFY_ZP(zp) \
-	if (__predict_false((zp)->z_sa_hdl == NULL)) { \
-		ZFS_EXIT((zp)->z_zfsvfs); \
-		return (EIO); \
-	} \
+#define	ZFS_VERIFY_ZP(zp)	ZFS_VERIFY_ZP_ERROR(zp, EIO)
 
 /*
  * Macros for dealing with dmu_buf_hold

include/os/linux/kernel/linux/blkdev_compat.h

@@ -495,21 +495,45 @@ blk_queue_discard_granularity(struct request_queue *q, unsigned int dg)
 }
 
 /*
+ * 5.19 API,
+ *   bdev_max_discard_sectors()
+ *
+ * 2.6.32 API,
+ *   blk_queue_discard()
+ */
+static inline boolean_t
+bdev_discard_supported(struct block_device *bdev)
+{
+#if defined(HAVE_BDEV_MAX_DISCARD_SECTORS)
+	return (!!bdev_max_discard_sectors(bdev));
+#elif defined(HAVE_BLK_QUEUE_DISCARD)
+	return (!!blk_queue_discard(bdev_get_queue(bdev)));
+#else
+#error "Unsupported kernel"
+#endif
+}
+
+/*
+ * 5.19 API,
+ *   bdev_max_secure_erase_sectors()
+ *
  * 4.8 API,
  *   blk_queue_secure_erase()
  *
  * 2.6.36 - 4.7 API,
  *   blk_queue_secdiscard()
  */
-static inline int
-blk_queue_discard_secure(struct request_queue *q)
+static inline boolean_t
+bdev_secure_discard_supported(struct block_device *bdev)
 {
-#if defined(HAVE_BLK_QUEUE_SECURE_ERASE)
-	return (blk_queue_secure_erase(q));
+#if defined(HAVE_BDEV_MAX_SECURE_ERASE_SECTORS)
+	return (!!bdev_max_secure_erase_sectors(bdev));
+#elif defined(HAVE_BLK_QUEUE_SECURE_ERASE)
+	return (!!blk_queue_secure_erase(bdev_get_queue(bdev)));
 #elif defined(HAVE_BLK_QUEUE_SECDISCARD)
-	return (blk_queue_secdiscard(q));
+	return (!!blk_queue_secdiscard(bdev_get_queue(bdev)));
 #else
-	return (0);
+#error "Unsupported kernel"
 #endif
 }
 
@@ -527,7 +551,10 @@ blk_generic_start_io_acct(struct request_queue *q __attribute__((unused)),
     struct gendisk *disk __attribute__((unused)),
     int rw __attribute__((unused)), struct bio *bio)
 {
-#if defined(HAVE_DISK_IO_ACCT)
+#if defined(HAVE_BDEV_IO_ACCT)
+	return (bdev_start_io_acct(bio->bi_bdev, bio_sectors(bio),
+	    bio_op(bio), jiffies));
+#elif defined(HAVE_DISK_IO_ACCT)
 	return (disk_start_io_acct(disk, bio_sectors(bio), bio_op(bio)));
 #elif defined(HAVE_BIO_IO_ACCT)
 	return (bio_start_io_acct(bio));
@@ -550,7 +577,9 @@ blk_generic_end_io_acct(struct request_queue *q __attribute__((unused)),
     struct gendisk *disk __attribute__((unused)),
     int rw __attribute__((unused)), struct bio *bio, unsigned long start_time)
 {
-#if defined(HAVE_DISK_IO_ACCT)
+#if defined(HAVE_BDEV_IO_ACCT)
+	bdev_end_io_acct(bio->bi_bdev, bio_op(bio), start_time);
+#elif defined(HAVE_DISK_IO_ACCT)
 	disk_end_io_acct(disk, bio_op(bio), start_time);
 #elif defined(HAVE_BIO_IO_ACCT)
 	bio_end_io_acct(bio, start_time);

include/os/linux/kernel/linux/simd_x86.h

@@ -87,7 +87,9 @@
 
 #if defined(HAVE_KERNEL_FPU_API_HEADER)
 #include <asm/fpu/api.h>
+#if defined(HAVE_KERNEL_FPU_INTERNAL_HEADER)
 #include <asm/fpu/internal.h>
+#endif
 #if defined(HAVE_KERNEL_FPU_XCR_HEADER)
 #include <asm/fpu/xcr.h>
 #endif

include/sys/dbuf.h

@@ -321,13 +321,12 @@ typedef struct dmu_buf_impl {
 	uint8_t db_dirtycnt;
 } dmu_buf_impl_t;
 
-/* Note: the dbuf hash table is exposed only for the mdb module */
-#define	DBUF_MUTEXES 2048
-#define	DBUF_HASH_MUTEX(h, idx) (&(h)->hash_mutexes[(idx) & (DBUF_MUTEXES-1)])
+#define	DBUF_RWLOCKS 8192
+#define	DBUF_HASH_RWLOCK(h, idx) (&(h)->hash_rwlocks[(idx) & (DBUF_RWLOCKS-1)])
 typedef struct dbuf_hash_table {
 	uint64_t hash_table_mask;
 	dmu_buf_impl_t **hash_table;
-	kmutex_t hash_mutexes[DBUF_MUTEXES] ____cacheline_aligned;
+	krwlock_t hash_rwlocks[DBUF_RWLOCKS] ____cacheline_aligned;
 } dbuf_hash_table_t;
 
 typedef void (*dbuf_prefetch_fn)(void *, boolean_t);

include/sys/dnode.h

@@ -616,7 +616,7 @@ _NOTE(CONSTCOND) } while (0)
 #else
 
 #define	dprintf_dnode(db, fmt, ...)
-#define	DNODE_VERIFY(dn)
+#define	DNODE_VERIFY(dn)		((void) sizeof ((uintptr_t)(dn)))
 #define	FREE_VERIFY(db, start, end, tx)
 
 #endif

include/sys/spa_impl.h

@@ -370,6 +370,7 @@ struct spa {
 	boolean_t	spa_is_root;		/* pool is root */
 	int		spa_minref;		/* num refs when first opened */
 	spa_mode_t	spa_mode;		/* SPA_MODE_{READ|WRITE} */
+	boolean_t	spa_read_spacemaps;	/* spacemaps available if ro */
 	spa_log_state_t spa_log_state;		/* log state */
 	uint64_t	spa_autoexpand;		/* lun expansion on/off */
 	ddt_t		*spa_ddt[ZIO_CHECKSUM_FUNCTIONS]; /* in-core DDTs */

include/sys/zap.h

@@ -110,7 +110,12 @@ typedef enum zap_flags {
 	 * already randomly distributed.
 	 */
 	ZAP_FLAG_PRE_HASHED_KEY = 1 << 2,
+#if defined(__linux__) && defined(_KERNEL)
+} zfs_zap_flags_t;
+#define	zap_flags_t	zfs_zap_flags_t
+#else
 } zap_flags_t;
+#endif
 
 /*
  * Create a new zapobj with no attributes and return its object number.

lib/libspl/os/freebsd/mnttab.c

@@ -39,6 +39,7 @@ __FBSDID("$FreeBSD$");
 
 #include <ctype.h>
 #include <errno.h>
+#include <pthread.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
@@ -136,6 +137,7 @@ statfs2mnttab(struct statfs *sfs, struct mnttab *mp)
 	mp->mnt_mntopts = gmntopts;
 }
 
+static pthread_rwlock_t gsfs_lock = PTHREAD_RWLOCK_INITIALIZER;
 static struct statfs *gsfs = NULL;
 static int allfs = 0;
 
@@ -145,6 +147,8 @@ statfs_init(void)
 	struct statfs *sfs;
 	int error;
 
+	(void) pthread_rwlock_wrlock(&gsfs_lock);
+
 	if (gsfs != NULL) {
 		free(gsfs);
 		gsfs = NULL;
@@ -162,6 +166,7 @@ statfs_init(void)
 	sfs = realloc(gsfs, allfs * sizeof (gsfs[0]));
 	if (sfs != NULL)
 		gsfs = sfs;
+	(void) pthread_rwlock_unlock(&gsfs_lock);
 	return (0);
 fail:
 	error = errno;
@@ -169,6 +174,7 @@ fail:
 		free(gsfs);
 	gsfs = NULL;
 	allfs = 0;
+	(void) pthread_rwlock_unlock(&gsfs_lock);
 	return (error);
 }
 
@@ -181,6 +187,8 @@ getmntany(FILE *fd __unused, struct mnttab *mgetp, struct mnttab *mrefp)
 	if (error != 0)
 		return (error);
 
+	(void) pthread_rwlock_rdlock(&gsfs_lock);
+
 	for (i = 0; i < allfs; i++) {
 		if (mrefp->mnt_special != NULL &&
 		    strcmp(mrefp->mnt_special, gsfs[i].f_mntfromname) != 0) {
@@ -195,8 +203,10 @@ getmntany(FILE *fd __unused, struct mnttab *mgetp, struct mnttab *mrefp)
 			continue;
 		}
 		statfs2mnttab(&gsfs[i], mgetp);
+		(void) pthread_rwlock_unlock(&gsfs_lock);
 		return (0);
 	}
+	(void) pthread_rwlock_unlock(&gsfs_lock);
 	return (-1);
 }
 
@@ -214,9 +224,13 @@ getmntent(FILE *fp, struct mnttab *mp)
 		if (error != 0)
 			return (error);
 	}
-	if (nfs >= allfs)
+	(void) pthread_rwlock_rdlock(&gsfs_lock);
+	if (nfs >= allfs) {
+		(void) pthread_rwlock_unlock(&gsfs_lock);
 		return (-1);
+	}
 	statfs2mnttab(&gsfs[nfs], mp);
+	(void) pthread_rwlock_unlock(&gsfs_lock);
 	if (lseek(fileno(fp), 1, SEEK_CUR) == -1)
 		return (errno);
 	return (0);

lib/libzfs/libzfs_dataset.c

@@ -449,14 +449,19 @@ make_dataset_handle_common(zfs_handle_t *zhp, zfs_cmd_t *zc)
 	 * We've managed to open the dataset and gather statistics.  Determine
 	 * the high-level type.
 	 */
-	if (zhp->zfs_dmustats.dds_type == DMU_OST_ZVOL)
+	if (zhp->zfs_dmustats.dds_type == DMU_OST_ZVOL) {
 		zhp->zfs_head_type = ZFS_TYPE_VOLUME;
-	else if (zhp->zfs_dmustats.dds_type == DMU_OST_ZFS)
+	} else if (zhp->zfs_dmustats.dds_type == DMU_OST_ZFS) {
 		zhp->zfs_head_type = ZFS_TYPE_FILESYSTEM;
-	else if (zhp->zfs_dmustats.dds_type == DMU_OST_OTHER)
+	} else if (zhp->zfs_dmustats.dds_type == DMU_OST_OTHER) {
+		errno = EINVAL;
 		return (-1);
-	else
+	} else if (zhp->zfs_dmustats.dds_inconsistent) {
+		errno = EBUSY;
+		return (-1);
+	} else {
 		abort();
+	}
 
 	if (zhp->zfs_dmustats.dds_is_snapshot)
 		zhp->zfs_type = ZFS_TYPE_SNAPSHOT;

lib/libzfs/libzfs_diff.c

@@ -122,9 +122,9 @@ stream_bytes(FILE *fp, const char *string)
 
