aboutsummaryrefslogtreecommitdiff
path: root/stand/libsa/zfs
diff options
context:
space:
mode:
Diffstat (limited to 'stand/libsa/zfs')
-rw-r--r--stand/libsa/zfs/Makefile.inc138
-rw-r--r--stand/libsa/zfs/devicename_stubs.c47
-rw-r--r--stand/libsa/zfs/libzfs.h8
-rw-r--r--stand/libsa/zfs/nvlist.c3
-rw-r--r--stand/libsa/zfs/spl/sys/blake3.h17
-rw-r--r--stand/libsa/zfs/spl/sys/zfs_context.h32
-rw-r--r--stand/libsa/zfs/zfs.c201
-rw-r--r--stand/libsa/zfs/zfsimpl.c448
-rw-r--r--stand/libsa/zfs/zstd_shim.c45
9 files changed, 531 insertions, 408 deletions
diff --git a/stand/libsa/zfs/Makefile.inc b/stand/libsa/zfs/Makefile.inc
index 2a06db0d2f5d..2e9d5679f71f 100644
--- a/stand/libsa/zfs/Makefile.inc
+++ b/stand/libsa/zfs/Makefile.inc
@@ -1,96 +1,84 @@
-# $FreeBSD$
-
.PATH: ${ZFSSRC}
.PATH: ${SYSDIR}/crypto/skein
.PATH: ${ZFSOSSRC}/spl
+.PATH: ${OZFS}/module/zstd
.PATH: ${OZFS}/module/zstd/lib/common
.PATH: ${OZFS}/module/zstd/lib/compress
.PATH: ${OZFS}/module/zstd/lib/decompress
-ZFSSRC= zfs.c nvlist.c skein.c skein_block.c list.c
-ZFSSRC+= zstd_shim.c
-ZFSSRC+= entropy_common.c error_private.c
-ZFSSRC+= fse_compress.c fse_decompress.c hist.c
-ZFSSRC+= huf_compress.c huf_decompress.c pool.c xxhash.c
-ZFSSRC+= zstd_common.c zstd_compress.c zstd_compress_literals.c
-ZFSSRC+= zstd_compress_sequences.c zstd_compress_superblock.c
-ZFSSRC+= zstd_ddict.c zstd_decompress.c zstd_decompress_block.c
-ZFSSRC+= zstd_double_fast.c zstd_fast.c zstd_lazy.c zstd_ldm.c
-ZFSSRC+= zstd_opt.c
+.PATH: ${OZFS}/module/icp/asm-aarch64/blake3
+.PATH: ${OZFS}/module/icp/algs/blake3
+ZFS_SRC= zfs.c nvlist.c skein.c skein_block.c list.c
+ZFS_SRC+= zfs_zstd.c
+ZFS_SRC+= blake3.c blake3_generic.c blake3_impl.c
+ZSTD_SRC+= entropy_common.c error_private.c
+ZSTD_SRC+= fse_decompress.c hist.c
+ZSTD_SRC+= huf_decompress.c pool.c xxhash.c
+ZSTD_SRC+= zstd_common.c
+ZSTD_SRC+= zstd_ddict.c zstd_decompress.c zstd_decompress_block.c
+ZSTD_SRC+= zstd_double_fast.c zstd_fast.c zstd_lazy.c zstd_ldm.c
-CFLAGS+= -DHAS_ZSTD_ZFS
-SRCS+= ${ZFSSRC}
+SRCS+= ${ZFS_SRC} ${ZSTD_SRC} ${ZFS_SRC_AS}
-CFLAGS+= -I${LDRSRC}
-CFLAGS+= -I${SYSDIR}/cddl/boot/zfs
-CFLAGS+= -I${SYSDIR}/crypto/skein
+#
+# Any file that needs the FreeBSD overrides that are in
+# include/os/freebssd/spl/XXX needs to have these added to
+# CFLAGS_EARLY.file.c. In general, we try to build out of the OpenZFS tree
+# unaltered. There's a problem, though, that since we're building for a
+# standalone environment that's neither userland nor kernel, we sometimes need
+# special code and that's handled by the 'nested' includes where we either setup
+# something just-so before we include the include/XXX file, or if we need to
+# tweak something defined in that file.
+#
-ZFS_EARLY= -I${ZFSOSINC} \
+ZFS_EARLY= -I${ZFSSRC}/spl \
+ -I${ZFSOSINC} \
-I${ZFSOSINC}/spl \
-I${ZFSOSINC}/zfs
-.for i in ${ZFSSRC}
-CFLAGS.$i+= -include ${ZFSOSINC}/spl/sys/ccompile.h
+#
+# For all files, though, we prepend the sys/ccompile.h file to the build which
+# has a bunch of defines that are present in OpenSolaris / Illumos, but missing
+# from FreeBSD.
+#
+.for i in ${ZFS_SRC} ${ZSTD_SRC}
+CFLAGS.$i+= -include ${ZFSOSINC}/spl/sys/ccompile.h -Wformat -Wall -I${OZFS}/include \
+ -DNEED_SOLARIS_BOOLEAN
.endfor
+CFLAGS_EARLY.blake3.c+= ${ZFS_EARLY} -DOMIT_SIMD
+CFLAGS_EARLY.blake3_generic.c+= ${ZFS_EARLY} -DOMIT_SIMD
+CFLAGS_EARLY.blake3_impl.c+= ${ZFS_EARLY} -DOMIT_SIMD
CFLAGS_EARLY.list.c+= ${ZFS_EARLY}
-CFLAGS_EARLY.zstd_shim.c+= ${ZFS_EARLY}
+CFLAGS_EARLY.zfs_zstd.c+= ${ZFS_EARLY}
+CFLAGS_EARLY.nvlist.c+= ${ZFS_EARLY}
+CFLAGS_EARLY.zfs.c += ${ZFS_EARLY}
-# Can't use the early flags because there's two conflicting definitions of boolean_t in
-# the zfs code that need to be unified.
-CFLAGS.nvlist.c+= -I${ZFSOSINC}/spl
-CFLAGS.zfs.c+= -I${ZFSOSINC}/spl \
+#
+# zfs.c is special: we need to define HAS_ZSTD_ZFS to get zfssubr.c to initialize zstd
+# properly. We need to have the cddl boot compat directory in the include path for zfssubr.c
+# to be found, and we need a couple of other include paths for skein and lz4. Finally we
+# temporarily need LDRSRC to pick up part.h until libsa has a way to look into partitions
+# or enumerate them...
