summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--Makefile2
-rw-r--r--common/Makefile1
-rw-r--r--common/cmd_zfs.c236
-rw-r--r--doc/README.zfs29
-rw-r--r--fs/Makefile1
-rw-r--r--fs/zfs/Makefile47
-rw-r--r--fs/zfs/dev.c137
-rw-r--r--fs/zfs/zfs.c2395
-rw-r--r--fs/zfs/zfs_fletcher.c87
-rw-r--r--fs/zfs/zfs_lzjb.c97
-rw-r--r--fs/zfs/zfs_sha256.c148
-rw-r--r--include/config_cmd_all.h1
-rw-r--r--include/zfs/dmu.h120
-rw-r--r--include/zfs/dmu_objset.h43
-rw-r--r--include/zfs/dnode.h81
-rw-r--r--include/zfs/dsl_dataset.h53
-rw-r--r--include/zfs/dsl_dir.h49
-rw-r--r--include/zfs/sa_impl.h35
-rw-r--r--include/zfs/spa.h292
-rw-r--r--include/zfs/uberblock_impl.h57
-rw-r--r--include/zfs/vdev_impl.h70
-rw-r--r--include/zfs/zap_impl.h111
-rw-r--r--include/zfs/zap_leaf.h103
-rw-r--r--include/zfs/zfs.h122
-rw-r--r--include/zfs/zfs_acl.h55
-rw-r--r--include/zfs/zfs_znode.h71
-rw-r--r--include/zfs/zil.h57
-rw-r--r--include/zfs/zio.h92
-rw-r--r--include/zfs/zio_checksum.h50
-rw-r--r--include/zfs_common.h109
30 files changed, 4750 insertions, 1 deletions
diff --git a/Makefile b/Makefile
index 20fa27220ac..73c8e39abc0 100644
--- a/Makefile
+++ b/Makefile
@@ -247,7 +247,7 @@ endif
LIBS += arch/$(ARCH)/lib/lib$(ARCH).o
LIBS += fs/cramfs/libcramfs.o fs/fat/libfat.o fs/fdos/libfdos.o fs/jffs2/libjffs2.o \
fs/reiserfs/libreiserfs.o fs/ext2/libext2fs.o fs/yaffs2/libyaffs2.o \
- fs/ubifs/libubifs.o
+ fs/ubifs/libubifs.o fs/zfs/libzfs.o
LIBS += net/libnet.o
LIBS += disk/libdisk.o
LIBS += drivers/bios_emulator/libatibiosemu.o
diff --git a/common/Makefile b/common/Makefile
index 483eb4daa30..3d62775dc00 100644
--- a/common/Makefile
+++ b/common/Makefile
@@ -162,6 +162,7 @@ endif
COBJS-$(CONFIG_CMD_XIMG) += cmd_ximg.o
COBJS-$(CONFIG_YAFFS2) += cmd_yaffs2.o
COBJS-$(CONFIG_CMD_SPL) += cmd_spl.o
+COBJS-$(CONFIG_CMD_ZFS) += cmd_zfs.o
# others
ifdef CONFIG_DDR_SPD
diff --git a/common/cmd_zfs.c b/common/cmd_zfs.c
new file mode 100644
index 00000000000..a6ea2c07cdd
--- /dev/null
+++ b/common/cmd_zfs.c
@@ -0,0 +1,236 @@
+/*
+ *
+ * ZFS filesystem porting to Uboot by
+ * Jorgen Lundman <lundman at lundman.net>
+ *
+ * zfsfs support
+ * made from existing GRUB Sources by Sun, GNU and others.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of
+ * the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston,
+ * MA 02111-1307 USA
+ *
+ */
+
+#include <common.h>
+#include <part.h>
+#include <config.h>
+#include <command.h>
+#include <image.h>
+#include <linux/ctype.h>
+#include <asm/byteorder.h>
+#include <zfs_common.h>
+#include <linux/stat.h>
+#include <malloc.h>
+
+#if defined(CONFIG_CMD_USB) && defined(CONFIG_USB_STORAGE)
+#include <usb.h>
+#endif
+
+#if !defined(CONFIG_DOS_PARTITION) && !defined(CONFIG_EFI_PARTITION)
+#error DOS or EFI partition support must be selected
+#endif
+
+#define DOS_PART_MAGIC_OFFSET 0x1fe
+#define DOS_FS_TYPE_OFFSET 0x36
+#define DOS_FS32_TYPE_OFFSET 0x52
+