 	while ((c = *string++) != '\0') {
 		if (c > ' ' && c != '\\' && c < '\177') {
-			(void) fprintf(fp, "%c", c);
+			(void) fputc(c, fp);
 		} else {
-			(void) fprintf(fp, "\\%04o", (uint8_t)c);
+			(void) fprintf(fp, "\\%04hho", (uint8_t)c);
 		}
 	}
 }
@@ -176,8 +176,13 @@ print_what(FILE *fp, mode_t what)
 static void
 print_cmn(FILE *fp, differ_info_t *di, const char *file)
 {
-	stream_bytes(fp, di->dsmnt);
-	stream_bytes(fp, file);
+	if (!di->no_mangle) {
+		stream_bytes(fp, di->dsmnt);
+		stream_bytes(fp, file);
+	} else {
+		(void) fputs(di->dsmnt, fp);
+		(void) fputs(file, fp);
+	}
 }
 
 static void
@@ -752,6 +757,7 @@ zfs_show_diffs(zfs_handle_t *zhp, int outfd, const char *fromsnap,
 	di.scripted = (flags & ZFS_DIFF_PARSEABLE);
 	di.classify = (flags & ZFS_DIFF_CLASSIFY);
 	di.timestamped = (flags & ZFS_DIFF_TIMESTAMP);
+	di.no_mangle = (flags & ZFS_DIFF_NO_MANGLE);
 
 	di.outputfd = outfd;
 	di.datafd = pipefd[0];

lib/libzfs/libzfs_import.c

@@ -146,10 +146,10 @@ zpool_clear_label(int fd)
 	struct stat64 statbuf;
 	int l;
 	vdev_label_t *label;
-	l2arc_dev_hdr_phys_t *l2dhdr;
+	l2arc_dev_hdr_phys_t *l2dhdr = NULL;
 	uint64_t size;
-	int labels_cleared = 0, header_cleared = 0;
-	boolean_t clear_l2arc_header = B_FALSE;
+	int labels_cleared = 0;
+	boolean_t clear_l2arc_header = B_FALSE, header_cleared = B_FALSE;
 
 	if (fstat64_blk(fd, &statbuf) == -1)
 		return (0);
@@ -219,13 +219,10 @@ zpool_clear_label(int fd)
 	}
 
 	/* Clear the L2ARC header. */
-	if (clear_l2arc_header) {
-		memset(l2dhdr, 0, sizeof (l2arc_dev_hdr_phys_t));
-		if (pwrite64(fd, l2dhdr, sizeof (l2arc_dev_hdr_phys_t),
-		    VDEV_LABEL_START_SIZE) == sizeof (l2arc_dev_hdr_phys_t)) {
-			header_cleared++;
-		}
-	}
+	if (clear_l2arc_header &&
+	    pwrite64(fd, l2dhdr, sizeof (l2arc_dev_hdr_phys_t),
+	    VDEV_LABEL_START_SIZE) == sizeof (l2arc_dev_hdr_phys_t))
+			header_cleared = B_TRUE;
 
 	free(label);
 	free(l2dhdr);
@@ -233,6 +230,9 @@ zpool_clear_label(int fd)
 	if (labels_cleared == 0)
 		return (-1);
 
+	if (clear_l2arc_header && !header_cleared)
+		return (-1);
+
 	return (0);
 }
 

lib/libzfs/libzfs_mount.c

@@ -1348,7 +1348,7 @@ zfs_mount_task(void *arg)
 	    sizeof (mountpoint), NULL, NULL, 0, B_FALSE) == 0);
 
 	if (mp->mnt_func(handles[idx], mp->mnt_data) != 0)
-		return;
+		goto out;
 
 	/*
 	 * We dispatch tasks to mount filesystems with mountpoints underneath
@@ -1369,6 +1369,8 @@ zfs_mount_task(void *arg)
 		zfs_dispatch_mount(mp->mnt_hdl, handles, num_handles, i,
 		    mp->mnt_func, mp->mnt_data, mp->mnt_tp);
 	}
+
+out:
 	free(mp);
 }
 

lib/libzfs/libzfs_sendrecv.c

@@ -2133,7 +2133,6 @@ zfs_send(zfs_handle_t *zhp, const char *fromsnap, const char *tosnap,
 	avl_tree_t *fsavl = NULL;
 	static uint64_t holdseq;
 	int spa_version;
-	int featureflags = 0;
 	FILE *fout;
 
 	(void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN,
@@ -2145,17 +2144,23 @@ zfs_send(zfs_handle_t *zhp, const char *fromsnap, const char *tosnap,
 		return (zfs_error(zhp->zfs_hdl, EZFS_NOENT, errbuf));
 	}
 
-	if (zhp->zfs_type == ZFS_TYPE_FILESYSTEM) {
-		uint64_t version;
-		version = zfs_prop_get_int(zhp, ZFS_PROP_VERSION);
-		if (version >= ZPL_VERSION_SA) {
-			featureflags |= DMU_BACKUP_FEATURE_SA_SPILL;
+	if (fromsnap) {
+		char full_fromsnap_name[ZFS_MAX_DATASET_NAME_LEN];
+		if (snprintf(full_fromsnap_name, sizeof (full_fromsnap_name),
+		    "%s@%s", zhp->zfs_name, fromsnap) >=
+		    sizeof (full_fromsnap_name)) {
+			err = EINVAL;
+			goto stderr_out;
 		}
+		zfs_handle_t *fromsnapn = zfs_open(zhp->zfs_hdl,
+		    full_fromsnap_name, ZFS_TYPE_SNAPSHOT);
+		if (fromsnapn == NULL) {
+			err = -1;
+			goto err_out;
+		}
+		zfs_close(fromsnapn);
 	}
 
-	if (flags->holds)
-		featureflags |= DMU_BACKUP_FEATURE_HOLDS;
-
 	if (flags->replicate || flags->doall || flags->props ||
 	    flags->holds || flags->backup) {
 		char full_tosnap_name[ZFS_MAX_DATASET_NAME_LEN];
@@ -2522,8 +2527,7 @@ zfs_send_one(zfs_handle_t *zhp, const char *from, int fd, sendflags_t *flags,
 
 	if (flags->progress) {
 		void *status = NULL;
-		if (err != 0)
-			(void) pthread_cancel(ptid);
+		(void) pthread_cancel(ptid);
 		(void) pthread_join(ptid, &status);
 		int error = (int)(uintptr_t)status;
 		if (error != 0 && status != PTHREAD_CANCELED)

lib/libzutil/zutil_import.c

@@ -1791,16 +1791,13 @@ zpool_find_config(void *hdl, const char *target, nvlist_t **configp,
 	nvlist_t *match = NULL;
 	nvlist_t *config = NULL;
 	char *sepp = NULL;
-	char sep = '\0';
 	int count = 0;
 	char *targetdup = strdup(target);
 
 	*configp = NULL;
 
-	if ((sepp = strpbrk(targetdup, "/@")) != NULL) {
-		sep = *sepp;
+	if ((sepp = strpbrk(targetdup, "/@")) != NULL)
 		*sepp = '\0';
-	}
 
 	pools = zpool_search_import(hdl, args, pco);
 

man/Makefile.am

@@ -15,6 +15,7 @@ dist_man_MANS = \
 	man4/spl.4 \
 	man4/zfs.4 \
 	\
+	man7/dracut.zfs.7 \
 	man7/zpool-features.7 \
 	man7/zfsconcepts.7 \
 	man7/zfsprops.7 \

man/man4/zfs.4

@@ -1581,12 +1581,12 @@ Allow no-operation writes.
 The occurrence of nopwrites will further depend on other pool properties
 .Pq i.a. the checksumming and compression algorithms .
 .
-.It Sy zfs_dmu_offset_next_sync Ns = Ns Sy 0 Ns | Ns 1 Pq int
+.It Sy zfs_dmu_offset_next_sync Ns = Ns Sy 1 Ns | Ns 0 Pq int
 Enable forcing TXG sync to find holes.
-When enabled forces ZFS to act like prior versions when
+When enabled forces ZFS to sync data when
 .Sy SEEK_HOLE No or Sy SEEK_DATA
-flags are used, which, when a dnode is dirty,
-causes TXGs to be synced so that this data can be found.
+flags are used allowing holes in a file to be accurately reported.
+When disabled holes will not be reported in recently dirtied files.
 .
 .It Sy zfs_pd_bytes_max Ns = Ns Sy 52428800 Ns B Po 50MB Pc Pq int
 The number of bytes which should be prefetched during a pool traversal, like