+#
+CFLAGS.zfs.c+= -DHAS_ZSTD_ZFS \
+ -I${SYSDIR}/cddl/boot/zfs \
+ -I${LDRSRC} \
+ -I${SYSDIR}/crypto/skein \
-I${SRCTOP}/sys/cddl/contrib/opensolaris/common/lz4
-CFLAGS.entropy_common.c= -U__BMI__
-CFLAGS.entropy_common.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.error_private.c= -U__BMI__
-CFLAGS.error_private.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.fse_compress.c= -U__BMI__
-CFLAGS.fse_compress.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.fse_decompress.c= -U__BMI__
-CFLAGS.fse_decompress.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.hist.c= -U__BMI__
-CFLAGS.hist.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.huf_compress.c= -U__BMI__
-CFLAGS.huf_compress.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.huf_decompress.c= -U__BMI__
-CFLAGS.huf_decompress.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.pool.c= -U__BMI__
-CFLAGS.pool.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.xxhash.c= -U__BMI__
-CFLAGS.xxhash.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.zstd_common.c= -U__BMI__
-CFLAGS.zstd_common.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.zstd_compress.c= -U__BMI__
-CFLAGS.zstd_compress.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.zstd_compress_literals.c= -U__BMI__
-CFLAGS.zstd_compress_literals.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.zstd_compress_sequences.c= -U__BMI__
-CFLAGS.zstd_compress_sequences.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.zstd_compress_superblock.c= -U__BMI__
-CFLAGS.zstd_compress_superblock.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.zstd_double_fast.c= -U__BMI__
-CFLAGS.zstd_double_fast.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.zstd_fast.c= -U__BMI__
-CFLAGS.zstd_fast.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.zstd_lazy.c= -U__BMI__
-CFLAGS.zstd_lazy.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.zstd_ldm.c= -U__BMI__
-CFLAGS.zstd_ldm.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.zstd_opt.c= -U__BMI__
-CFLAGS.zstd_opt.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.zstd_ddict.c= -U__BMI__
-CFLAGS.zstd_ddict.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.zstd_decompress.c= -U__BMI__
-CFLAGS.zstd_decompress.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.zstd_decompress_block.c= -U__BMI__
-CFLAGS.zstd_decompress_block.c+= ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
-CFLAGS.zstd_shim.c+= -DIN_BASE -DIN_LIBSA -I${OZFS}/include
+
+#
+# ZSTD coding style has some issues, so suppress clang's warnings. Also, zstd's
+# use of BMI instrucitons is broken in this environment, so avoid them.
+#
+.for i in ${ZSTD_SRC}
+CFLAGS.$i+= -U__BMI__ ${NO_WBITWISE_INSTEAD_OF_LOGICAL}
+.endfor
+
+CFLAGS.zfs_zstd.c+= -DIN_BASE -DIN_LIBSA
+
+CFLAGS.blake3_impl.c+= -I${OZFS}/module/icp/algs/blake3 -I${OZFS}/module/icp/include -DIN_LIBSA
# Do not unroll skein loops, reduce code size
CFLAGS.skein_block.c+= -DSKEIN_LOOP=111
-CFLAGS+= -I${SYSDIR}/contrib/openzfs/include
-CFLAGS+= -I${SYSDIR}/contrib/openzfs/include/os/freebsd/zfs
-CFLAGS.zfs.c+= -I${SYSDIR}/cddl/contrib/opensolaris/common/lz4
-
-CFLAGS+= -Wformat -Wall
+CWARNFLAGS.zfs.c+= ${NO_WDANGLING_POINTER}
diff --git a/stand/libsa/zfs/devicename_stubs.c b/stand/libsa/zfs/devicename_stubs.c
deleted file mode 100644
index 41bf907e0cf1..000000000000
--- a/stand/libsa/zfs/devicename_stubs.c
+++ /dev/null
@@ -1,47 +0,0 @@
-/*-
- * Copyright (c) 2012 Andriy Gapon <avg@FreeBSD.org>
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- * 1. Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- * 2. Redistributions in binary form must reproduce the above copyright
- * notice, this list of conditions and the following disclaimer in the
- * documentation and/or other materials provided with the distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
- * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
- * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
- * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
- * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
- * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
- * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
- * SUCH DAMAGE.
- */
-
-#include <sys/cdefs.h>
-__FBSDID("$FreeBSD$");
-
-#include <stand.h>
-#include "libzfs.h"
-
-__attribute__((weak))
-int
-zfs_parsedev(struct zfs_devdesc *dev, const char *devspec, const char **path)
-{
- return (EINVAL);
-}
-
-__attribute__((weak))
-char *
-zfs_fmtdev(void *vdev)
-{
- static char buf[128];
-
- return (buf);
-}
diff --git a/stand/libsa/zfs/libzfs.h b/stand/libsa/zfs/libzfs.h
index e8676c0d53b8..35af864e7e0b 100644
--- a/stand/libsa/zfs/libzfs.h
+++ b/stand/libsa/zfs/libzfs.h
@@ -22,8 +22,6 @@
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
- *
- * $FreeBSD$
*/
#ifndef _BOOT_LIBZFS_H_
@@ -48,10 +46,8 @@ struct zfs_devdesc {
uint64_t root_guid;
};
-int zfs_parsedev(struct zfs_devdesc *dev, const char *devspec,
- const char **path);
-char *zfs_fmtdev(void *vdev);
-int zfs_probe_dev(const char *devname, uint64_t *pool_guid);
+char *zfs_fmtdev(struct devdesc *);
+int zfs_probe_dev(const char *devname, uint64_t *pool_guid, bool part_too);
int zfs_list(const char *name);
int zfs_get_bootonce(void *, const char *, char *, size_t);
int zfs_get_bootenv(void *, nvlist_t **);
diff --git a/stand/libsa/zfs/nvlist.c b/stand/libsa/zfs/nvlist.c
index e5e0abecb274..6cb496a57b26 100644
--- a/stand/libsa/zfs/nvlist.c
+++ b/stand/libsa/zfs/nvlist.c
@@ -23,9 +23,6 @@
* SUCH DAMAGE.
*/
-#include <sys/cdefs.h>
-__FBSDID("$FreeBSD$");
-
#include <sys/param.h>
#include <sys/endian.h>
#include <sys/stdint.h>
diff --git a/stand/libsa/zfs/spl/sys/blake3.h b/stand/libsa/zfs/spl/sys/blake3.h
new file mode 100644
index 000000000000..aa2d3b7e55cc
--- /dev/null
+++ b/stand/libsa/zfs/spl/sys/blake3.h
@@ -0,0 +1,17 @@
+/*
+ * Copyright 2022, Netflix, Inc
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+/*
+ * Gross and ugly hack to cope with upstream's sys/blake3.h not being standalone
+ * safe.
+ */
+#define _KERNEL
+
+#include_next <sys/blake3.h>
+
+#undef _KERNEL
diff --git a/stand/libsa/zfs/spl/sys/zfs_context.h b/stand/libsa/zfs/spl/sys/zfs_context.h
new file mode 100644
index 000000000000..9f12955dd05f
--- /dev/null
+++ b/stand/libsa/zfs/spl/sys/zfs_context.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2022, Netflix, Inc
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+/* TODO: openzfs/include/sys/uio_impl.h must not be included in stand */
+#ifndef _SYS_UIO_IMPL_H
+#define _SYS_UIO_IMPL_H
+#endif
+
+/*
+ * sys/atomic.h must be included after sys/sysmacros.h. The latter includes
+ * machine/atomic.h, which interferes. Sadly, upstream includes them in the
+ * wrong order, so we include it here to fix that.
+ */
+#include <sys/sysmacros.h>
+
+#include_next <sys/zfs_context.h>
+
+#define SYSCTL_HANDLER_ARGS void
+
+/*
+ * Not sure why I need these, but including the canonical stand.h fails because
+ * the normal string.h doesn't like all the other shenanigans in this environment.
+ */
+void *memcpy(void *dst, const void *src, size_t len);
+void *memset(void *dest, int c, size_t len);
+void *memmem(const void *big, size_t big_len, const void *little,
+ size_t little_len);
diff --git a/stand/libsa/zfs/zfs.c b/stand/libsa/zfs/zfs.c
index 633ef3b18784..70a102f6425d 100644
--- a/stand/libsa/zfs/zfs.c
+++ b/stand/libsa/zfs/zfs.c
@@ -22,13 +22,8 @@
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
- *
- * $FreeBSD$
*/
-#include <sys/cdefs.h>
-__FBSDID("$FreeBSD$");
-
/*
* Stand-alone file reading package.