+static int do_zfs_load(cmd_tbl_t *cmdtp, int flag, int argc, char *argv[])
+{
+ char *filename = NULL;
+ char *ep;
+ int dev;
+ unsigned long part = 1;
+ ulong addr = 0;
+ ulong part_length;
+ disk_partition_t info;
+ char buf[12];
+ unsigned long count;
+ const char *addr_str;
+ struct zfs_file zfile;
+ struct device_s vdev;
+
+ if (argc < 3)
+ return CMD_RET_USAGE;
+
+ count = 0;
+ addr = simple_strtoul(argv[3], NULL, 16);
+ filename = getenv("bootfile");
+ switch (argc) {
+ case 3:
+ addr_str = getenv("loadaddr");
+ if (addr_str != NULL)
+ addr = simple_strtoul(addr_str, NULL, 16);
+ else
+ addr = CONFIG_SYS_LOAD_ADDR;
+
+ break;
+ case 4:
+ break;
+ case 5:
+ filename = argv[4];
+ break;
+ case 6:
+ filename = argv[4];
+ count = simple_strtoul(argv[5], NULL, 16);
+ break;
+
+ default:
+ return cmd_usage(cmdtp);
+ }
+
+ if (!filename) {
+ puts("** No boot file defined **\n");
+ return 1;
+ }
+
+ dev = (int)simple_strtoul(argv[2], &ep, 16);
+ zfs_dev_desc = get_dev(argv[1], dev);
+ if (zfs_dev_desc == NULL) {
+ printf("** Block device %s %d not supported\n", argv[1], dev);
+ return 1;
+ }
+
+ if (*ep) {
+ if (*ep != ':') {
+ puts("** Invalid boot device, use `dev[:part]' **\n");
+ return 1;
+ }
+ part = simple_strtoul(++ep, NULL, 16);
+ }
+
+ if (part != 0) {
+ if (get_partition_info(zfs_dev_desc, part, &info)) {
+ printf("** Bad partition %lu **\n", part);
+ return 1;
+ }
+
+ if (strncmp((char *)info.type, BOOT_PART_TYPE,
+ strlen(BOOT_PART_TYPE)) != 0) {
+ printf("** Invalid partition type \"%s\" (expect \"" BOOT_PART_TYPE "\")\n",
+ info.type);
+ return 1;
+ }
+ printf("Loading file \"%s\" "
+ "from %s device %d:%lu %s\n",
+ filename, argv[1], dev, part, info.name);
+ } else {
+ printf("Loading file \"%s\" from %s device %d\n",
+ filename, argv[1], dev);
+ }
+
+ part_length = zfs_set_blk_dev(zfs_dev_desc, part);
+ if (part_length == 0) {
+ printf("**Bad partition - %s %d:%lu **\n", argv[1], dev, part);
+ return 1;
+ }
+
+ vdev.part_length = part_length;
+
+ memset(&zfile, 0, sizeof(zfile));
+ zfile.device = &vdev;
+ if (zfs_open(&zfile, filename)) {
+ printf("** File not found %s\n", filename);
+ return 1;
+ }
+
+ if ((count < zfile.size) && (count != 0))
+ zfile.size = (uint64_t)count;
+
+ if (zfs_read(&zfile, (char *)addr, zfile.size) != zfile.size) {
+ printf("** Unable to read \"%s\" from %s %d:%lu **\n",
+ filename, argv[1], dev, part);
+ zfs_close(&zfile);
+ return 1;
+ }
+
+ zfs_close(&zfile);
+
+ /* Loading ok, update default load address */
+ load_addr = addr;
+
+ printf("%llu bytes read\n", zfile.size);
+ sprintf(buf, "%llX", zfile.size);
+ setenv("filesize", buf);
+
+ return 0;
+}
+
+
+int zfs_print(const char *entry, const struct zfs_dirhook_info *data)
+{
+ printf("%s %s\n",
+ data->dir ? "<DIR> " : " ",
+ entry);
+ return 0; /* 0 continue, 1 stop */
+}
+
+
+
+static int do_zfs_ls(cmd_tbl_t *cmdtp, int flag, int argc, char *argv[])
+{
+ const char *filename = "/";
+ int dev;
+ unsigned long part = 1;