man/man7/dracut.zfs.7

@@ -0,0 +1,278 @@
+.\" SPDX-License-Identifier: 0BSD
+.\"
+.Dd April 4, 2022
+.Dt DRACUT.ZFS 7
+.Os
+.
+.Sh NAME
+.Nm dracut.zfs
+.Nd overview of ZFS dracut hooks
+.
+.Sh SYNOPSIS
+.Bd -literal -compact
+                      parse-zfs.sh \(-> dracut-cmdline.service
+                          |                     \(da
+                          |                     …
+                          |                     \(da
+                          \e\(em\(em\(em\(em\(em\(em\(em\(em\(-> dracut-initqueue.service
+                                                |                      zfs-import-opts.sh
+   zfs-load-module.service                      \(da                          |       |
+     |                  |                sysinit.target                    \(da       |
+     \(da                  |                       |        zfs-import-scan.service   \(da
+zfs-import-scan.service \(da                       \(da           | zfs-import-cache.service
+     |   zfs-import-cache.service         basic.target      |     |
+     \e__________________|                       |           \(da     \(da
+                        \(da                       |     zfs-load-key.sh
+     zfs-env-bootfs.service                     |         |
+                        \(da                       \(da         \(da
+                 zfs-import.target \(-> dracut-pre-mount.service
+                        |          \(ua            |
+                        | dracut-zfs-generator  |
+                        |  ____________________/|
+                        |/                      \(da
+                        |                   sysroot.mount \(<-\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em dracut-zfs-generator
+                        |                       |                                        \(da   |
+                        |                       \(da            sysroot-{usr,etc,lib,&c.}.mount |
+                        |             initrd-root-fs.target \(<-\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em or                 \(da
+                        |                       |              zfs-nonroot-necessities.service
+                        |                       \(da                                 |
+                        \(da             dracut-mount.service                        |
+       zfs-snapshot-bootfs.service              |                                 |
+                        |                       \(da                                 |
+                        \(da                       …                                 |
+       zfs-rollback-bootfs.service              |                                 |
+                        |                       \(da                                 |
+                        |               sysroot-usr.mount \(<-\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em\(em/
+                        |                       |
+                        |                       \(da
+                        |                initrd-fs.target
+                        \e______________________ |
+                                               \e|
+                                                \(da
+        export-zfs.sh                      initrd.target
+              |                                 |
+              \(da                                 \(da
+   dracut-shutdown.service                      …
+                                                |
+                                                \(da
+                 zfs-needshutdown.sh \(-> initrd-cleanup.service
+.Ed
+.Pp
+Compare
+.Xr dracut.bootup 7
+for the full flowchart.
+.
+.Sh DESCRIPTION
+Under dracut, booting with
+.No ZFS-on- Ns Pa /
+is facilitated by a number of hooks in the
+.Nm 90zfs
+module.
+.Pp
+Booting into a ZFS dataset requires
+.Sy mountpoint Ns = Ns Pa /
+to be set on the dataset containing the root filesystem (henceforth "the boot dataset") and at the very least either the
+.Sy bootfs
+property to be set to that dataset, or the
+.Sy root=
+kernel cmdline (or dracut drop-in) argument to specify it.
+.Pp
+All children of the boot dataset with
+.Sy canmount Ns = Ns Sy on
+with
+.Sy mountpoint Ns s
+matching
+.Pa /etc , /bin , /lib , /lib?? , /libx32 , No and Pa /usr
+globs are deemed essential and will be mounted as well.
+.Pp
+.Xr zfs-mount-generator 8
+is recommended for proper functioning of the system afterward (correct mount properties, remounting, &c.).
+.
+.Sh CMDLINE
+.Ss Standard
+.Bl -tag -compact -width ".Sy root=zfs:AUTO , root=zfs: , root=zfs , Op Sy root="
+.It Sy root=zfs:\& Ns Ar dataset , Sy root=ZFS= Ns Ar dataset
+Use
+.Ar dataset
+as the boot dataset.
+All pluses
+.Pq Sq +
+are replaced with spaces
+.Pq Sq \  .
+.
+.It Sy root=zfs:AUTO , root=zfs:\& , root=zfs , Op Sy root=
+After import, search for the first pool with the
+.Sy bootfs
+property set, use its value as-if specified as the
+.Ar dataset
+above.
+.
+.It Sy rootfstype=zfs root= Ns Ar dataset
+Equivalent to
+.Sy root=zfs:\& Ns Ar dataset .
+.
+.It Sy rootfstype=zfs Op Sy root=
+Equivalent to
+.Sy root=zfs:AUTO .
+.
+.It Sy rootflags= Ns Ar flags
+Mount the boot dataset with
+.Fl o Ar flags ;
+cf.\&
+.Sx Temporary Mount Point Properties
+in
+.Xr zfsprops 7 .
+These properties will not last, since all filesystems will be re-mounted from the real root.
+.
+.It Sy debug
+If specified,
+.Nm dracut-zfs-generator
+logs to the journal.
+.El
+.Pp
+Be careful about setting neither
+.Sy rootfstype=zfs
+nor
+.Sy root=zfs:\& Ns Ar dataset
+\(em other automatic boot selection methods, like
+.Nm systemd-gpt-auto-generator
+and
+.Nm systemd-fstab-generator
+might take precedent.
+.
+.Ss ZFS-specific
+.Bl -tag -compact -width ".Sy bootfs.snapshot Ns Op Sy = Ns Ar snapshot-name"
+.It Sy bootfs.snapshot Ns Op Sy = Ns Ar snapshot-name
+Execute
+.Nm zfs Cm snapshot Ar boot-dataset Ns Sy @ Ns Ar snapshot-name
+before pivoting to the real root.
+.Ar snapshot-name
+defaults to the current kernel release.
+.
+.It Sy bootfs.rollback Ns Op Sy = Ns Ar snapshot-name
+Execute
+.Nm zfs Cm snapshot Fl Rf Ar boot-dataset Ns Sy @ Ns Ar snapshot-name
+before pivoting to the real root.
+.Ar snapshot-name
+defaults to the current kernel release.
+.
+.It Sy spl_hostid= Ns Ar host-id
+Use
+.Xr zgenhostid 8
+to set the host ID to
+.Ar host-id ;
+otherwise,
+.Pa /etc/hostid
+inherited from the real root is used.
+.
+.It Sy zfs_force , zfs.force , zfsforce
+Appends
+.Fl f
+to all
+.Nm zpool Cm import
+invocations; primarily useful in conjunction with
+.Sy spl_hostid= ,
+or if no host ID was inherited.
+.El
+.
+.Sh FILES
+.Bl -tag -width 0
+.It Pa parse-zfs.sh Pq Sy cmdline
+Processes
+.Sy spl_hostid= .
+If
+.Sy root=
+matches a known pattern, above, provides
+.Pa /dev/root
+and delays the initqueue until
+.Xr zfs 4
+is loaded,
+.
+.It Pa zfs-import-opts.sh Pq Nm systemd No environment generator
+Turns
+.Sy zfs_force , zfs.force , No or Sy zfsforce
+into
+.Ev ZPOOL_IMPORT_OPTS Ns = Ns Fl f
+for
+.Pa zfs-import-scan.service
+or
+.Pa zfs-import-cache.service .
+.
+.It Pa zfs-load-key.sh Pq Sy pre-mount
+Loads encryption keys for the boot dataset and its essential descendants.
+.Bl -tag -compact -offset 4n -width ".Sy keylocation Ns = Ns Sy https:// Ns Ar URL , Sy keylocation Ns = Ns Sy http:// Ns Ar URL"
+.It Sy keylocation Ns = Ns Sy prompt
+Is prompted for via
+.Nm systemd-ask-password
+thrice.
+.
+.It Sy keylocation Ns = Ns Sy https:// Ns Ar URL , Sy keylocation Ns = Ns Sy http:// Ns Ar URL
+.Pa network-online.target
+is started before loading.
+.
+.It Sy keylocation Ns = Ns Sy file:// Ns Ar path
+If
+.Ar path
+doesn't exist,
+.Nm udevadm No is Cm settle Ns d .
+If it still doesn't, it's waited for for up to
+.Sy 10 Ns s .
+.El
+.
+.It Pa zfs-env-bootfs.service Pq Nm systemd No service
+After pool import, sets
+.Ev BOOTFS Ns =
+in the systemd environment to the first non-null
+.Sy bootfs
+value in iteration order.
+.
+.It Pa dracut-zfs-generator Pq Nm systemd No generator
+Generates
+.Pa sysroot.mount Pq using Sy rootflags= , No if any .
+If an explicit boot dataset was specified, also generates essential mountpoints
+.Pq Pa sysroot-etc.mount , sysroot-bin.mount , No &c.\& ,
+otherwise generates
+.Pa zfs-nonroot-necessities.service
+which mounts them explicitly after
+.Pa /sysroot
+using
+.Ev BOOTFS Ns = .
+.
+.It Pa zfs-snapshot-bootfs.service , zfs-rollback-bootfs.service Pq Nm systemd No services
+Consume
+.Sy bootfs.snapshot
+and
+.Sy bootfs.rollback
+as described in
+.Sx CMDLINE  .
+Use
+.Ev BOOTFS Ns =
+if no explicit boot dataset was specified.
+.
+.It Pa zfs-needshutdown.sh Pq Sy cleanup
+If any pools were imported, signals that shutdown hooks are required.
+.
+.It Pa export-zfs.sh Pq Sy shutdown
+Forcibly exports all pools.
+.
+.It Pa /etc/hostid , /etc/zfs/zpool.cache , /etc/zfs/vdev_id.conf Pq regular files
+Included verbatim, hostonly.
+.
+.It Pa mount-zfs.sh Pq Sy mount
+Does nothing on
+.Nm systemd
+systems
+.Pq if Pa dracut-zfs-generator No succeeded .
+Otherwise, loads encryption key for the boot dataset from the console or via plymouth.
+It may not work at all!
+.El
+.
+.Sh SEE ALSO
+.Xr dracut.bootup 7 ,
+.Xr zfsprops 7 ,
+.Xr zpoolprops 7 ,
+.Xr dracut-shutdown.service 8 ,
+.Xr systemd-fstab-generator 8 ,
+.Xr systemd-gpt-auto-generator 8 ,
+.Xr zfs-mount-generator 8 ,
+.Xr zgenhostid 8

man/man8/zfs-diff.8

@@ -39,7 +39,7 @@
 .Sh SYNOPSIS
 .Nm zfs
 .Cm diff
-.Op Fl FHt
+.Op Fl FHth
 .Ar snapshot Ar snapshot Ns | Ns Ar filesystem
 .
 .Sh DESCRIPTION
@@ -92,6 +92,10 @@ Give more parsable tab-separated output, without header lines and without
 arrows.
 .It Fl t
 Display the path's inode change time as the first column of output.
+.It Fl h
+Do not
+.Sy \e0 Ns Ar ooo Ns -escape
+non-ASCII paths.
 .El
 .
 .Sh SEE ALSO

man/man8/zfs-mount-generator.8.in

@@ -142,22 +142,11 @@ ZEDLET, if enabled
 .Pq see Xr zed 8 .
 .
 .Sh ENVIRONMENT
-The
+If the
 .Sy ZFS_DEBUG
-environment variable can either be
-.Sy 0
-(default),
-.Sy 1
-(print summary accounting information at the end), or at least
-.Sy 2
-(print accounting information for each subprocess as it finishes).
-.
-If not present,
-.Pa /proc/cmdline
-is additionally checked for
-.Qq debug ,
-in which case the debug level is set to
-.Sy 2 .
+environment variable is nonzero
+.Pq or unset and Pa /proc/cmdline No contains Qq Sy debug ,
+print summary accounting information at the end.
 .
 .Sh EXAMPLES
 To begin, enable tracking for the pool:

man/man8/zfs-set.8

@@ -170,8 +170,9 @@ inherited.
 .It Fl r
 Recursively inherit the given property for all children.
 .It Fl S
-Revert the property to the received value if one exists; otherwise operate as
-if the
+Revert the property to the received value, if one exists;
+otherwise, for non-inheritable properties, to the default;
+otherwise, operate as if the
 .Fl S
 option was not specified.
 .El

man/man8/zpool-import.8

@@ -71,7 +71,8 @@
 .Xc
 Lists pools available to import.
 If the
-.Fl d or
+.Fl d
+or
 .Fl c
 options are not specified, this command searches for devices using libblkid
 on Linux and geom on

module/Makefile.bsd

@@ -320,6 +320,7 @@ beforeinstall:
 
 
 CFLAGS.gcc+= -Wno-pointer-to-int-cast
+CFLAGS.clang+= ${NO_WUNUSED_BUT_SET_VARIABLE}
 
 CFLAGS.lapi.c= -Wno-cast-qual
 CFLAGS.lcompat.c= -Wno-cast-qual

module/icp/algs/modes/gcm.c

@@ -798,7 +798,7 @@ static gcm_impl_ops_t *gcm_supp_impl[ARRAY_SIZE(gcm_all_impl)];
  * fallback to the fastest generic implementation.
  */
 const gcm_impl_ops_t *
-gcm_impl_get_ops()
+gcm_impl_get_ops(void)
 {
 	if (!kfpu_allowed())
 		return (&gcm_generic_impl);

module/os/freebsd/zfs/abd_os.c

@@ -113,7 +113,6 @@ static kstat_t *abd_ksp;
  * memory by only using a single zero buffer for the scatter chunks.
  */
 abd_t *abd_zero_scatter = NULL;
-static char *abd_zero_buf = NULL;
 
 static uint_t
 abd_chunkcnt_for_bytes(size_t size)
@@ -241,18 +240,16 @@ abd_free_struct_impl(abd_t *abd)
 
 /*
  * Allocate scatter ABD of size SPA_MAXBLOCKSIZE, where
- * each chunk in the scatterlist will be set to abd_zero_buf.
+ * each chunk in the scatterlist will be set to the same area.
  */
+_Static_assert(ZERO_REGION_SIZE >= PAGE_SIZE, "zero_region too small");
 static void
 abd_alloc_zero_scatter(void)
 {
 	uint_t i, n;
 
 	n = abd_chunkcnt_for_bytes(SPA_MAXBLOCKSIZE);
-	abd_zero_buf = kmem_cache_alloc(abd_chunk_cache, KM_PUSHPAGE);
-	bzero(abd_zero_buf, PAGE_SIZE);
 	abd_zero_scatter = abd_alloc_struct(SPA_MAXBLOCKSIZE);
-
 	abd_zero_scatter->abd_flags |= ABD_FLAG_OWNER | ABD_FLAG_ZEROS;
 	abd_zero_scatter->abd_size = SPA_MAXBLOCKSIZE;
 
@@ -260,7 +257,7 @@ abd_alloc_zero_scatter(void)
 
 	for (i = 0; i < n; i++) {
 		ABD_SCATTER(abd_zero_scatter).abd_chunks[i] =
-		    abd_zero_buf;
+		    __DECONST(void *, zero_region);
 	}
 