*/
@@ -66,6 +61,9 @@ static void zfs_bootenv_initial(const char *envname, spa_t *spa,
static void zfs_checkpoints_initial(spa_t *spa, const char *name,
const char *dsname);
+static int zfs_parsedev(struct devdesc **idev, const char *devspec,
+ const char **path);
+
struct devsw zfs_dev;
struct fs_ops zfs_fsops = {
@@ -108,7 +106,8 @@ struct zfs_be_entry {
static int
zfs_open(const char *upath, struct open_file *f)
{
- struct zfsmount *mount = (struct zfsmount *)f->f_devdata;
+ struct devdesc *dev = f->f_devdata;
+ struct zfsmount *mount = dev->d_opendata;
struct file *fp;
int rc;
@@ -149,7 +148,8 @@ zfs_close(struct open_file *f)
static int
zfs_read(struct open_file *f, void *start, size_t size, size_t *resid /* out */)
{
- const spa_t *spa = ((struct zfsmount *)f->f_devdata)->spa;
+ struct devdesc *dev = f->f_devdata;
+ const spa_t *spa = ((struct zfsmount *)dev->d_opendata)->spa;
struct file *fp = (struct file *)f->f_fsdata;
struct stat sb;
size_t n;
@@ -213,7 +213,8 @@ zfs_seek(struct open_file *f, off_t offset, int where)
static int
zfs_stat(struct open_file *f, struct stat *sb)
{
- const spa_t *spa = ((struct zfsmount *)f->f_devdata)->spa;
+ struct devdesc *dev = f->f_devdata;
+ const spa_t *spa = ((struct zfsmount *)dev->d_opendata)->spa;
struct file *fp = (struct file *)f->f_fsdata;
return (zfs_dnode_stat(spa, &fp->f_dnode, sb));
@@ -222,7 +223,8 @@ zfs_stat(struct open_file *f, struct stat *sb)
static int
zfs_readdir(struct open_file *f, struct dirent *d)
{
- const spa_t *spa = ((struct zfsmount *)f->f_devdata)->spa;
+ struct devdesc *dev = f->f_devdata;
+ const spa_t *spa = ((struct zfsmount *)dev->d_opendata)->spa;
struct file *fp = (struct file *)f->f_fsdata;
mzap_ent_phys_t mze;
struct stat sb;
@@ -366,59 +368,77 @@ zfs_readdir(struct open_file *f, struct dirent *d)
}
}
+static spa_t *
+spa_find_by_dev(struct zfs_devdesc *dev)
+{
+
+ if (dev->dd.d_dev->dv_type != DEVT_ZFS)
+ return (NULL);
+
+ if (dev->pool_guid == 0)
+ return (STAILQ_FIRST(&zfs_pools));
+
+ return (spa_find_by_guid(dev->pool_guid));
+}
+
/*
* if path is NULL, create mount structure, but do not add it to list.
*/
static int
zfs_mount(const char *dev, const char *path, void **data)
{
- struct zfs_devdesc *zfsdev;
+ struct zfs_devdesc *zfsdev = NULL;
spa_t *spa;
- struct zfsmount *mnt;
+ struct zfsmount *mnt = NULL;
int rv;
errno = 0;
- zfsdev = malloc(sizeof(*zfsdev));
- if (zfsdev == NULL)
- return (errno);
-
- rv = zfs_parsedev(zfsdev, dev + 3, NULL);
+ rv = zfs_parsedev((struct devdesc **)&zfsdev, dev, NULL);
if (rv != 0) {
- free(zfsdev);
return (rv);
}
spa = spa_find_by_dev(zfsdev);
- if (spa == NULL)
- return (ENXIO);
+ if (spa == NULL) {
+ rv = ENXIO;
+ goto err;
+ }
mnt = calloc(1, sizeof(*mnt));
- if (mnt != NULL && path != NULL)
+ if (mnt == NULL) {
+ rv = ENOMEM;
+ goto err;
+ }
+
+ if (mnt->path != NULL) {
mnt->path = strdup(path);
- rv = errno;
+ if (mnt->path == NULL) {
+ rv = ENOMEM;
+ goto err;
+ }
+ }
- if (mnt != NULL)
- rv = zfs_mount_impl(spa, zfsdev->root_guid, mnt);
- free(zfsdev);
+ rv = zfs_mount_impl(spa, zfsdev->root_guid, mnt);
- if (rv == 0 && mnt != NULL && mnt->objset.os_type != DMU_OST_ZFS) {
+ if (rv == 0 && mnt->objset.os_type != DMU_OST_ZFS) {
printf("Unexpected object set type %ju\n",
(uintmax_t)mnt->objset.os_type);
rv = EIO;
}
-
+err:
if (rv != 0) {
if (mnt != NULL)
free(mnt->path);
free(mnt);
+ free(zfsdev);
return (rv);
}
- if (mnt != NULL) {
- *data = mnt;
- if (path != NULL)
- STAILQ_INSERT_TAIL(&zfsmount, mnt, next);
- }
+ *data = mnt;
+ if (path != NULL)
+ STAILQ_INSERT_TAIL(&zfsmount, mnt, next);
+
+ free(zfsdev);
return (rv);
}
@@ -773,35 +793,12 @@ zfs_probe_partition(void *arg, const char *partname,
int
zfs_get_bootenv(void *vdev, nvlist_t **benvp)
{
- struct zfs_devdesc *dev = (struct zfs_devdesc *)vdev;
- nvlist_t *benv = NULL;
- vdev_t *vd;
spa_t *spa;
- if (dev->dd.d_dev->dv_type != DEVT_ZFS)
- return (ENOTSUP);
-
- if ((spa = spa_find_by_dev(dev)) == NULL)
+ if ((spa = spa_find_by_dev((struct zfs_devdesc *)vdev)) == NULL)
return (ENXIO);
- if (spa->spa_bootenv == NULL) {
- STAILQ_FOREACH(vd, &spa->spa_root_vdev->v_children,
- v_childlink) {
- benv = vdev_read_bootenv(vd);
-
- if (benv != NULL)
- break;
- }
- spa->spa_bootenv = benv;
- } else {
- benv = spa->spa_bootenv;
- }
-
- if (benv == NULL)
- return (ENOENT);
-
- *benvp = benv;
- return (0);
+ return (zfs_get_bootenv_spa(spa, benvp));
}
/*
@@ -810,22 +807,12 @@ zfs_get_bootenv(void *vdev, nvlist_t **benvp)
int
zfs_set_bootenv(void *vdev, nvlist_t *benv)
{
- struct zfs_devdesc *dev = (struct zfs_devdesc *)vdev;
spa_t *spa;
- vdev_t *vd;
- if (dev->dd.d_dev->dv_type != DEVT_ZFS)
- return (ENOTSUP);
-
- if ((spa = spa_find_by_dev(dev)) == NULL)
+ if ((spa = spa_find_by_dev((struct zfs_devdesc *)vdev)) == NULL)
return (ENXIO);
- STAILQ_FOREACH(vd, &spa->spa_root_vdev->v_children, v_childlink) {
- vdev_write_bootenv(vd, benv);
- }
-
- spa->spa_bootenv = benv;
- return (0);
+ return (zfs_set_bootenv_spa(spa, benv));
}
/*
@@ -835,27 +822,12 @@ zfs_set_bootenv(void *vdev, nvlist_t *benv)
int
zfs_get_bootonce(void *vdev, const char *key, char *buf, size_t size)
{
- nvlist_t *benv;
- char *result = NULL;
- int result_size, rv;
-
- if ((rv = zfs_get_bootenv(vdev, &benv)) != 0)
- return (rv);
+ spa_t *spa;
- if ((rv = nvlist_find(benv, key, DATA_TYPE_STRING, NULL,
- &result, &result_size)) == 0) {
- if (result_size == 0) {
- /* ignore empty string */
- rv = ENOENT;
- } else {
- size = MIN((size_t)result_size + 1, size);
- strlcpy(buf, result, size);
- }
- (void) nvlist_remove(benv, key, DATA_TYPE_STRING);
- (void) zfs_set_bootenv(vdev, benv);
- }
+ if ((spa = spa_find_by_dev((struct zfs_devdesc *)vdev)) == NULL)
+ return (ENXIO);
- return (rv);
+ return (zfs_get_bootonce_spa(spa, key, buf, size));
}
/*
@@ -1292,8 +1264,12 @@ zfs_nvstore_unset_impl(void *vdev, const char *name, bool unset_env)
rv = zfs_set_bootenv(vdev, spa->spa_bootenv);
}
- if (unset_env)
- env_discard(env_getenv(name));
+ if (unset_env) {
+ struct env_var *ev = env_getenv(name);
+
+ if (ev != NULL)
+ env_discard(ev);
+ }
return (rv);
}
@@ -1490,7 +1466,7 @@ zfs_attach_nvstore(void *vdev)
}
int
-zfs_probe_dev(const char *devname, uint64_t *pool_guid)
+zfs_probe_dev(const char *devname, uint64_t *pool_guid, bool parts_too)
{
struct ptable *table;
struct zfs_probe_args pa;
@@ -1506,6 +1482,8 @@ zfs_probe_dev(const char *devname, uint64_t *pool_guid)
ret = zfs_probe(pa.fd, pool_guid);
if (ret == 0)
return (0);
+ if (!parts_too)
+ return (ENXIO);
/* Probe each partition */
ret = ioctl(pa.fd, DIOCGMEDIASIZE, &mediasz);
@@ -1583,11 +1561,10 @@ zfs_dev_open(struct open_file *f, ...)