+ char *ep;
+ int part_length;
+ struct device_s vdev;
+
+ if (argc < 3)
+ return cmd_usage(cmdtp);
+
+ dev = (int)simple_strtoul(argv[2], &ep, 16);
+ zfs_dev_desc = get_dev(argv[1], dev);
+
+ if (zfs_dev_desc == NULL) {
+ printf("\n** Block device %s %d not supported\n", argv[1], dev);
+ return 1;
+ }
+
+ if (*ep) {
+ if (*ep != ':') {
+ puts("\n** Invalid boot device, use `dev[:part]' **\n");
+ return 1;
+ }
+ part = simple_strtoul(++ep, NULL, 16);
+ }
+
+ if (argc == 4)
+ filename = argv[3];
+
+ part_length = zfs_set_blk_dev(zfs_dev_desc, part);
+ if (part_length == 0) {
+ printf("** Bad partition - %s %d:%lu **\n", argv[1], dev, part);
+ return 1;
+ }
+
+ vdev.part_length = part_length;
+
+ zfs_ls(&vdev, filename,
+ zfs_print);
+
+ return 0;
+}
+
+
+U_BOOT_CMD(zfsls, 4, 1, do_zfs_ls,
+ "list files in a directory (default /)",
+ "<interface> <dev[:part]> [directory]\n"
+ " - list files from 'dev' on 'interface' in a '/DATASET/@/$dir/'");
+
+U_BOOT_CMD(zfsload, 6, 0, do_zfs_load,
+ "load binary file from a ZFS filesystem",
+ "<interface> <dev[:part]> [addr] [filename] [bytes]\n"
+ " - load binary file '/DATASET/@/$dir/$file' from 'dev' on 'interface'\n"
+ " to address 'addr' from ZFS filesystem");
diff --git a/doc/README.zfs b/doc/README.zfs
new file mode 100644
index 00000000000..f5998f2f887
--- /dev/null
+++ b/doc/README.zfs
@@ -0,0 +1,29 @@
+This patch series adds support for ZFS listing and load to u-boot.
+
+To Enable zfs ls and load commands, modify the board specific config file with
+#define CONFIG_CMD_ZFS
+
+Steps to test:
+
+1. After applying the patch, zfs specific commands can be seen
+ in the boot loader prompt using
+ UBOOT #help
+
+ zfsload- load binary file from a ZFS file system
+ zfsls - list files in a directory (default /)
+
+2. To list the files in zfs pool, device or partition, execute
+ zfsls <interface> <dev[:part]> [POOL/@/dir/file]
+ For example:
+ UBOOT #zfsls mmc 0:5 /rpool/@/usr/bin/
+
+3. To read and load a file from an ZFS formatted partition to RAM, execute
+ zfsload <interface> <dev[:part]> [addr] [filename] [bytes]
+ For example:
+ UBOOT #zfsload mmc 2:2 0x30007fc0 /rpool/@/boot/uImage
+
+References :
+ -- ZFS GRUB sources from Solaris GRUB-0.97
+ -- GRUB Bazaar repository
+
+Jorgen Lundman <lundman at lundman.net> 2012.
diff --git a/fs/Makefile b/fs/Makefile
index 22aad126bc2..28da76e9591 100644
--- a/fs/Makefile
+++ b/fs/Makefile
@@ -30,6 +30,7 @@ subdirs-$(CONFIG_CMD_JFFS2) += jffs2
subdirs-$(CONFIG_CMD_REISER) += reiserfs
subdirs-$(CONFIG_YAFFS2) += yaffs2
subdirs-$(CONFIG_CMD_UBIFS) += ubifs
+subdirs-$(CONFIG_CMD_ZFS) += zfs
SUBDIRS := $(subdirs-y)
diff --git a/fs/zfs/Makefile b/fs/zfs/Makefile
new file mode 100644
index 00000000000..938fc5e5331
--- /dev/null
+++ b/fs/zfs/Makefile
@@ -0,0 +1,47 @@
+#
+# (C) Copyright 2012
+# Jorgen Lundman <lundman at lundman.net>
+#
+# See file CREDITS for list of people who contributed to this
+# project.