 	ABDSTAT_BUMP(abdstat_scatter_cnt);
@@ -275,7 +272,6 @@ abd_free_zero_scatter(void)
 
 	abd_free_struct(abd_zero_scatter);
 	abd_zero_scatter = NULL;
-	kmem_cache_free(abd_chunk_cache, abd_zero_buf);
 }
 
 static int

module/os/freebsd/zfs/spa_os.c

@@ -183,7 +183,6 @@ spa_import_rootpool(const char *name, bool checkpointrewind)
 	spa_t *spa;
 	vdev_t *rvd;
 	nvlist_t *config, *nvtop;
-	uint64_t txg;
 	char *pname;
 	int error;
 
@@ -196,7 +195,6 @@ spa_import_rootpool(const char *name, bool checkpointrewind)
 	if (config != NULL) {
 		pname = fnvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME);
 		VERIFY0(strcmp(name, pname));
-		txg = fnvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_TXG);
 
 		if ((spa = spa_lookup(pname)) != NULL) {
 			/*

module/os/freebsd/zfs/vdev_geom.c

@@ -1132,8 +1132,12 @@ vdev_geom_fill_unmap_cb(void *buf, size_t len, void *priv)
 	vm_offset_t addr = (vm_offset_t)buf;
 	vm_offset_t end = addr + len;
 
-	if (bp->bio_ma_n == 0)
+	if (bp->bio_ma_n == 0) {
 		bp->bio_ma_offset = addr & PAGE_MASK;
+		addr &= ~PAGE_MASK;
+	} else {
+		ASSERT0(P2PHASE(addr, PAGE_SIZE));
+	}
 	do {
 		bp->bio_ma[bp->bio_ma_n++] =
 		    PHYS_TO_VM_PAGE(pmap_kextract(addr));

module/os/freebsd/zfs/zfs_vnops_os.c

@@ -97,6 +97,10 @@
 
 VFS_SMR_DECLARE;
 
+#if __FreeBSD_version < 1300103
+#define	NDFREE_PNBUF(ndp)	NDFREE((ndp), NDF_ONLY_PNBUF)
+#endif
+
 #if __FreeBSD_version >= 1300047
 #define	vm_page_wire_lock(pp)
 #define	vm_page_wire_unlock(pp)
@@ -4058,8 +4062,8 @@ zfs_getpages(struct vnode *vp, vm_page_t *ma, int count, int *rbehind,
 	int pgsin_b, pgsin_a;
 	int error;
 
-	ZFS_ENTER(zfsvfs);
-	ZFS_VERIFY_ZP(zp);
+	ZFS_ENTER_ERROR(zfsvfs, zfs_vm_pagerret_error);
+	ZFS_VERIFY_ZP_ERROR(zp, zfs_vm_pagerret_error);
 
 	start = IDX_TO_OFF(ma[0]->pindex);
 	end = IDX_TO_OFF(ma[count - 1]->pindex + 1);
@@ -4183,19 +4187,18 @@ zfs_putpages(struct vnode *vp, vm_page_t *ma, size_t len, int flags,
 	int		err;
 	int		i;
 
-	ZFS_ENTER(zfsvfs);
-	ZFS_VERIFY_ZP(zp);
-
 	object = vp->v_object;
-	pcount = btoc(len);
-	ncount = pcount;
-
 	KASSERT(ma[0]->object == object, ("mismatching object"));
 	KASSERT(len > 0 && (len & PAGE_MASK) == 0, ("unexpected length"));
 
+	pcount = btoc(len);
+	ncount = pcount;
 	for (i = 0; i < pcount; i++)
 		rtvals[i] = zfs_vm_pagerret_error;
 
+	ZFS_ENTER_ERROR(zfsvfs, zfs_vm_pagerret_error);
+	ZFS_VERIFY_ZP_ERROR(zp, zfs_vm_pagerret_error);
+
 	off = IDX_TO_OFF(ma[0]->pindex);
 	blksz = zp->z_blksz;
 	lo_off = rounddown(off, blksz);
@@ -5234,6 +5237,11 @@ zfs_freebsd_pathconf(struct vop_pathconf_args *ap)
 	case _PC_NAME_MAX:
 		*ap->a_retval = NAME_MAX;
 		return (0);
+#if __FreeBSD_version >= 1400032
+	case _PC_DEALLOC_PRESENT:
+		*ap->a_retval = 1;
+		return (0);
+#endif
 	case _PC_PIPE_BUF:
 		if (ap->a_vp->v_type == VDIR || ap->a_vp->v_type == VFIFO) {
 			*ap->a_retval = PIPE_BUF;
@@ -5357,7 +5365,7 @@ zfs_getextattr_dir(struct vop_getextattr_args *ap, const char *attrname)
 #endif
 	error = vn_open_cred(&nd, &flags, 0, VN_OPEN_INVFS, ap->a_cred, NULL);
 	vp = nd.ni_vp;
-	NDFREE(&nd, NDF_ONLY_PNBUF);
+	NDFREE_PNBUF(&nd);
 	if (error != 0)
 		return (error);
 
@@ -5430,7 +5438,7 @@ zfs_getextattr(struct vop_getextattr_args *ap)
 
 	error = ENOENT;
 	ZFS_ENTER(zfsvfs);
-	ZFS_VERIFY_ZP(zp)
+	ZFS_VERIFY_ZP(zp);
 	rw_enter(&zp->z_xattr_lock, RW_READER);
 	if (zfsvfs->z_use_sa && zp->z_is_sa)
 		error = zfs_getextattr_sa(ap, attrname);
@@ -5475,12 +5483,12 @@ zfs_deleteextattr_dir(struct vop_deleteextattr_args *ap, const char *attrname)
 	error = namei(&nd);
 	vp = nd.ni_vp;
 	if (error != 0) {
-		NDFREE(&nd, NDF_ONLY_PNBUF);
+		NDFREE_PNBUF(&nd);
 		return (error);
 	}
 
 	error = VOP_REMOVE(nd.ni_dvp, vp, &nd.ni_cnd);
-	NDFREE(&nd, NDF_ONLY_PNBUF);
+	NDFREE_PNBUF(&nd);
 
 	vput(nd.ni_dvp);
 	if (vp == nd.ni_dvp)
@@ -5605,7 +5613,7 @@ zfs_setextattr_dir(struct vop_setextattr_args *ap, const char *attrname)
 	error = vn_open_cred(&nd, &flags, 0600, VN_OPEN_INVFS, ap->a_cred,
 	    NULL);
 	vp = nd.ni_vp;
-	NDFREE(&nd, NDF_ONLY_PNBUF);
+	NDFREE_PNBUF(&nd);
 	if (error != 0)
 		return (error);
 
@@ -5760,7 +5768,7 @@ zfs_listextattr_dir(struct vop_listextattr_args *ap, const char *attrprefix)
 #endif
 	error = namei(&nd);
 	vp = nd.ni_vp;
-	NDFREE(&nd, NDF_ONLY_PNBUF);
+	NDFREE_PNBUF(&nd);
 	if (error != 0)
 		return (error);
 
@@ -6073,6 +6081,55 @@ zfs_vptocnp(struct vop_vptocnp_args *ap)
 	return (error);
 }
 
+#if __FreeBSD_version >= 1400032
+static int
+zfs_deallocate(struct vop_deallocate_args *ap)
+{
+	znode_t *zp = VTOZ(ap->a_vp);
+	zfsvfs_t *zfsvfs = zp->z_zfsvfs;
+	zilog_t *zilog;
+	off_t off, len, file_sz;
+	int error;
+
+	ZFS_ENTER(zfsvfs);
+	ZFS_VERIFY_ZP(zp);
+
+	/*
+	 * Callers might not be able to detect properly that we are read-only,
+	 * so check it explicitly here.
+	 */
+	if (zfs_is_readonly(zfsvfs)) {
+		ZFS_EXIT(zfsvfs);
+		return (SET_ERROR(EROFS));
+	}
+
+	zilog = zfsvfs->z_log;
+	off = *ap->a_offset;
+	len = *ap->a_len;
+	file_sz = zp->z_size;
+	if (off + len > file_sz)
+		len = file_sz - off;
+	/* Fast path for out-of-range request. */
+	if (len <= 0) {
+		*ap->a_len = 0;
+		ZFS_EXIT(zfsvfs);
+		return (0);
+	}
+
+	error = zfs_freesp(zp, off, len, O_RDWR, TRUE);
+	if (error == 0) {
+		if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS ||
+		    (ap->a_ioflag & IO_SYNC) != 0)
+			zil_commit(zilog, zp->z_id);
+		*ap->a_offset = off + len;
+		*ap->a_len = 0;
+	}
+
+	ZFS_EXIT(zfsvfs);
+	return (error);
+}
+#endif
+
 struct vop_vector zfs_vnodeops;
 struct vop_vector zfs_fifoops;
 struct vop_vector zfs_shareops;
@@ -6092,6 +6149,9 @@ struct vop_vector zfs_vnodeops = {
 #endif
 	.vop_access =		zfs_freebsd_access,
 	.vop_allocate =		VOP_EINVAL,
+#if __FreeBSD_version >= 1400032
+	.vop_deallocate =	zfs_deallocate,
+#endif
 	.vop_lookup =		zfs_cache_lookup,
 	.vop_cachedlookup =	zfs_freebsd_cachedlookup,
 	.vop_getattr =		zfs_freebsd_getattr,

module/os/freebsd/zfs/zfs_znode.c

@@ -575,7 +575,6 @@ zfs_mknode(znode_t *dzp, vattr_t *vap, dmu_tx_t *tx, cred_t *cr,
 	dmu_buf_t	*db;
 	timestruc_t	now;
 	uint64_t	gen, obj;
-	int		err;
 	int		bonuslen;
 	int		dnodesize;
 	sa_handle_t	*sa_hdl;
@@ -815,12 +814,11 @@ zfs_mknode(znode_t *dzp, vattr_t *vap, dmu_tx_t *tx, cred_t *cr,
 		VERIFY0(zfs_aclset_common(*zpp, acl_ids->z_aclp, cr, tx));
 	}
 	if (!(flag & IS_ROOT_NODE)) {
-		vnode_t *vp;
-
-		vp = ZTOV(*zpp);
+		vnode_t *vp = ZTOV(*zpp);
 		vp->v_vflag |= VV_FORCEINSMQ;
-		err = insmntque(vp, zfsvfs->z_vfs);
+		int err = insmntque(vp, zfsvfs->z_vfs);
 		vp->v_vflag &= ~VV_FORCEINSMQ;
+		(void) err;
 		KASSERT(err == 0, ("insmntque() failed: error %d", err));
 	}
 	kmem_free(sa_attrs, sizeof (sa_bulk_attr_t) * ZPL_END);
@@ -934,11 +932,9 @@ zfs_zget(zfsvfs_t *zfsvfs, uint64_t obj_num, znode_t **zpp)
 	znode_t		*zp;
 	vnode_t		*vp;
 	sa_handle_t	*hdl;
-	struct thread	*td;
 	int locked;
 	int err;
 
-	td = curthread;
 	getnewvnode_reserve_();
 again:
 	*zpp = NULL;
@@ -964,7 +960,7 @@ again:
 
 	hdl = dmu_buf_get_user(db);
 	if (hdl != NULL) {
-		zp  = sa_get_userdata(hdl);
+		zp = sa_get_userdata(hdl);
 