rv = 0;
/* This device is not set as currdev, mount us private copy. */
if (mount == NULL)
- rv = zfs_mount(zfs_fmtdev(dev), NULL, (void **)&mount);
+ rv = zfs_mount(devformat(&dev->dd), NULL, (void **)&mount);
if (rv == 0) {
- f->f_devdata = mount;
- free(dev);
+ dev->dd.d_opendata = mount;
}
return (rv);
}
@@ -1595,25 +1572,18 @@ zfs_dev_open(struct open_file *f, ...)
static int
zfs_dev_close(struct open_file *f)
{
+ struct devdesc *dev;
struct zfsmount *mnt, *mount;
- mnt = f->f_devdata;
+ dev = f->f_devdata;
+ mnt = dev->d_opendata;
STAILQ_FOREACH(mount, &zfsmount, next) {
if (mnt->spa->spa_guid == mount->spa->spa_guid)
break;
}
- /*
- * devclose() will free f->f_devdata, but since we do have
- * pointer to zfsmount structure in f->f_devdata, and
- * zfs_unmount() will also free the zfsmount structure,
- * we will get double free. To prevent double free,
- * we must set f_devdata to NULL there.
- */
- if (mount != NULL)
- f->f_devdata = NULL;
-
+ /* XXX */
return (0);
}
@@ -1633,11 +1603,13 @@ struct devsw zfs_dev = {
.dv_close = zfs_dev_close,
.dv_ioctl = noioctl,
.dv_print = zfs_dev_print,
- .dv_cleanup = NULL
+ .dv_cleanup = nullsys,
+ .dv_fmtdev = zfs_fmtdev,
+ .dv_parsedev = zfs_parsedev,
};
-int
-zfs_parsedev(struct zfs_devdesc *dev, const char *devspec, const char **path)
+static int
+zfs_parsedev(struct devdesc **idev, const char *devspec, const char **path)
{
static char rootname[ZFS_MAXNAMELEN];
static char poolname[ZFS_MAXNAMELEN];
@@ -1646,8 +1618,9 @@ zfs_parsedev(struct zfs_devdesc *dev, const char *devspec, const char **path)
const char *np;
const char *sep;
int rv;
+ struct zfs_devdesc *dev;
- np = devspec;
+ np = devspec + 3; /* Skip the leading 'zfs' */
if (*np != ':')
return (EINVAL);
np++;
@@ -1670,18 +1643,24 @@ zfs_parsedev(struct zfs_devdesc *dev, const char *devspec, const char **path)
spa = spa_find_by_name(poolname);
if (!spa)
return (ENXIO);
+ dev = malloc(sizeof(*dev));
+ if (dev == NULL)
+ return (ENOMEM);
dev->pool_guid = spa->spa_guid;
rv = zfs_lookup_dataset(spa, rootname, &dev->root_guid);
- if (rv != 0)
+ if (rv != 0) {
+ free(dev);
return (rv);
+ }
if (path != NULL)
*path = (*end == '\0') ? end : end + 1;
dev->dd.d_dev = &zfs_dev;
+ *idev = &dev->dd;
return (0);
}
char *
-zfs_fmtdev(void *vdev)
+zfs_fmtdev(struct devdesc *vdev)
{
static char rootname[ZFS_MAXNAMELEN];
static char buf[2 * ZFS_MAXNAMELEN + 8];
@@ -1689,7 +1668,7 @@ zfs_fmtdev(void *vdev)
spa_t *spa;
buf[0] = '\0';
- if (dev->dd.d_dev->dv_type != DEVT_ZFS)
+ if (vdev->d_dev->dv_type != DEVT_ZFS)
return (buf);
/* Do we have any pools? */
diff --git a/stand/libsa/zfs/zfsimpl.c b/stand/libsa/zfs/zfsimpl.c
index 865294dafed4..f15d9b016068 100644
--- a/stand/libsa/zfs/zfsimpl.c
+++ b/stand/libsa/zfs/zfsimpl.c
@@ -24,9 +24,6 @@
* SUCH DAMAGE.
*/
-#include <sys/cdefs.h>
-__FBSDID("$FreeBSD$");
-
/*
* Stand-alone ZFS file reader.
*/
@@ -110,36 +107,27 @@ typedef struct indirect_vsd {
} indirect_vsd_t;
/*
- * List of all vdevs, chained through v_alllink.
- */
-static vdev_list_t zfs_vdevs;
-
-/*
- * List of ZFS features supported for read
+ * List of supported read-incompatible ZFS features. Do not add here features
+ * marked as ZFEATURE_FLAG_READONLY_COMPAT, they are irrelevant for read-only!