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU General Public License as
+# published by the Free Software Foundation; either version 2 of
+# the License, or (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software
+# Foundation, Inc., 59 Temple Place, Suite 330, Boston,
+# MA 02111-1307 USA
+#
+
+include $(TOPDIR)/config.mk
+
+LIB = $(obj)libzfs.o
+
+AOBJS =
+COBJS-$(CONFIG_CMD_ZFS) := dev.o zfs.o zfs_fletcher.o zfs_sha256.o zfs_lzjb.o
+
+SRCS := $(AOBJS:.o=.S) $(COBJS-y:.o=.c)
+OBJS := $(addprefix $(obj),$(AOBJS) $(COBJS-y))
+
+
+all: $(LIB) $(AOBJS)
+
+$(LIB): $(obj).depend $(OBJS)
+ $(call cmd_link_o_target, $(OBJS))
+
+#########################################################################
+
+# defines $(obj).depend target
+include $(SRCTREE)/rules.mk
+
+sinclude $(obj).depend
+
+#########################################################################
diff --git a/fs/zfs/dev.c b/fs/zfs/dev.c
new file mode 100644
index 00000000000..d68372c8089
--- /dev/null
+++ b/fs/zfs/dev.c
@@ -0,0 +1,137 @@
+/*
+ *
+ * based on code of fs/reiserfs/dev.c by
+ *
+ * (C) Copyright 2003 - 2004
+ * Sysgo AG, <www.elinos.com>, Pavel Bartusek <pba@sysgo.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+
+#include <common.h>
+#include <config.h>
+#include <zfs_common.h>
+
+static block_dev_desc_t *zfs_block_dev_desc;
+static disk_partition_t part_info;
+
+int zfs_set_blk_dev(block_dev_desc_t *rbdd, int part)
+{
+ zfs_block_dev_desc = rbdd;
+
+ if (part == 0) {
+ /* disk doesn't use partition table */
+ part_info.start = 0;
+ part_info.size = rbdd->lba;
+ part_info.blksz = rbdd->blksz;
+ } else {
+ if (get_partition_info(zfs_block_dev_desc, part, &part_info))
+ return 0;
+ }
+
+ return part_info.size;
+}
+
+/* err */
+int zfs_devread(int sector, int byte_offset, int byte_len, char *buf)
+{
+ short sec_buffer[SECTOR_SIZE/sizeof(short)];
+ char *sec_buf = (char *)sec_buffer;
+ unsigned block_len;
+
+ /*
+ * Check partition boundaries
+ */
+ if ((sector < 0) ||
+ ((sector + ((byte_offset + byte_len - 1) >> SECTOR_BITS)) >=
+ part_info.size)) {
+ /* errnum = ERR_OUTSIDE_PART; */
+ printf(" ** zfs_devread() read outside partition sector %d\n", sector);
+ return 1;
+ }
+
+ /*
+ * Get the read to the beginning of a partition.