 		/*
 		 * Since "SA" does immediate eviction we
@@ -1491,12 +1487,16 @@ zfs_free_range(znode_t *zp, uint64_t off, uint64_t len)
 	error = dmu_free_long_range(zfsvfs->z_os, zp->z_id, off, len);
 
 	if (error == 0) {
+#if __FreeBSD_version >= 1400032
+		vnode_pager_purge_range(ZTOV(zp), off, off + len);
+#else
 		/*
-		 * In FreeBSD we cannot free block in the middle of a file,
-		 * but only at the end of a file, so this code path should
-		 * never happen.
+		 * Before __FreeBSD_version 1400032 we cannot free block in the
+		 * middle of a file, but only at the end of a file, so this code
+		 * path should never happen.
 		 */
 		vnode_pager_setsize(ZTOV(zp), off);
+#endif
 	}
 
 	zfs_rangelock_exit(lr);

module/os/freebsd/zfs/zvol_os.c

@@ -1063,7 +1063,7 @@ zvol_cdev_ioctl(struct cdev *dev, ulong_t cmd, caddr_t data,
 	zvol_state_t *zv;
 	zfs_locked_range_t *lr;
 	off_t offset, length;
-	int i, error;
+	int error;
 	boolean_t sync;
 
 	zv = dev->si_drv2;
@@ -1072,7 +1072,6 @@ zvol_cdev_ioctl(struct cdev *dev, ulong_t cmd, caddr_t data,
 	KASSERT(zv->zv_open_count > 0,
 	    ("Device with zero access count in %s", __func__));
 
-	i = IOCPARM_LEN(cmd);
 	switch (cmd) {
 	case DIOCGSECTORSIZE:
 		*(uint32_t *)data = DEV_BSIZE;

module/os/linux/spl/spl-kmem-cache.c

@@ -1421,7 +1421,7 @@ EXPORT_SYMBOL(spl_kmem_cache_reap_now);
  * it should do no harm.
  */
 int
-spl_kmem_cache_reap_active()
+spl_kmem_cache_reap_active(void)
 {
 	return (0);
 }

module/os/linux/zfs/abd_os.c

@@ -184,8 +184,11 @@ abd_t *abd_zero_scatter = NULL;
 
 struct page;
 /*
- * abd_zero_page we will be an allocated zero'd PAGESIZE buffer, which is
- * assigned to set each of the pages of abd_zero_scatter.
+ * _KERNEL   - Will point to ZERO_PAGE if it is available or it will be
+ *             an allocated zero'd PAGESIZE buffer.
+ * Userspace - Will be an allocated zero'ed PAGESIZE buffer.
+ *
+ * abd_zero_page is assigned to each of the pages of abd_zero_scatter.
  */
 static struct page *abd_zero_page = NULL;
 
@@ -465,15 +468,19 @@ abd_alloc_zero_scatter(void)
 	struct scatterlist *sg = NULL;
 	struct sg_table table;
 	gfp_t gfp = __GFP_NOWARN | GFP_NOIO;
-	gfp_t gfp_zero_page = gfp | __GFP_ZERO;
 	int nr_pages = abd_chunkcnt_for_bytes(SPA_MAXBLOCKSIZE);
 	int i = 0;
 
+#if defined(HAVE_ZERO_PAGE_GPL_ONLY)
+	gfp_t gfp_zero_page = gfp | __GFP_ZERO;
 	while ((abd_zero_page = __page_cache_alloc(gfp_zero_page)) == NULL) {
 		ABDSTAT_BUMP(abdstat_scatter_page_alloc_retry);
 		schedule_timeout_interruptible(1);
 	}
 	abd_mark_zfs_page(abd_zero_page);
+#else
+	abd_zero_page = ZERO_PAGE(0);
+#endif /* HAVE_ZERO_PAGE_GPL_ONLY */
 
 	while (sg_alloc_table(&table, nr_pages, gfp)) {
 		ABDSTAT_BUMP(abdstat_scatter_sg_table_retry);
@@ -612,7 +619,6 @@ abd_alloc_zero_scatter(void)
 	ABD_SCATTER(abd_zero_scatter).abd_offset = 0;
 	ABD_SCATTER(abd_zero_scatter).abd_nents = nr_pages;
 	abd_zero_scatter->abd_size = SPA_MAXBLOCKSIZE;
-	zfs_refcount_create(&abd_zero_scatter->abd_children);
 	ABD_SCATTER(abd_zero_scatter).abd_sgl = vmem_alloc(nr_pages *
 	    sizeof (struct scatterlist), KM_SLEEP);
 
@@ -694,8 +700,10 @@ abd_free_zero_scatter(void)
 	abd_zero_scatter = NULL;
 	ASSERT3P(abd_zero_page, !=, NULL);
 #if defined(_KERNEL)
+#if defined(HAVE_ZERO_PAGE_GPL_ONLY)
 	abd_unmark_zfs_page(abd_zero_page);
 	__free_page(abd_zero_page);
+#endif /* HAVE_ZERO_PAGE_GPL_ONLY */
 #else
 	umem_free(abd_zero_page, PAGESIZE);
 #endif /* _KERNEL */

module/os/linux/zfs/vdev_disk.c

@@ -304,8 +304,6 @@ vdev_disk_open(vdev_t *v, uint64_t *psize, uint64_t *max_psize,
 		rw_exit(&vd->vd_lock);
 	}
 
-	struct request_queue *q = bdev_get_queue(vd->vd_bdev);
-
 	/*  Determine the physical block size */
 	int physical_block_size = bdev_physical_block_size(vd->vd_bdev);
 
@@ -316,13 +314,13 @@ vdev_disk_open(vdev_t *v, uint64_t *psize, uint64_t *max_psize,
 	v->vdev_nowritecache = B_FALSE;
 
 	/* Set when device reports it supports TRIM. */
-	v->vdev_has_trim = !!blk_queue_discard(q);
+	v->vdev_has_trim = bdev_discard_supported(vd->vd_bdev);
 
 	/* Set when device reports it supports secure TRIM. */
-	v->vdev_has_securetrim = !!blk_queue_discard_secure(q);
+	v->vdev_has_securetrim = bdev_secure_discard_supported(vd->vd_bdev);
 
 	/* Inform the ZIO pipeline that we are non-rotational */
-	v->vdev_nonrot = blk_queue_nonrot(q);
+	v->vdev_nonrot = blk_queue_nonrot(bdev_get_queue(vd->vd_bdev));
 
 	/* Physical volume size in bytes for the partition */
 	*psize = bdev_capacity(vd->vd_bdev);
@@ -460,6 +458,13 @@ vdev_submit_bio_impl(struct bio *bio)
 #define	preempt_schedule_notrace(x) preempt_schedule(x)
 #endif
 
+/*
+ * As for the Linux 5.18 kernel bio_alloc() expects a block_device struct
+ * as an argument removing the need to set it with bio_set_dev().  This
+ * removes the need for all of the following compatibility code.
+ */
+#if !defined(HAVE_BIO_ALLOC_4ARG)
+
 #ifdef HAVE_BIO_SET_DEV
 #if defined(CONFIG_BLK_CGROUP) && defined(HAVE_BIO_SET_DEV_GPL_ONLY)
 /*
@@ -467,8 +472,11 @@ vdev_submit_bio_impl(struct bio *bio)
  * blkg_tryget() to use rcu_read_lock() instead of rcu_read_lock_sched().
  * As a side effect the function was converted to GPL-only.  Define our
  * own version when needed which uses rcu_read_lock_sched().
+ *
+ * The Linux 5.17 kernel split linux/blk-cgroup.h into a private and a public
+ * part, moving blkg_tryget into the private one. Define our own version.
  */
-#if defined(HAVE_BLKG_TRYGET_GPL_ONLY)
+#if defined(HAVE_BLKG_TRYGET_GPL_ONLY) || !defined(HAVE_BLKG_TRYGET)
 static inline bool
 vdev_blkg_tryget(struct blkcg_gq *blkg)
 {
@@ -493,7 +501,7 @@ vdev_blkg_tryget(struct blkcg_gq *blkg)
 
 	return (rc);
 }
-#elif defined(HAVE_BLKG_TRYGET)
+#else
 #define	vdev_blkg_tryget(bg)	blkg_tryget(bg)
 #endif
 #ifdef HAVE_BIO_SET_DEV_MACRO
@@ -553,6 +561,7 @@ bio_set_dev(struct bio *bio, struct block_device *bdev)
 	bio->bi_bdev = bdev;
 }
 #endif /* HAVE_BIO_SET_DEV */
+#endif /* !HAVE_BIO_ALLOC_4ARG */
 
 static inline void
 vdev_submit_bio(struct bio *bio)
@@ -563,6 +572,36 @@ vdev_submit_bio(struct bio *bio)
 	current->bio_list = bio_list;
 }
 
+static inline struct bio *
+vdev_bio_alloc(struct block_device *bdev, gfp_t gfp_mask,
+    unsigned short nr_vecs)
+{
+	struct bio *bio;
+
+#ifdef HAVE_BIO_ALLOC_4ARG
+	bio = bio_alloc(bdev, nr_vecs, 0, gfp_mask);
+#else
+	bio = bio_alloc(gfp_mask, nr_vecs);
+	if (likely(bio != NULL))
+		bio_set_dev(bio, bdev);
+#endif
+
+	return (bio);
+}
+
+static inline unsigned int
+vdev_bio_max_segs(zio_t *zio, int bio_size, uint64_t abd_offset)
+{
+	unsigned long nr_segs = abd_nr_pages_off(zio->io_abd,
+	    bio_size, abd_offset);
+
+#ifdef HAVE_BIO_MAX_SEGS
+	return (bio_max_segs(nr_segs));
+#else
+	return (MIN(nr_segs, BIO_MAX_PAGES));
+#endif
+}
+
 static int
 __vdev_disk_physio(struct block_device *bdev, zio_t *zio,
     size_t io_size, uint64_t io_offset, int rw, int flags)
@@ -574,6 +613,7 @@ __vdev_disk_physio(struct block_device *bdev, zio_t *zio,
 	int bio_count = 16;
 	int error = 0;
 	struct blk_plug plug;
+	unsigned short nr_vecs;
 
 	/*
 	 * Accessing outside the block device is never allowed.
@@ -623,15 +663,8 @@ retry:
 			goto retry;
 		}
 
-		/* bio_alloc() with __GFP_WAIT never returns NULL */
-#ifdef HAVE_BIO_MAX_SEGS
-		dr->dr_bio[i] = bio_alloc(GFP_NOIO, bio_max_segs(
-		    abd_nr_pages_off(zio->io_abd, bio_size, abd_offset)));
-#else
-		dr->dr_bio[i] = bio_alloc(GFP_NOIO,
-		    MIN(abd_nr_pages_off(zio->io_abd, bio_size, abd_offset),
-		    BIO_MAX_PAGES));
-#endif
+		nr_vecs = vdev_bio_max_segs(zio, bio_size, abd_offset);
+		dr->dr_bio[i] = vdev_bio_alloc(bdev, GFP_NOIO, nr_vecs);
 		if (unlikely(dr->dr_bio[i] == NULL)) {
 			vdev_disk_dio_free(dr);
 			return (SET_ERROR(ENOMEM));
@@ -640,7 +673,6 @@ retry:
 		/* Matching put called by vdev_disk_physio_completion */
 		vdev_disk_dio_get(dr);
 
-		bio_set_dev(dr->dr_bio[i], bdev);
 		BIO_BI_SECTOR(dr->dr_bio[i]) = bio_offset >> 9;
 		dr->dr_bio[i]->bi_end_io = vdev_disk_physio_completion;
 		dr->dr_bio[i]->bi_private = dr;
@@ -704,14 +736,12 @@ vdev_disk_io_flush(struct block_device *bdev, zio_t *zio)
 	if (!q)
 		return (SET_ERROR(ENXIO));
 
-	bio = bio_alloc(GFP_NOIO, 0);
-	/* bio_alloc() with __GFP_WAIT never returns NULL */
+	bio = vdev_bio_alloc(bdev, GFP_NOIO, 0);
 	if (unlikely(bio == NULL))
 		return (SET_ERROR(ENOMEM));
 
 	bio->bi_end_io = vdev_disk_io_flush_completion;
 	bio->bi_private = zio;
-	bio_set_dev(bio, bdev);
 	bio_set_flush(bio);
 	vdev_submit_bio(bio);
 	invalidate_bdev(bdev);
@@ -719,12 +749,38 @@ vdev_disk_io_flush(struct block_device *bdev, zio_t *zio)
 	return (0);
 }
 