*/
static const char *features_for_read[] = {
- "org.illumos:lz4_compress",
- "com.delphix:hole_birth",
- "com.delphix:extensible_dataset",
+ "com.datto:bookmark_v2",
+ "com.datto:encryption",
+ "com.delphix:bookmark_written",
+ "com.delphix:device_removal",
"com.delphix:embedded_data",
- "org.open-zfs:large_blocks",
+ "com.delphix:extensible_dataset",
+ "com.delphix:head_errlog",
+ "com.delphix:hole_birth",
+ "com.joyent:multi_vdev_crash_dump",
+ "com.klarasystems:vdev_zaps_v2",
+ "org.freebsd:zstd_compress",
+ "org.illumos:lz4_compress",
"org.illumos:sha512",
"org.illumos:skein",
+ "org.open-zfs:large_blocks",
+ "org.openzfs:blake3",
"org.zfsonlinux:large_dnode",
- "com.joyent:multi_vdev_crash_dump",
- "com.delphix:spacemap_histogram",
- "com.delphix:zpool_checkpoint",
- "com.delphix:spacemap_v2",
- "com.datto:encryption",
- "com.datto:bookmark_v2",
- "org.zfsonlinux:allocation_classes",
- "com.datto:resilver_defer",
- "com.delphix:device_removal",
- "com.delphix:obsolete_counts",
- "com.intel:allocation_classes",
- "org.freebsd:zstd_compress",
- "com.delphix:bookmark_written",
- "com.delphix:head_errlog",
NULL
};
@@ -174,7 +162,6 @@ vdev_indirect_mapping_entry_phys_t *
static void
zfs_init(void)
{
- STAILQ_INIT(&zfs_vdevs);
STAILQ_INIT(&zfs_pools);
dnode_cache_buf = malloc(SPA_MAXBLOCKSIZE);
@@ -446,7 +433,7 @@ vdev_indirect_mapping_entry(vdev_indirect_mapping_t *vim, uint64_t index)
*
* It's possible that the given offset will not be in the mapping table
* (i.e. no mapping entries contain this offset), in which case, the
- * return value value depends on the "next_if_missing" parameter.
+ * return value depends on the "next_if_missing" parameter.
*
* If the offset is not found in the table and "next_if_missing" is
* B_FALSE, then NULL will always be returned. The behavior is intended
@@ -846,16 +833,27 @@ vdev_replacing_read(vdev_t *vdev, const blkptr_t *bp, void *buf,
return (kid->v_read(kid, bp, buf, offset, bytes));
}
+/*
+ * List of vdevs that were fully initialized from their own label, but later a
+ * newer label was found that obsoleted the stale label, freeing its
+ * configuration tree. We keep those vdevs around, since a new configuration
+ * may include them.
+ */
+static vdev_list_t orphans = STAILQ_HEAD_INITIALIZER(orphans);
+
static vdev_t *
-vdev_find(uint64_t guid)
+vdev_find(vdev_list_t *list, uint64_t guid)
{
- vdev_t *vdev;
+ vdev_t *vdev, *safe;
- STAILQ_FOREACH(vdev, &zfs_vdevs, v_alllink)
+ STAILQ_FOREACH_SAFE(vdev, list, v_childlink, safe) {
if (vdev->v_guid == guid)
return (vdev);
+ if ((vdev = vdev_find(&vdev->v_children, guid)) != NULL)
+ return (vdev);
+ }
- return (0);
+ return (NULL);
}
static vdev_t *
@@ -864,6 +862,11 @@ vdev_create(uint64_t guid, vdev_read_t *_read)
vdev_t *vdev;
vdev_indirect_config_t *vic;
+ if ((vdev = vdev_find(&orphans, guid))) {
+ STAILQ_REMOVE(&orphans, vdev, vdev, v_childlink);
+ return (vdev);
+ }
+
vdev = calloc(1, sizeof(vdev_t));
if (vdev != NULL) {
STAILQ_INIT(&vdev->v_children);
@@ -878,7 +881,6 @@ vdev_create(uint64_t guid, vdev_read_t *_read)
if (_read != NULL) {
vic = &vdev->vdev_indirect_config;
vic->vic_prev_indirect_vdev = UINT64_MAX;
- STAILQ_INSERT_TAIL(&zfs_vdevs, vdev, v_alllink);
}
}
@@ -1042,22 +1044,19 @@ vdev_init(uint64_t guid, const nvlist_t *nvlist, vdev_t **vdevp)
* STAILQ_INSERT_AFTER.
*/
static vdev_t *
-vdev_find_previous(vdev_t *top_vdev, vdev_t *vdev)
+vdev_find_previous(vdev_t *top_vdev, uint64_t id)
{
vdev_t *v, *previous;
- if (STAILQ_EMPTY(&top_vdev->v_children))
- return (NULL);
-
previous = NULL;
STAILQ_FOREACH(v, &top_vdev->v_children, v_childlink) {
- if (v->v_id > vdev->v_id)
+ if (v->v_id > id)
return (previous);
- if (v->v_id == vdev->v_id)
+ if (v->v_id == id)
return (v);
- if (v->v_id < vdev->v_id)
+ if (v->v_id < id)
previous = v;
}
return (previous);
@@ -1079,7 +1078,7 @@ vdev_child_count(vdev_t *vdev)
/*
* Insert vdev into top_vdev children list. List is ordered by v_id.
*/
-static void
+static vdev_t *
vdev_insert(vdev_t *top_vdev, vdev_t *vdev)
{
vdev_t *previous;
@@ -1092,7 +1091,7 @@ vdev_insert(vdev_t *top_vdev, vdev_t *vdev)
* so we can use either STAILQ_INSERT_HEAD or STAILQ_INSERT_AFTER
* as STAILQ does not have insert before.
*/
- previous = vdev_find_previous(top_vdev, vdev);
+ previous = vdev_find_previous(top_vdev, vdev->v_id);
if (previous == NULL) {
STAILQ_INSERT_HEAD(&top_vdev->v_children, vdev, v_childlink);
@@ -1101,7 +1100,8 @@ vdev_insert(vdev_t *top_vdev, vdev_t *vdev)
* This vdev was configured from label config,
* do not insert duplicate.