+ */
+ sector += byte_offset >> SECTOR_BITS;
+ byte_offset &= SECTOR_SIZE - 1;
+
+ debug(" <%d, %d, %d>\n", sector, byte_offset, byte_len);
+
+ if (zfs_block_dev_desc == NULL) {
+ printf("** Invalid Block Device Descriptor (NULL)\n");
+ return 1;
+ }
+
+ if (byte_offset != 0) {
+ /* read first part which isn't aligned with start of sector */
+ if (zfs_block_dev_desc->block_read(zfs_block_dev_desc->dev,
+ part_info.start + sector, 1,
+ (unsigned long *) sec_buf) != 1) {
+ printf(" ** zfs_devread() read error **\n");
+ return 1;
+ }
+ memcpy(buf, sec_buf + byte_offset,
+ min(SECTOR_SIZE - byte_offset, byte_len));
+ buf += min(SECTOR_SIZE - byte_offset, byte_len);
+ byte_len -= min(SECTOR_SIZE - byte_offset, byte_len);
+ sector++;
+ }
+
+ if (byte_len == 0)
+ return 0;
+
+ /* read sector aligned part */
+ block_len = byte_len & ~(SECTOR_SIZE - 1);
+
+ if (block_len == 0) {
+ u8 p[SECTOR_SIZE];
+
+ block_len = SECTOR_SIZE;
+ zfs_block_dev_desc->block_read(zfs_block_dev_desc->dev,
+ part_info.start + sector,
+ 1, (unsigned long *)p);
+ memcpy(buf, p, byte_len);
+ return 0;
+ }
+
+ if (zfs_block_dev_desc->block_read(zfs_block_dev_desc->dev,
+ part_info.start + sector,
+ block_len / SECTOR_SIZE,
+ (unsigned long *) buf) !=
+ block_len / SECTOR_SIZE) {
+ printf(" ** zfs_devread() read error - block\n");
+ return 1;
+ }
+
+ block_len = byte_len & ~(SECTOR_SIZE - 1);
+ buf += block_len;
+ byte_len -= block_len;
+ sector += block_len / SECTOR_SIZE;
+
+ if (byte_len != 0) {
+ /* read rest of data which are not in whole sector */
+ if (zfs_block_dev_desc->
+ block_read(zfs_block_dev_desc->dev,
+ part_info.start + sector, 1,
+ (unsigned long *) sec_buf) != 1) {
+ printf(" ** zfs_devread() read error - last part\n");
+ return 1;
+ }
+ memcpy(buf, sec_buf, byte_len);
+ }
+ return 0;
+}
diff --git a/fs/zfs/zfs.c b/fs/zfs/zfs.c
new file mode 100644
index 00000000000..360f723c846
--- /dev/null
+++ b/fs/zfs/zfs.c
@@ -0,0 +1,2395 @@
+/*
+ *
+ * ZFS filesystem ported to u-boot by
+ * Jorgen Lundman <lundman at lundman.net>
+ *
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004
+ * Free Software Foundation, Inc.
+ * Copyright 2004 Sun Microsystems, Inc.
+ *
+ * GRUB is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * GRUB is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with GRUB. If not, see <http://www.gnu.org/licenses/>.
+ *
+ */
+
+#include <common.h>
+#include <malloc.h>
+#include <linux/stat.h>
+#include <linux/time.h>
+#include <linux/ctype.h>
+#include <asm/byteorder.h>
+#include "zfs_common.h"
+
+block_dev_desc_t *zfs_dev_desc;
+
+/*
+ * The zfs plug-in routines for GRUB are:
+ *
+ * zfs_mount() - locates a valid uberblock of the root pool and reads
+ * in its MOS at the memory address MOS.
+ *
+ * zfs_open() - locates a plain file object by following the MOS
+ * and places its dnode at the memory address DNODE.
+ *
+ * zfs_read() - read in the data blocks pointed by the DNODE.
+ *
+ */
+
+#include <zfs/zfs.h>
+#include <zfs/zio.h>
+#include <zfs/dnode.h>
+#include <zfs/uberblock_impl.h>
+#include <zfs/vdev_impl.h>
+#include <zfs/zio_checksum.h>
+#include <zfs/zap_impl.h>
+#include <zfs/zap_leaf.h>
+#include <zfs/zfs_znode.h>
+#include <zfs/dmu.h>
+#include <zfs/dmu_objset.h>
+#include <zfs/sa_impl.h>
+#include <zfs/dsl_dir.h>
+#include <zfs/dsl_dataset.h>
+
+
+#define ZPOOL_PROP_BOOTFS "bootfs"
+
+
+/*
+ * For nvlist manipulation. (from nvpair.h)
+ */
+#define NV_ENCODE_NATIVE 0
+#define NV_ENCODE_XDR 1
+#define NV_BIG_ENDIAN 0
+#define NV_LITTLE_ENDIAN 1
+#define DATA_TYPE_UINT64 8
+#define DATA_TYPE_STRING 9
+#define DATA_TYPE_NVLIST 19
+#define DATA_TYPE_NVLIST_ARRAY 20
+
+
+/*
+ * Macros to get fields in a bp or DVA.