+static int
+vdev_disk_io_trim(zio_t *zio)
+{
+	vdev_t *v = zio->io_vd;
+	vdev_disk_t *vd = v->vdev_tsd;
+
+#if defined(HAVE_BLKDEV_ISSUE_SECURE_ERASE)
+	if (zio->io_trim_flags & ZIO_TRIM_SECURE) {
+		return (-blkdev_issue_secure_erase(vd->vd_bdev,
+		    zio->io_offset >> 9, zio->io_size >> 9, GFP_NOFS));
+	} else {
+		return (-blkdev_issue_discard(vd->vd_bdev,
+		    zio->io_offset >> 9, zio->io_size >> 9, GFP_NOFS));
+	}
+#elif defined(HAVE_BLKDEV_ISSUE_DISCARD)
+	unsigned long trim_flags = 0;
+#if defined(BLKDEV_DISCARD_SECURE)
+	if (zio->io_trim_flags & ZIO_TRIM_SECURE)
+		trim_flags |= BLKDEV_DISCARD_SECURE;
+#endif
+	return (-blkdev_issue_discard(vd->vd_bdev,
+	    zio->io_offset >> 9, zio->io_size >> 9, GFP_NOFS, trim_flags));
+#else
+#error "Unsupported kernel"
+#endif
+}
+
 static void
 vdev_disk_io_start(zio_t *zio)
 {
 	vdev_t *v = zio->io_vd;
 	vdev_disk_t *vd = v->vdev_tsd;
-	unsigned long trim_flags = 0;
 	int rw, error;
 
 	/*
@@ -797,14 +853,7 @@ vdev_disk_io_start(zio_t *zio)
 		break;
 
 	case ZIO_TYPE_TRIM:
-#if defined(BLKDEV_DISCARD_SECURE)
-		if (zio->io_trim_flags & ZIO_TRIM_SECURE)
-			trim_flags |= BLKDEV_DISCARD_SECURE;
-#endif
-		zio->io_error = -blkdev_issue_discard(vd->vd_bdev,
-		    zio->io_offset >> 9, zio->io_size >> 9, GFP_NOFS,
-		    trim_flags);
-
+		zio->io_error = vdev_disk_io_trim(zio);
 		rw_exit(&vd->vd_lock);
 		zio_interrupt(zio);
 		return;

module/os/linux/zfs/zfs_sysfs.c

@@ -65,16 +65,15 @@
 /*
  * A zfs_mod_kobj_t represents a zfs kobject under '/sys/module/zfs'
  */
-struct zfs_mod_kobj;
 typedef struct zfs_mod_kobj zfs_mod_kobj_t;
-
 struct zfs_mod_kobj {
 	struct kobject		zko_kobj;
 	struct kobj_type	zko_kobj_type;
 	struct sysfs_ops	zko_sysfs_ops;
 	size_t			zko_attr_count;
 	struct attribute	*zko_attr_list;		/* allocated */
-	struct attribute	**zko_default_attrs;	/* allocated */
+	struct attribute_group	zko_default_group;	/* .attrs allocated */
+	const struct attribute_group	*zko_default_groups[2];
 	size_t			zko_child_count;
 	zfs_mod_kobj_t		*zko_children;		/* allocated */
 };
@@ -126,10 +125,10 @@ zfs_kobj_release(struct kobject *kobj)
 		zkobj->zko_attr_list = NULL;
 	}
 
-	if (zkobj->zko_default_attrs != NULL) {
-		kmem_free(zkobj->zko_default_attrs,
+	if (zkobj->zko_default_group.attrs != NULL) {
+		kmem_free(zkobj->zko_default_group.attrs,
 		    DEFAULT_ATTR_SIZE(zkobj->zko_attr_count));
-		zkobj->zko_default_attrs = NULL;
+		zkobj->zko_default_group.attrs = NULL;
 	}
 
 	if (zkobj->zko_child_count != 0) {
@@ -153,11 +152,12 @@ zfs_kobj_add_attr(zfs_mod_kobj_t *zkobj, int attr_num, const char *attr_name)
 {
 	VERIFY3U(attr_num, <, zkobj->zko_attr_count);
 	ASSERT(zkobj->zko_attr_list);
-	ASSERT(zkobj->zko_default_attrs);
+	ASSERT(zkobj->zko_default_group.attrs);
 
 	zkobj->zko_attr_list[attr_num].name = attr_name;
 	zkobj->zko_attr_list[attr_num].mode = 0444;
-	zkobj->zko_default_attrs[attr_num] = &zkobj->zko_attr_list[attr_num];
+	zkobj->zko_default_group.attrs[attr_num] =
+	    &zkobj->zko_attr_list[attr_num];
 	sysfs_attr_init(&zkobj->zko_attr_list[attr_num]);
 }
 
@@ -175,9 +175,9 @@ zfs_kobj_init(zfs_mod_kobj_t *zkobj, int attr_cnt, int child_cnt,
 			return (ENOMEM);
 	}
 	/* this will always have at least one slot for NULL termination */
-	zkobj->zko_default_attrs = kmem_zalloc(DEFAULT_ATTR_SIZE(attr_cnt),
-	    KM_SLEEP);
-	if (zkobj->zko_default_attrs == NULL) {
+	zkobj->zko_default_group.attrs =
+	    kmem_zalloc(DEFAULT_ATTR_SIZE(attr_cnt), KM_SLEEP);
+	if (zkobj->zko_default_group.attrs == NULL) {
 		if (zkobj->zko_attr_list != NULL) {
 			kmem_free(zkobj->zko_attr_list,
 			    ATTR_TABLE_SIZE(attr_cnt));
@@ -185,14 +185,19 @@ zfs_kobj_init(zfs_mod_kobj_t *zkobj, int attr_cnt, int child_cnt,
 		return (ENOMEM);
 	}
 	zkobj->zko_attr_count = attr_cnt;
-	zkobj->zko_kobj_type.default_attrs = zkobj->zko_default_attrs;
+	zkobj->zko_default_groups[0] = &zkobj->zko_default_group;
+#ifdef HAVE_SYSFS_DEFAULT_GROUPS
+	zkobj->zko_kobj_type.default_groups = zkobj->zko_default_groups;
+#else
+	zkobj->zko_kobj_type.default_attrs = zkobj->zko_default_group.attrs;
+#endif
 
 	if (child_cnt > 0) {
 		zkobj->zko_children = kmem_zalloc(CHILD_TABLE_SIZE(child_cnt),
 		    KM_SLEEP);
 		if (zkobj->zko_children == NULL) {
-			if (zkobj->zko_default_attrs != NULL) {
-				kmem_free(zkobj->zko_default_attrs,
+			if (zkobj->zko_default_group.attrs != NULL) {
+				kmem_free(zkobj->zko_default_group.attrs,
 				    DEFAULT_ATTR_SIZE(attr_cnt));
 			}
 			if (zkobj->zko_attr_list != NULL) {
@@ -214,9 +219,9 @@ zfs_kobj_init(zfs_mod_kobj_t *zkobj, int attr_cnt, int child_cnt,
 static int
 zfs_kobj_add(zfs_mod_kobj_t *zkobj, struct kobject *parent, const char *name)
 {
-	/* zko_default_attrs must be NULL terminated */
-	ASSERT(zkobj->zko_default_attrs != NULL);
-	ASSERT(zkobj->zko_default_attrs[zkobj->zko_attr_count] == NULL);
+	/* zko_default_group.attrs must be NULL terminated */
+	ASSERT(zkobj->zko_default_group.attrs != NULL);
+	ASSERT(zkobj->zko_default_group.attrs[zkobj->zko_attr_count] == NULL);
 
 	kobject_init(&zkobj->zko_kobj, &zkobj->zko_kobj_type);
 	return (kobject_add(&zkobj->zko_kobj, parent, name));

module/os/linux/zfs/zfs_uio.c

@@ -248,7 +248,7 @@ zfs_uio_prefaultpages(ssize_t n, zfs_uio_t *uio)
 			/* touch each page in this segment. */
 			p = iov->iov_base + skip;
 			while (cnt) {
-				if (get_user(tmp, (uint8_t *)p))
+				if (copy_from_user(&tmp, p, 1))
 					return (EFAULT);
 				ulong_t incr = MIN(cnt, PAGESIZE);
 				p += incr;
@@ -256,7 +256,7 @@ zfs_uio_prefaultpages(ssize_t n, zfs_uio_t *uio)
 			}
 			/* touch the last byte in case it straddles a page. */
 			p--;
-			if (get_user(tmp, (uint8_t *)p))
+			if (copy_from_user(&tmp, p, 1))
 				return (EFAULT);
 		}
 	}

module/os/linux/zfs/zfs_vnops_os.c

@@ -3594,7 +3594,11 @@ zfs_putpage(struct inode *ip, struct page *pp, struct writeback_control *wbc)
 			dmu_tx_wait(tx);
 
 		dmu_tx_abort(tx);
+#ifdef HAVE_VFS_FILEMAP_DIRTY_FOLIO
+		filemap_dirty_folio(page_mapping(pp), page_folio(pp));
+#else
 		__set_page_dirty_nobuffers(pp);
+#endif
 		ClearPageError(pp);
 		end_page_writeback(pp);
 		zfs_rangelock_exit(lr);

module/os/linux/zfs/zpl_file.c

@@ -33,9 +33,13 @@
 #include <sys/zfs_vfsops.h>
 #include <sys/zfs_vnops.h>
 #include <sys/zfs_project.h>
-#ifdef HAVE_VFS_SET_PAGE_DIRTY_NOBUFFERS
+#if defined(HAVE_VFS_SET_PAGE_DIRTY_NOBUFFERS) || \
+    defined(HAVE_VFS_FILEMAP_DIRTY_FOLIO)
 #include <linux/pagemap.h>
 #endif
+#ifdef HAVE_VFS_FILEMAP_DIRTY_FOLIO
+#include <linux/writeback.h>
+#endif
 
 /*
  * When using fallocate(2) to preallocate space, inflate the requested
@@ -413,6 +417,8 @@ zpl_aio_write(struct kiocb *kiocb, const struct iovec *iov,
 	if (ret)
 		return (ret);
 
+	kiocb->ki_pos = pos;
+
 	zfs_uio_t uio;
 	zfs_uio_iovec_init(&uio, iov, nr_segs, kiocb->ki_pos, UIO_USERSPACE,
 	    count, 0);
@@ -629,11 +635,19 @@ zpl_readpage_common(struct page *pp)
 	return (error);
 }
 
+#ifdef HAVE_VFS_READ_FOLIO
+static int
+zpl_read_folio(struct file *filp, struct folio *folio)
+{
+	return (zpl_readpage_common(&folio->page));
+}
+#else
 static int
 zpl_readpage(struct file *filp, struct page *pp)
 {
 	return (zpl_readpage_common(pp));
 }
+#endif
 
 static int
 zpl_readpage_filler(void *data, struct page *pp)
@@ -647,12 +661,29 @@ zpl_readpage_filler(void *data, struct page *pp)
  * paging.  For simplicity, the code relies on read_cache_pages() to
  * correctly lock each page for IO and call zpl_readpage().
  */
+#ifdef HAVE_VFS_READPAGES
 static int
 zpl_readpages(struct file *filp, struct address_space *mapping,
     struct list_head *pages, unsigned nr_pages)
 {
 	return (read_cache_pages(mapping, pages, zpl_readpage_filler, NULL));
 }
+#else
+static void
+zpl_readahead(struct readahead_control *ractl)
+{
+	struct page *page;
+
+	while ((page = readahead_page(ractl)) != NULL) {
+		int ret;
+
+		ret = zpl_readpage_filler(NULL, page);
+		put_page(page);
+		if (ret)
+			break;
+	}
+}
+#endif
 
 static int
 zpl_putpage(struct page *pp, struct writeback_control *wbc, void *data)
@@ -764,11 +795,13 @@ zpl_fallocate_common(struct inode *ip, int mode, loff_t offset, loff_t len)
 	if (mode & (test_mode)) {
 		flock64_t bf;
 