*/
- return;
+ free(vdev);
+ return (previous);
} else {
STAILQ_INSERT_AFTER(&top_vdev->v_children, previous, vdev,
v_childlink);
@@ -1110,24 +1110,28 @@ vdev_insert(vdev_t *top_vdev, vdev_t *vdev)
count = vdev_child_count(top_vdev);
if (top_vdev->v_nchildren < count)
top_vdev->v_nchildren = count;
+ return (vdev);
}
static int
-vdev_from_nvlist(spa_t *spa, uint64_t top_guid, const nvlist_t *nvlist)
+vdev_from_nvlist(spa_t *spa, uint64_t top_guid, uint64_t label_guid,
+ uint64_t txg, const nvlist_t *nvlist)
{
vdev_t *top_vdev, *vdev;
nvlist_t **kids = NULL;
int rc, nkids;
/* Get top vdev. */
- top_vdev = vdev_find(top_guid);
+ top_vdev = vdev_find(&spa->spa_root_vdev->v_children, top_guid);
if (top_vdev == NULL) {
rc = vdev_init(top_guid, nvlist, &top_vdev);
if (rc != 0)
return (rc);
top_vdev->v_spa = spa;
top_vdev->v_top = top_vdev;
- vdev_insert(spa->spa_root_vdev, top_vdev);
+ top_vdev->v_label = label_guid;
+ top_vdev->v_txg = txg;
+ (void )vdev_insert(spa->spa_root_vdev, top_vdev);
}
/* Add children if there are any. */
@@ -1148,7 +1152,7 @@ vdev_from_nvlist(spa_t *spa, uint64_t top_guid, const nvlist_t *nvlist)
vdev->v_spa = spa;
vdev->v_top = top_vdev;
- vdev_insert(top_vdev, vdev);
+ vdev = vdev_insert(top_vdev, vdev);
}
} else {
/*
@@ -1167,28 +1171,6 @@ done:
return (rc);
}
-static int
-vdev_init_from_label(spa_t *spa, const nvlist_t *nvlist)
-{
- uint64_t pool_guid, top_guid;
- nvlist_t *vdevs;
- int rc;
-
- if (nvlist_find(nvlist, ZPOOL_CONFIG_POOL_GUID, DATA_TYPE_UINT64,
- NULL, &pool_guid, NULL) ||
- nvlist_find(nvlist, ZPOOL_CONFIG_TOP_GUID, DATA_TYPE_UINT64,
- NULL, &top_guid, NULL) ||
- nvlist_find(nvlist, ZPOOL_CONFIG_VDEV_TREE, DATA_TYPE_NVLIST,
- NULL, &vdevs, NULL)) {
- printf("ZFS: can't find vdev details\n");
- return (ENOENT);
- }
-
- rc = vdev_from_nvlist(spa, top_guid, vdevs);
- nvlist_destroy(vdevs);
- return (rc);
-}
-
static void
vdev_set_state(vdev_t *vdev)
{
@@ -1235,14 +1217,14 @@ vdev_set_state(vdev_t *vdev)
}
static int
-vdev_update_from_nvlist(uint64_t top_guid, const nvlist_t *nvlist)
+vdev_update_from_nvlist(vdev_t *root, uint64_t top_guid, const nvlist_t *nvlist)
{
vdev_t *vdev;
nvlist_t **kids = NULL;
int rc, nkids;
/* Update top vdev. */
- vdev = vdev_find(top_guid);
+ vdev = vdev_find(&root->v_children, top_guid);
if (vdev != NULL)
vdev_set_initial_state(vdev, nvlist);
@@ -1258,7 +1240,7 @@ vdev_update_from_nvlist(uint64_t top_guid, const nvlist_t *nvlist)
if (rc != 0)
break;
- vdev = vdev_find(guid);
+ vdev = vdev_find(&root->v_children, guid);
if (vdev != NULL)
vdev_set_initial_state(vdev, kids[i]);
}
@@ -1274,6 +1256,19 @@ vdev_update_from_nvlist(uint64_t top_guid, const nvlist_t *nvlist)
return (rc);
}
+static void
+vdev_free(struct vdev *vdev)
+{
+ struct vdev *kid, *safe;
+
+ STAILQ_FOREACH_SAFE(kid, &vdev->v_children, v_childlink, safe)
+ vdev_free(kid);
+ if (vdev->v_phys_read != NULL)
+ STAILQ_INSERT_HEAD(&orphans, vdev, v_childlink);
+ else
+ free(vdev);
+}
+
static int
vdev_init_from_nvlist(spa_t *spa, const nvlist_t *nvlist)
{
@@ -1317,14 +1312,16 @@ vdev_init_from_nvlist(spa_t *spa, const nvlist_t *nvlist)
NULL, &guid, NULL);
if (rc != 0)
break;
- vdev = vdev_find(guid);
+ vdev = vdev_find(&spa->spa_root_vdev->v_children, guid);
/*
* Top level vdev is missing, create it.
+ * XXXGL: how can this happen?
*/
if (vdev == NULL)
- rc = vdev_from_nvlist(spa, guid, kids[i]);
+ rc = vdev_from_nvlist(spa, guid, 0, 0, kids[i]);
else
- rc = vdev_update_from_nvlist(guid, kids[i]);
+ rc = vdev_update_from_nvlist(spa->spa_root_vdev, guid,
+ kids[i]);
if (rc != 0)
break;
}
@@ -1342,6 +1339,53 @@ vdev_init_from_nvlist(spa_t *spa, const nvlist_t *nvlist)
return (rc);
}
+static bool
+nvlist_find_child_guid(const nvlist_t *nvlist, uint64_t guid)
+{
+ nvlist_t **kids = NULL;
+ int nkids, i;
+ bool rv = false;
+
+ if (nvlist_find(nvlist, ZPOOL_CONFIG_CHILDREN, DATA_TYPE_NVLIST_ARRAY,
+ &nkids, &kids, NULL) != 0)
+ nkids = 0;
+
+ for (i = 0; i < nkids; i++) {
+ uint64_t kid_guid;
+
+ if (nvlist_find(kids[i], ZPOOL_CONFIG_GUID, DATA_TYPE_UINT64,
+ NULL, &kid_guid, NULL) != 0)
+ break;
+ if (kid_guid == guid)
+ rv = true;
+ else
+ rv = nvlist_find_child_guid(kids[i], guid);
+ if (rv)
+ break;
+ }
+
+ for (i = 0; i < nkids; i++)
+ nvlist_destroy(kids[i]);
+ free(kids);
+
+ return (rv);
+}
+
+static bool
+nvlist_find_vdev_guid(const nvlist_t *nvlist, uint64_t guid)
+{
+ nvlist_t *vdevs;
+ bool rv;
+
+ if (nvlist_find(nvlist, ZPOOL_CONFIG_VDEV_TREE, DATA_TYPE_NVLIST, NULL,
+ &vdevs, NULL) != 0)
+ return (false);
+ rv = nvlist_find_child_guid(vdevs, guid);
+ nvlist_destroy(vdevs);
+
+ return (rv);
+}
+
static spa_t *
spa_find_by_guid(uint64_t guid)
{
@@ -1367,19 +1411,6 @@ spa_find_by_name(const char *name)
}
static spa_t *
-spa_find_by_dev(struct zfs_devdesc *dev)
-{
-
- if (dev->dd.d_dev->dv_type != DEVT_ZFS)
- return (NULL);
-
- if (dev->pool_guid == 0)
- return (STAILQ_FIRST(&zfs_pools));
-
- return (spa_find_by_guid(dev->pool_guid));
-}
-
-static spa_t *
spa_create(uint64_t guid, const char *name)
{
spa_t *spa;
@@ -1399,7 +1430,7 @@ spa_create(uint64_t guid, const char *name)
free(spa);
return (NULL);
}
- spa->spa_root_vdev->v_name = strdup("root");
+ spa->spa_root_vdev->v_name = spa->spa_name;
STAILQ_INSERT_TAIL(&zfs_pools, spa, spa_link);
return (spa);
@@ -1701,14 +1732,14 @@ static int
vdev_write_bootenv_impl(vdev_t *vdev, vdev_boot_envblock_t *be)
{
vdev_t *kid;
- int rv = 0, rc;
+ int rv = 0, err;
STAILQ_FOREACH(kid, &vdev->v_children, v_childlink) {
if (kid->v_state != VDEV_STATE_HEALTHY)
continue;
- rc = vdev_write_bootenv_impl(kid, be);
- if (rv == 0)
- rv = rc;
+ err = vdev_write_bootenv_impl(kid, be);
+ if (err != 0)
+ rv = err;
}
/*
@@ -1718,12 +1749,12 @@ vdev_write_bootenv_impl(vdev_t *vdev, vdev_boot_envblock_t *be)
return (rv);
for (int l = 0; l < VDEV_LABELS; l++) {
- rc = vdev_label_write(vdev, l, be,
+ err = vdev_label_write(vdev, l, be,
offsetof(vdev_label_t, vl_be));
- if (rc != 0) {
+ if (err != 0) {
printf("failed to write bootenv to %s label %d: %d\n",
- vdev->v_name ? vdev->v_name : "unknown", l, rc);
- rv = rc;
+ vdev->v_name ? vdev->v_name : "unknown", l, err);
+ rv = err;
}
}
return (rv);
@@ -2023,11 +2054,10 @@ vdev_probe(vdev_phys_read_t *_read, vdev_phys_write_t *_write, void *priv,
{
vdev_t vtmp;
spa_t *spa;
- vdev_t *vdev;
- nvlist_t *nvl;
+ vdev_t *vdev, *top;
+ nvlist_t *nvl, *vdevs;
uint64_t val;
- uint64_t guid, vdev_children;
- uint64_t pool_txg, pool_guid;
+ uint64_t guid, pool_guid, top_guid, txg;
const char *pool_name;
int rc, namelen;
@@ -2083,13 +2113,18 @@ vdev_probe(vdev_phys_read_t *_read, vdev_phys_write_t *_write, void *priv,
}
if (nvlist_find(nvl, ZPOOL_CONFIG_POOL_TXG, DATA_TYPE_UINT64,
- NULL, &pool_txg, NULL) != 0 ||
+ NULL, &txg, NULL) != 0 ||
+ txg == 0 ||
+ nvlist_find(nvl, ZPOOL_CONFIG_TOP_GUID, DATA_TYPE_UINT64,
+ NULL, &top_guid, NULL) != 0 ||
nvlist_find(nvl, ZPOOL_CONFIG_POOL_GUID, DATA_TYPE_UINT64,
NULL, &pool_guid, NULL) != 0 ||
nvlist_find(nvl, ZPOOL_CONFIG_POOL_NAME, DATA_TYPE_STRING,
- NULL, &pool_name, &namelen) != 0) {
+ NULL, &pool_name, &namelen) != 0 ||
+ nvlist_find(nvl, ZPOOL_CONFIG_GUID, DATA_TYPE_UINT64,
+ NULL, &guid, NULL) != 0) {
/*
- * Cache and spare devices end up here - just ignore
+ * Cache, spare and replaced devices end up here - just ignore
* them.
*/
nvlist_destroy(nvl);
@@ -2103,8 +2138,6 @@ vdev_probe(vdev_phys_read_t *_read, vdev_phys_write_t *_write, void *priv,
if (spa == NULL) {
char *name;
- nvlist_find(nvl, ZPOOL_CONFIG_VDEV_CHILDREN,
- DATA_TYPE_UINT64, NULL, &vdev_children, NULL);
name = malloc(namelen + 1);
if (name == NULL) {
nvlist_destroy(nvl);
@@ -2118,10 +2151,47 @@ vdev_probe(vdev_phys_read_t *_read, vdev_phys_write_t *_write, void *priv,
nvlist_destroy(nvl);
return (ENOMEM);
}
- spa->spa_root_vdev->v_nchildren = vdev_children;
}
- if (pool_txg > spa->spa_txg)
- spa->spa_txg = pool_txg;
+
+ /*
+ * Check if configuration is already known. If configuration is known
+ * and txg numbers don't match, we got 2x2 scenarios here. First, is
+ * the label being read right now _newer_ than the one read before.
+ * Second, is the vdev that provided the stale label _present_ in the
+ * newer configuration. If neither is true, we completely ignore the
+ * label.
+ */
+ STAILQ_FOREACH(top, &spa->spa_root_vdev->v_children, v_childlink)
+ if (top->v_guid == top_guid) {
+ bool newer, present;
+
+ if (top->v_txg == txg)
+ break;
+ newer = (top->v_txg < txg);
+ present = newer ?
+ nvlist_find_vdev_guid(nvl, top->v_label) :
+ (vdev_find(&top->v_children, guid) != NULL);
+ printf("ZFS: pool %s vdev %s %s stale label from "
+ "0x%jx@0x%jx, %s 0x%jx@0x%jx\n",
+ spa->spa_name, top->v_name,
+ present ? "using" : "ignoring",
+ newer ? top->v_label : guid,
+ newer ? top->v_txg : txg,
+ present ? "referred by" : "using",
+ newer ? guid : top->v_label,
+ newer ? txg : top->v_txg);
+ if (newer) {
+ STAILQ_REMOVE(&spa->spa_root_vdev->v_children,
+ top, vdev, v_childlink);
+ vdev_free(top);
+ break;
+ } else if (present) {
+ break;
+ } else {
+ nvlist_destroy(nvl);
+ return (EIO);
+ }
+ }
/*
* Get the vdev tree and create our in-core copy of it.
@@ -2129,19 +2199,22 @@ vdev_probe(vdev_phys_read_t *_read, vdev_phys_write_t *_write, void *priv,
* be some kind of alias (overlapping slices, dangerously dedicated
* disks etc).
*/
- if (nvlist_find(nvl, ZPOOL_CONFIG_GUID, DATA_TYPE_UINT64,
- NULL, &guid, NULL) != 0) {
- nvlist_destroy(nvl);
- return (EIO);
- }
- vdev = vdev_find(guid);
+ vdev = vdev_find(&spa->spa_root_vdev->v_children, guid);
/* Has this vdev already been inited? */
if (vdev && vdev->v_phys_read) {
nvlist_destroy(nvl);
return (EIO);
}
- rc = vdev_init_from_label(spa, nvl);
+ if (nvlist_find(nvl, ZPOOL_CONFIG_VDEV_TREE, DATA_TYPE_NVLIST, NULL,
+ &vdevs, NULL)) {
+ printf("ZFS: can't find vdev details\n");
+ nvlist_destroy(nvl);
+ return (ENOENT);
+ }
+
+ rc = vdev_from_nvlist(spa, top_guid, guid, txg, vdevs);
+ nvlist_destroy(vdevs);
nvlist_destroy(nvl);
if (rc != 0)
return (rc);
@@ -2150,7 +2223,7 @@ vdev_probe(vdev_phys_read_t *_read, vdev_phys_write_t *_write, void *priv,
* We should already have created an incomplete vdev for this
* vdev. Find it and initialise it with our read proc.
*/
- vdev = vdev_find(guid);
+ vdev = vdev_find(&spa->spa_root_vdev->v_children, guid);
if (vdev != NULL) {
vdev->v_phys_read = _read;
vdev->v_phys_write = _write;
@@ -3067,11 +3140,12 @@ zfs_rlookup(const spa_t *spa, uint64_t objnum, char *result)
char name[256];
char component[256];
uint64_t dir_obj, parent_obj, child_dir_zapobj;
- dnode_phys_t child_dir_zap, dataset, dir, parent;
+ dnode_phys_t child_dir_zap, snapnames_zap, dataset, dir, parent;
dsl_dir_phys_t *dd;
dsl_dataset_phys_t *ds;
char *p;
int len;
+ boolean_t issnap = B_FALSE;
p = &name[sizeof(name) - 1];
*p = '\0';
@@ -3082,6 +3156,8 @@ zfs_rlookup(const spa_t *spa, uint64_t objnum, char *result)
}
ds = (dsl_dataset_phys_t *)&dataset.dn_bonus;
dir_obj = ds->ds_dir_obj;
+ if (ds->ds_snapnames_zapobj == 0)
+ issnap = B_TRUE;
for (;;) {
if (objset_get_dnode(spa, spa->spa_mos, dir_obj, &dir) != 0)
@@ -3097,6 +3173,34 @@ zfs_rlookup(const spa_t *spa, uint64_t objnum, char *result)
&parent) != 0)
return (EIO);
dd = (dsl_dir_phys_t *)&parent.dn_bonus;
+ if (issnap == B_TRUE) {
+ /*
+ * The dataset we are looking up is a snapshot
+ * the dir_obj is the parent already, we don't want
+ * the grandparent just yet. Reset to the parent.