-		if (offset > olen)
-			goto out_unmark;
+		if (mode & FALLOC_FL_KEEP_SIZE) {
+			if (offset > olen)
+				goto out_unmark;
 
-		if (offset + len > olen)
-			len = olen - offset;
+			if (offset + len > olen)
+				len = olen - offset;
+		}
 		bf.l_type = F_WRLCK;
 		bf.l_whence = SEEK_SET;
 		bf.l_start = offset;
@@ -1027,14 +1060,25 @@ zpl_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
 
 
 const struct address_space_operations zpl_address_space_operations = {
+#ifdef HAVE_VFS_READPAGES
 	.readpages	= zpl_readpages,
+#else
+	.readahead	= zpl_readahead,
+#endif
+#ifdef HAVE_VFS_READ_FOLIO
+	.read_folio	= zpl_read_folio,
+#else
 	.readpage	= zpl_readpage,
+#endif
 	.writepage	= zpl_writepage,
 	.writepages	= zpl_writepages,
 	.direct_IO	= zpl_direct_IO,
 #ifdef HAVE_VFS_SET_PAGE_DIRTY_NOBUFFERS
 	.set_page_dirty = __set_page_dirty_nobuffers,
 #endif
+#ifdef HAVE_VFS_FILEMAP_DIRTY_FOLIO
+	.dirty_folio	= filemap_dirty_folio,
+#endif
 };
 
 const struct file_operations zpl_file_operations = {

module/os/linux/zfs/zvol_os.c

@@ -903,22 +903,17 @@ zvol_alloc(dev_t dev, const char *name)
 	zso->zvo_disk->major = zvol_major;
 	zso->zvo_disk->events = DISK_EVENT_MEDIA_CHANGE;
 
+	/*
+	 * Setting ZFS_VOLMODE_DEV disables partitioning on ZVOL devices.
+	 * This is accomplished by limiting the number of minors for the
+	 * device to one and explicitly disabling partition scanning.
+	 */
 	if (volmode == ZFS_VOLMODE_DEV) {
-		/*
-		 * ZFS_VOLMODE_DEV disable partitioning on ZVOL devices: set
-		 * gendisk->minors = 1 as noted in include/linux/genhd.h.
-		 * Also disable extended partition numbers (GENHD_FL_EXT_DEVT)
-		 * and suppresses partition scanning (GENHD_FL_NO_PART_SCAN)
-		 * setting gendisk->flags accordingly.
-		 */
 		zso->zvo_disk->minors = 1;
-#if defined(GENHD_FL_EXT_DEVT)
-		zso->zvo_disk->flags &= ~GENHD_FL_EXT_DEVT;
-#endif
-#if defined(GENHD_FL_NO_PART_SCAN)
-		zso->zvo_disk->flags |= GENHD_FL_NO_PART_SCAN;
-#endif
+		zso->zvo_disk->flags &= ~ZFS_GENHD_FL_EXT_DEVT;
+		zso->zvo_disk->flags |= ZFS_GENHD_FL_NO_PART;
 	}
+
 	zso->zvo_disk->first_minor = (dev & MINORMASK);
 	zso->zvo_disk->fops = &zvol_ops;
 	zso->zvo_disk->private_data = zv;
@@ -1055,7 +1050,9 @@ zvol_os_create_minor(const char *name)
 	    (zvol_max_discard_blocks * zv->zv_volblocksize) >> 9);
 	blk_queue_discard_granularity(zv->zv_zso->zvo_queue,
 	    zv->zv_volblocksize);
+#ifdef QUEUE_FLAG_DISCARD
 	blk_queue_flag_set(QUEUE_FLAG_DISCARD, zv->zv_zso->zvo_queue);
+#endif
 #ifdef QUEUE_FLAG_NONROT
 	blk_queue_flag_set(QUEUE_FLAG_NONROT, zv->zv_zso->zvo_queue);
 #endif

module/zfs/arc.c

@@ -3430,7 +3430,6 @@ arc_hdr_realloc_crypt(arc_buf_hdr_t *hdr, boolean_t need_crypt)
 	arc_buf_hdr_t *nhdr;
 	arc_buf_t *buf;
 	kmem_cache_t *ncache, *ocache;
-	unsigned nsize, osize;
 
 	/*
 	 * This function requires that hdr is in the arc_anon state.
@@ -3447,14 +3446,10 @@ arc_hdr_realloc_crypt(arc_buf_hdr_t *hdr, boolean_t need_crypt)
 
 	if (need_crypt) {
 		ncache = hdr_full_crypt_cache;
-		nsize = sizeof (hdr->b_crypt_hdr);
 		ocache = hdr_full_cache;
-		osize = HDR_FULL_SIZE;
 	} else {
 		ncache = hdr_full_cache;
-		nsize = HDR_FULL_SIZE;
 		ocache = hdr_full_crypt_cache;
-		osize = sizeof (hdr->b_crypt_hdr);
 	}
 
 	nhdr = kmem_cache_alloc(ncache, KM_PUSHPAGE);
@@ -9312,26 +9307,37 @@ l2arc_apply_transforms(spa_t *spa, arc_buf_hdr_t *hdr, uint64_t asize,
 	}
 
 	if (compress != ZIO_COMPRESS_OFF && !HDR_COMPRESSION_ENABLED(hdr)) {
-		cabd = abd_alloc_for_io(asize, ismd);
-		tmp = abd_borrow_buf(cabd, asize);
+		/*
+		 * In some cases, we can wind up with size > asize, so
+		 * we need to opt for the larger allocation option here.
+		 *
+		 * (We also need abd_return_buf_copy in all cases because
+		 * it's an ASSERT() to modify the buffer before returning it
+		 * with arc_return_buf(), and all the compressors
+		 * write things before deciding to fail compression in nearly
+		 * every case.)
+		 */
+		cabd = abd_alloc_for_io(size, ismd);
+		tmp = abd_borrow_buf(cabd, size);
 
 		psize = zio_compress_data(compress, to_write, tmp, size,
 		    hdr->b_complevel);
 
-		if (psize >= size) {
-			abd_return_buf(cabd, tmp, asize);
+		if (psize >= asize) {
+			psize = HDR_GET_PSIZE(hdr);
+			abd_return_buf_copy(cabd, tmp, size);
 			HDR_SET_COMPRESS(hdr, ZIO_COMPRESS_OFF);
 			to_write = cabd;
-			abd_copy(to_write, hdr->b_l1hdr.b_pabd, size);
-			if (size != asize)
-				abd_zero_off(to_write, size, asize - size);
+			abd_copy(to_write, hdr->b_l1hdr.b_pabd, psize);
+			if (psize != asize)
+				abd_zero_off(to_write, psize, asize - psize);
 			goto encrypt;
 		}
 		ASSERT3U(psize, <=, HDR_GET_PSIZE(hdr));
 		if (psize < asize)
 			bzero((char *)tmp + psize, asize - psize);
 		psize = HDR_GET_PSIZE(hdr);
-		abd_return_buf_copy(cabd, tmp, asize);
+		abd_return_buf_copy(cabd, tmp, size);
 		to_write = cabd;
 	}
 

module/zfs/dbuf.c

@@ -338,18 +338,18 @@ dbuf_find(objset_t *os, uint64_t obj, uint8_t level, uint64_t blkid)
 	hv = dbuf_hash(os, obj, level, blkid);
 	idx = hv & h->hash_table_mask;
 
-	mutex_enter(DBUF_HASH_MUTEX(h, idx));
+	rw_enter(DBUF_HASH_RWLOCK(h, idx), RW_READER);
 	for (db = h->hash_table[idx]; db != NULL; db = db->db_hash_next) {
 		if (DBUF_EQUAL(db, os, obj, level, blkid)) {
 			mutex_enter(&db->db_mtx);
 			if (db->db_state != DB_EVICTING) {
-				mutex_exit(DBUF_HASH_MUTEX(h, idx));
+				rw_exit(DBUF_HASH_RWLOCK(h, idx));
 				return (db);
 			}
 			mutex_exit(&db->db_mtx);
 		}
 	}
-	mutex_exit(DBUF_HASH_MUTEX(h, idx));
+	rw_exit(DBUF_HASH_RWLOCK(h, idx));
 	return (NULL);
 }
 
@@ -392,13 +392,13 @@ dbuf_hash_insert(dmu_buf_impl_t *db)
 	hv = dbuf_hash(os, obj, level, blkid);
 	idx = hv & h->hash_table_mask;
 
-	mutex_enter(DBUF_HASH_MUTEX(h, idx));
+	rw_enter(DBUF_HASH_RWLOCK(h, idx), RW_WRITER);
 	for (dbf = h->hash_table[idx], i = 0; dbf != NULL;
 	    dbf = dbf->db_hash_next, i++) {
 		if (DBUF_EQUAL(dbf, os, obj, level, blkid)) {
 			mutex_enter(&dbf->db_mtx);
 			if (dbf->db_state != DB_EVICTING) {
-				mutex_exit(DBUF_HASH_MUTEX(h, idx));
+				rw_exit(DBUF_HASH_RWLOCK(h, idx));
 				return (dbf);
 			}
 			mutex_exit(&dbf->db_mtx);
@@ -416,7 +416,7 @@ dbuf_hash_insert(dmu_buf_impl_t *db)
 	mutex_enter(&db->db_mtx);
 	db->db_hash_next = h->hash_table[idx];
 	h->hash_table[idx] = db;
-	mutex_exit(DBUF_HASH_MUTEX(h, idx));
+	rw_exit(DBUF_HASH_RWLOCK(h, idx));
 	uint64_t he = atomic_inc_64_nv(&dbuf_stats.hash_elements.value.ui64);
 	DBUF_STAT_MAX(hash_elements_max, he);
 
@@ -473,13 +473,13 @@ dbuf_hash_remove(dmu_buf_impl_t *db)
 
 	/*
 	 * We mustn't hold db_mtx to maintain lock ordering:
-	 * DBUF_HASH_MUTEX > db_mtx.
+	 * DBUF_HASH_RWLOCK > db_mtx.
 	 */
 	ASSERT(zfs_refcount_is_zero(&db->db_holds));
 	ASSERT(db->db_state == DB_EVICTING);
 	ASSERT(!MUTEX_HELD(&db->db_mtx));
 
-	mutex_enter(DBUF_HASH_MUTEX(h, idx));
+	rw_enter(DBUF_HASH_RWLOCK(h, idx), RW_WRITER);
 	dbp = &h->hash_table[idx];
 	while ((dbf = *dbp) != db) {
 		dbp = &dbf->db_hash_next;
@@ -490,7 +490,7 @@ dbuf_hash_remove(dmu_buf_impl_t *db)
 	if (h->hash_table[idx] &&
 	    h->hash_table[idx]->db_hash_next == NULL)
 		DBUF_STAT_BUMPDOWN(hash_chains);
-	mutex_exit(DBUF_HASH_MUTEX(h, idx));
+	rw_exit(DBUF_HASH_RWLOCK(h, idx));
 	atomic_dec_64(&dbuf_stats.hash_elements.value.ui64);
 }
 