+ */
+ dd = (dsl_dir_phys_t *)&dir.dn_bonus;
+ /* Lookup the dataset to get the snapname ZAP */
+ if (objset_get_dnode(spa, spa->spa_mos,
+ dd->dd_head_dataset_obj, &dataset))
+ return (EIO);
+ ds = (dsl_dataset_phys_t *)&dataset.dn_bonus;
+ if (objset_get_dnode(spa, spa->spa_mos,
+ ds->ds_snapnames_zapobj, &snapnames_zap) != 0)
+ return (EIO);
+ /* Get the name of the snapshot */
+ if (zap_rlookup(spa, &snapnames_zap, component,
+ objnum) != 0)
+ return (EIO);
+ len = strlen(component);
+ p -= len;
+ memcpy(p, component, len);
+ --p;
+ *p = '@';
+ issnap = B_FALSE;
+ continue;
+ }
+
child_dir_zapobj = dd->dd_child_dir_zapobj;
if (objset_get_dnode(spa, spa->spa_mos, child_dir_zapobj,
&child_dir_zap) != 0)
@@ -3126,9 +3230,11 @@ zfs_lookup_dataset(const spa_t *spa, const char *name, uint64_t *objnum)
{
char element[256];
uint64_t dir_obj, child_dir_zapobj;
- dnode_phys_t child_dir_zap, dir;
+ dnode_phys_t child_dir_zap, snapnames_zap, dir, dataset;
dsl_dir_phys_t *dd;
+ dsl_dataset_phys_t *ds;
const char *p, *q;
+ boolean_t issnap = B_FALSE;
if (objset_get_dnode(spa, spa->spa_mos,
DMU_POOL_DIRECTORY_OBJECT, &dir))
@@ -3159,6 +3265,25 @@ zfs_lookup_dataset(const spa_t *spa, const char *name, uint64_t *objnum)
p += strlen(p);
}
+ if (issnap == B_TRUE) {
+ if (objset_get_dnode(spa, spa->spa_mos,
+ dd->dd_head_dataset_obj, &dataset))
+ return (EIO);
+ ds = (dsl_dataset_phys_t *)&dataset.dn_bonus;
+ if (objset_get_dnode(spa, spa->spa_mos,
+ ds->ds_snapnames_zapobj, &snapnames_zap) != 0)
+ return (EIO);
+ /* Actual loop condition #2. */
+ if (zap_lookup(spa, &snapnames_zap, element,
+ sizeof (dir_obj), 1, &dir_obj) != 0)
+ return (ENOENT);
+ *objnum = dir_obj;
+ return (0);
+ } else if ((q = strchr(element, '@')) != NULL) {
+ issnap = B_TRUE;
+ element[q - element] = '\0';
+ p = q + 1;
+ }
child_dir_zapobj = dd->dd_child_dir_zapobj;
if (objset_get_dnode(spa, spa->spa_mos, child_dir_zapobj,
&child_dir_zap) != 0)
@@ -3509,8 +3634,10 @@ zfs_spa_init(spa_t *spa)
return (EIO);
}
rc = load_nvlist(spa, config_object, &nvlist);
- if (rc != 0)
+ if (rc != 0) {
+ printf("ZFS: failed to load pool %s nvlist\n", spa->spa_name);
return (rc);
+ }
rc = zap_lookup(spa, &dir, DMU_POOL_ZPOOL_CHECKPOINT,
sizeof(uint64_t), sizeof(checkpoint) / sizeof(uint64_t),
@@ -3813,3 +3940,82 @@ done:
free(entry);
return (rc);
}
+
+/*
+ * Return either a cached copy of the bootenv, or read each of the vdev children
+ * looking for the bootenv. Cache what's found and return the results. Returns 0
+ * when benvp is filled in, and some errno when not.
+ */
+static int
+zfs_get_bootenv_spa(spa_t *spa, nvlist_t **benvp)
+{
+ vdev_t *vd;
+ nvlist_t *benv = NULL;
+
+ if (spa->spa_bootenv == NULL) {
+ STAILQ_FOREACH(vd, &spa->spa_root_vdev->v_children,
+ v_childlink) {
+ benv = vdev_read_bootenv(vd);
+
+ if (benv != NULL)
+ break;
+ }
+ spa->spa_bootenv = benv;
+ }
+ benv = spa->spa_bootenv;
+
+ if (benv == NULL)
+ return (ENOENT);
+
+ *benvp = benv;
+ return (0);
+}
+
+/*
+ * Store nvlist to pool label bootenv area. Also updates cached pointer in spa.
+ */
+static int
+zfs_set_bootenv_spa(spa_t *spa, nvlist_t *benv)
+{
+ vdev_t *vd;
+
+ STAILQ_FOREACH(vd, &spa->spa_root_vdev->v_children, v_childlink) {
+ vdev_write_bootenv(vd, benv);
+ }
+
+ spa->spa_bootenv = benv;
+ return (0);
+}
+
+/*
+ * Get bootonce value by key. The bootonce <key, value> pair is removed from the
+ * bootenv nvlist and the remaining nvlist is committed back to disk. This process
+ * the bootonce flag since we've reached the point in the boot that we've 'used'
+ * the BE. For chained boot scenarios, we may reach this point multiple times (but
+ * only remove it and return 0 the first time).
+ */
+static int
+zfs_get_bootonce_spa(spa_t *spa, const char *key, char *buf, size_t size)
+{
+ nvlist_t *benv;
+ char *result = NULL;
+ int result_size, rv;
+
+ if ((rv = zfs_get_bootenv_spa(spa, &benv)) != 0)
+ return (rv);
+
+ if ((rv = nvlist_find(benv, key, DATA_TYPE_STRING, NULL,
+ &result, &result_size)) == 0) {
+ if (result_size == 0) {
+ /* ignore empty string */
+ rv = ENOENT;
+ } else if (buf != NULL) {
+ size = MIN((size_t)result_size + 1, size);
+ strlcpy(buf, result, size);
+ }
+ (void)nvlist_remove(benv, key, DATA_TYPE_STRING);
+ (void)zfs_set_bootenv_spa(spa, benv);
+ }
+
+ return (rv);
+}
diff --git a/stand/libsa/zfs/zstd_shim.c b/stand/libsa/zfs/zstd_shim.c
deleted file mode 100644
index b1ba4babcf36..000000000000
--- a/stand/libsa/zfs/zstd_shim.c
+++ /dev/null
@@ -1,45 +0,0 @@
-/*-
- * Copyright (c) 2020 M. Warner Losh <imp@FreeBSD.org>
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- * 1. Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- * 2. Redistributions in binary form must reproduce the above copyright
- * notice, this list of conditions and the following disclaimer in the
- * documentation and/or other materials provided with the distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
- * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
- * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
- * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
- * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
- * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
- * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
- * SUCH DAMAGE.
- *
- */
-#include <sys/cdefs.h>
-__FBSDID("$FreeBSD$");
-
-/*
- * Small amount of shim code needed to get zfs_zstd.c to compile. These items
- * here should all be defined in the SPL or as part of libsa somewhere, but
- * aren't for reasons that haven't been tracked down yet. Ideally, they would
- * all go away and we'd compile zfs_zstd.c directly. Based on an original by
- * Matt Macey, but only the #include remains untouched from that.
- */
-
-#define ZFS_MODULE_PARAM_ARGS void
-typedef int boolean_t; /* This one may be tough to get rid of */
-
-/* TODO: openzfs/include/sys/uio_impl.h must not be included */
-#ifndef _SYS_UIO_IMPL_H
-#define _SYS_UIO_IMPL_H
-#endif
-
-#include <contrib/openzfs/module/zstd/zfs_zstd.c>
generated by cgit v1.2.3 (git 2.25.1) at 2025-09-23 03:21:37 +0000