@@ -851,8 +851,8 @@ retry:
 	    sizeof (dmu_buf_impl_t),
 	    0, dbuf_cons, dbuf_dest, NULL, NULL, NULL, 0);
 
-	for (i = 0; i < DBUF_MUTEXES; i++)
-		mutex_init(&h->hash_mutexes[i], NULL, MUTEX_DEFAULT, NULL);
+	for (i = 0; i < DBUF_RWLOCKS; i++)
+		rw_init(&h->hash_rwlocks[i], NULL, RW_DEFAULT, NULL);
 
 	dbuf_stats_init(h);
 
@@ -918,8 +918,8 @@ dbuf_fini(void)
 
 	dbuf_stats_destroy();
 
-	for (i = 0; i < DBUF_MUTEXES; i++)
-		mutex_destroy(&h->hash_mutexes[i]);
+	for (i = 0; i < DBUF_RWLOCKS; i++)
+		rw_destroy(&h->hash_rwlocks[i]);
 #if defined(_KERNEL)
 	/*
 	 * Large allocations which do not require contiguous pages
@@ -1465,10 +1465,8 @@ dbuf_read_impl(dmu_buf_impl_t *db, zio_t *zio, uint32_t flags,
 	zbookmark_phys_t zb;
 	uint32_t aflags = ARC_FLAG_NOWAIT;
 	int err, zio_flags;
-	boolean_t bonus_read;
 
 	err = zio_flags = 0;
-	bonus_read = B_FALSE;
 	DB_DNODE_ENTER(db);
 	dn = DB_DNODE(db);
 	ASSERT(!zfs_refcount_is_zero(&db->db_holds));
@@ -3886,7 +3884,7 @@ dmu_buf_get_user(dmu_buf_t *db_fake)
 }
 
 void
-dmu_buf_user_evict_wait()
+dmu_buf_user_evict_wait(void)
 {
 	taskq_wait(dbu_evict_taskq);
 }

module/zfs/dbuf_stats.c

@@ -137,7 +137,7 @@ dbuf_stats_hash_table_data(char *buf, size_t size, void *data)
 	if (size)
 		buf[0] = 0;
 
-	mutex_enter(DBUF_HASH_MUTEX(h, dsh->idx));
+	rw_enter(DBUF_HASH_RWLOCK(h, dsh->idx), RW_READER);
 	for (db = h->hash_table[dsh->idx]; db != NULL; db = db->db_hash_next) {
 		/*
 		 * Returning ENOMEM will cause the data and header functions
@@ -158,7 +158,7 @@ dbuf_stats_hash_table_data(char *buf, size_t size, void *data)
 
 		mutex_exit(&db->db_mtx);
 	}
-	mutex_exit(DBUF_HASH_MUTEX(h, dsh->idx));
+	rw_exit(DBUF_HASH_RWLOCK(h, dsh->idx));
 
 	return (error);
 }

module/zfs/dmu.c

@@ -73,9 +73,13 @@ int zfs_nopwrite_enabled = 1;
 unsigned long zfs_per_txg_dirty_frees_percent = 5;
 
 /*
- * Enable/disable forcing txg sync when dirty in dmu_offset_next.
+ * Enable/disable forcing txg sync when dirty checking for holes with lseek().
+ * By default this is enabled to ensure accurate hole reporting, it can result
+ * in a significant performance penalty for lseek(SEEK_HOLE) heavy workloads.
+ * Disabling this option will result in holes never being reported in dirty
+ * files which is always safe.
  */
-int zfs_dmu_offset_next_sync = 0;
+int zfs_dmu_offset_next_sync = 1;
 
 /*
  * Limit the amount we can prefetch with one call to this amount.  This
@@ -2110,8 +2114,8 @@ restart:
 		 * If the zfs_dmu_offset_next_sync module option is enabled
 		 * then strict hole reporting has been requested.  Dirty
 		 * dnodes must be synced to disk to accurately report all
-		 * holes.  When disabled (the default) dirty dnodes are
-		 * reported to not have any holes which is always safe.
+		 * holes.  When disabled dirty dnodes are reported to not
+		 * have any holes which is always safe.
 		 *
 		 * When called by zfs_holey_common() the zp->z_rangelock
 		 * is held to prevent zfs_write() and mmap writeback from

module/zfs/dsl_scan.c

@@ -1268,9 +1268,12 @@ dsl_scan_should_clear(dsl_scan_t *scn)
 		mutex_enter(&tvd->vdev_scan_io_queue_lock);
 		queue = tvd->vdev_scan_io_queue;
 		if (queue != NULL) {
-			/* # extents in exts_by_size = # in exts_by_addr */
+			/*
+			 * # of extents in exts_by_size = # in exts_by_addr.
+			 * B-tree efficiency is ~75%, but can be as low as 50%.
+			 */
 			mused += zfs_btree_numnodes(&queue->q_exts_by_size) *
-			    sizeof (range_seg_gap_t) + queue->q_sio_memused;
+			    3 * sizeof (range_seg_gap_t) + queue->q_sio_memused;
 		}
 		mutex_exit(&tvd->vdev_scan_io_queue_lock);
 	}

module/zfs/metaslab.c

@@ -2659,7 +2659,8 @@ metaslab_init(metaslab_group_t *mg, uint64_t id, uint64_t object,
 
 	/*
 	 * We only open space map objects that already exist. All others
-	 * will be opened when we finally allocate an object for it.
+	 * will be opened when we finally allocate an object for it. For
+	 * readonly pools there is no need to open the space map object.
 	 *
 	 * Note:
 	 * When called from vdev_expand(), we can't call into the DMU as
@@ -2668,7 +2669,8 @@ metaslab_init(metaslab_group_t *mg, uint64_t id, uint64_t object,
 	 * that case, the object parameter is zero though, so we won't
 	 * call into the DMU.
 	 */
-	if (object != 0) {
+	if (object != 0 && !(spa->spa_mode == SPA_MODE_READ &&
+	    !spa->spa_read_spacemaps)) {
 		error = space_map_open(&ms->ms_sm, mos, object, ms->ms_start,
 		    ms->ms_size, vd->vdev_ashift);
 

module/zfs/refcount.c

@@ -318,6 +318,14 @@ zfs_refcount_not_held(zfs_refcount_t *rc, const void *holder)
 	return (B_TRUE);
 }
 
+EXPORT_SYMBOL(zfs_refcount_create);
+EXPORT_SYMBOL(zfs_refcount_destroy);
+EXPORT_SYMBOL(zfs_refcount_is_zero);
+EXPORT_SYMBOL(zfs_refcount_count);
+EXPORT_SYMBOL(zfs_refcount_add);
+EXPORT_SYMBOL(zfs_refcount_remove);
+EXPORT_SYMBOL(zfs_refcount_held);
+
 /* BEGIN CSTYLED */
 ZFS_MODULE_PARAM(zfs, ,reference_tracking_enable, INT, ZMOD_RW,
 	"Track reference holders to refcount_t objects");

module/zfs/spa.c

@@ -180,6 +180,12 @@ boolean_t	spa_create_process = B_TRUE;	/* no process ==> no sysdc */
  */
 boolean_t	spa_load_verify_dryrun = B_FALSE;
 
+/*
+ * Allow read spacemaps in case of readonly import (spa_mode == SPA_MODE_READ).
+ * This is used by zdb for spacemaps verification.
+ */
+boolean_t	spa_mode_readable_spacemaps = B_FALSE;
+
 /*
  * This (illegal) pool name is used when temporarily importing a spa_t in order
  * to get the vdev stats associated with the imported devices.
@@ -1241,6 +1247,7 @@ spa_activate(spa_t *spa, spa_mode_t mode)
 
 	spa->spa_state = POOL_STATE_ACTIVE;
 	spa->spa_mode = mode;
+	spa->spa_read_spacemaps = spa_mode_readable_spacemaps;
 
 	spa->spa_normal_class = metaslab_class_create(spa, zfs_metaslab_ops);
 	spa->spa_log_class = metaslab_class_create(spa, zfs_metaslab_ops);

module/zfs/spa_log_spacemap.c

@@ -493,12 +493,6 @@ spa_log_summary_decrement_blkcount(spa_t *spa, uint64_t blocks_gone)
 	for (log_summary_entry_t *e = list_head(&spa->spa_log_summary);
 	    e != NULL; e = list_head(&spa->spa_log_summary)) {
 		if (e->lse_blkcount > blocks_gone) {
-			/*
-			 * Assert that we stopped at an entry that is not
-			 * obsolete.
-			 */
-			ASSERT(e->lse_mscount != 0);
-
 			e->lse_blkcount -= blocks_gone;
 			blocks_gone = 0;
 			break;

module/zfs/vdev.c

@@ -3147,6 +3147,12 @@ vdev_dtl_load(vdev_t *vd)
 	if (vd->vdev_ops->vdev_op_leaf && vd->vdev_dtl_object != 0) {
 		ASSERT(vdev_is_concrete(vd));
 
+		/*
+		 * If the dtl cannot be sync'd there is no need to open it.
+		 */
+		if (spa->spa_mode == SPA_MODE_READ && !spa->spa_read_spacemaps)
+			return (0);
+
 		error = space_map_open(&vd->vdev_dtl_sm, mos,
 		    vd->vdev_dtl_object, 0, -1ULL, 0);
 		if (error)

module/zfs/vdev_indirect.c

@@ -1423,11 +1423,6 @@ vdev_indirect_repair(zio_t *zio)
 {
 	indirect_vsd_t *iv = zio->io_vsd;
 
-	enum zio_flag flags = ZIO_FLAG_IO_REPAIR;
-
-	if (!(zio->io_flags & (ZIO_FLAG_SCRUB | ZIO_FLAG_RESILVER)))
-		flags |= ZIO_FLAG_SELF_HEAL;
-
 	if (!spa_writeable(zio->io_spa))
 		return;
 

module/zfs/vdev_removal.c

@@ -1517,10 +1517,6 @@ spa_vdev_remove_thread(void *arg)
 			 * specified by zfs_removal_suspend_progress. We do this
 			 * solely from the test suite or during debugging.
 			 */
-			uint64_t bytes_copied =
-			    spa->spa_removing_phys.sr_copied;
-			for (int i = 0; i < TXG_SIZE; i++)
-				bytes_copied += svr->svr_bytes_done[i];
 			while (zfs_removal_suspend_progress &&
 			    !svr->svr_thread_exit)
 				delay(hz);

module/zfs/zfs_fm.c

@@ -825,9 +825,6 @@ annotate_ecksum(nvlist_t *ereport, zio_bad_cksum_t *info,
 	const uint64_t *good;
 	const uint64_t *bad;
 
-	uint64_t allset = 0;
-	uint64_t allcleared = 0;
-
 	size_t nui64s = size / sizeof (uint64_t);
 
 	size_t inline_size;
@@ -929,9 +926,6 @@ annotate_ecksum(nvlist_t *ereport, zio_bad_cksum_t *info,
 			// bits set in good, but not in bad
 			cleared = (good[idx] & (~bad[idx]));
 
-			allset |= set;
-			allcleared |= cleared;
-
 			if (!no_inline) {
 				ASSERT3U(offset, <, inline_size);
 				eip->zei_bits_set[offset] = set;

module/zfs/zvol.c

@@ -1353,13 +1353,12 @@ static void
 zvol_rename_minors_impl(const char *oldname, const char *newname)
 {
 	zvol_state_t *zv, *zv_next;
-	int oldnamelen, newnamelen;
+	int oldnamelen;
 
 	if (zvol_inhibit_dev)
 		return;
 
 	oldnamelen = strlen(oldname);
-	newnamelen = strlen(newname);
 
 	rw_enter(&zvol_state_lock, RW_READER);