+ */
+#define P2PHASE(x, align) ((x) & ((align) - 1))
+#define DVA_OFFSET_TO_PHYS_SECTOR(offset) \
+ ((offset + VDEV_LABEL_START_SIZE) >> SPA_MINBLOCKSHIFT)
+
+/*
+ * return x rounded down to an align boundary
+ * eg, P2ALIGN(1200, 1024) == 1024 (1*align)
+ * eg, P2ALIGN(1024, 1024) == 1024 (1*align)
+ * eg, P2ALIGN(0x1234, 0x100) == 0x1200 (0x12*align)
+ * eg, P2ALIGN(0x5600, 0x100) == 0x5600 (0x56*align)
+ */
+#define P2ALIGN(x, align) ((x) & -(align))
+
+/*
+ * FAT ZAP data structures
+ */
+#define ZFS_CRC64_POLY 0xC96C5795D7870F42ULL /* ECMA-182, reflected form */
+#define ZAP_HASH_IDX(hash, n) (((n) == 0) ? 0 : ((hash) >> (64 - (n))))
+#define CHAIN_END 0xffff /* end of the chunk chain */
+
+/*
+ * The amount of space within the chunk available for the array is:
+ * chunk size - space for type (1) - space for next pointer (2)
+ */
+#define ZAP_LEAF_ARRAY_BYTES (ZAP_LEAF_CHUNKSIZE - 3)
+
+#define ZAP_LEAF_HASH_SHIFT(bs) (bs - 5)
+#define ZAP_LEAF_HASH_NUMENTRIES(bs) (1 << ZAP_LEAF_HASH_SHIFT(bs))
+#define LEAF_HASH(bs, h) \
+ ((ZAP_LEAF_HASH_NUMENTRIES(bs)-1) & \
+ ((h) >> (64 - ZAP_LEAF_HASH_SHIFT(bs)-l->l_hdr.lh_prefix_len)))
+
+/*
+ * The amount of space available for chunks is:
+ * block size shift - hash entry size (2) * number of hash
+ * entries - header space (2*chunksize)
+ */
+#define ZAP_LEAF_NUMCHUNKS(bs) \
+ (((1<<bs) - 2*ZAP_LEAF_HASH_NUMENTRIES(bs)) / \
+ ZAP_LEAF_CHUNKSIZE - 2)
+
+/*
+ * The chunks start immediately after the hash table. The end of the
+ * hash table is at l_hash + HASH_NUMENTRIES, which we simply cast to a
+ * chunk_t.
+ */
+#define ZAP_LEAF_CHUNK(l, bs, idx) \
+ ((zap_leaf_chunk_t *)(l->l_hash + ZAP_LEAF_HASH_NUMENTRIES(bs)))[idx]
+#define ZAP_LEAF_ENTRY(l, bs, idx) (&ZAP_LEAF_CHUNK(l, bs, idx).l_entry)
+
+
+/*
+ * Decompression Entry - lzjb
+ */
+#ifndef NBBY
+#define NBBY 8
+#endif
+
+
+
+typedef int zfs_decomp_func_t(void *s_start, void *d_start,
+ uint32_t s_len, uint32_t d_len);
+typedef struct decomp_entry {
+ char *name;
+ zfs_decomp_func_t *decomp_func;
+} decomp_entry_t;
+
+typedef struct dnode_end {
+ dnode_phys_t dn;
+ zfs_endian_t endian;
+} dnode_end_t;
+
+struct zfs_data {
+ /* cache for a file block of the currently zfs_open()-ed file */
+ char *file_buf;
+ uint64_t file_start;
+ uint64_t file_end;
+
+ /* XXX: ashift is per vdev, not per pool. We currently only ever touch
+ * a single vdev, but when/if raid-z or stripes are supported, this
+ * may need revision.
+ */
+ uint64_t vdev_ashift;
+ uint64_t label_txg;
+ uint64_t pool_guid;
+
+ /* cache for a dnode block */
+ dnode_phys_t *dnode_buf;
+ dnode_phys_t *dnode_mdn;
+ uint64_t dnode_start;
+ uint64_t dnode_end;
+ zfs_endian_t dnode_endian;
+
+ uberblock_t current_uberblock;
+
+ dnode_end_t mos;
+ dnode_end_t mdn;
+ dnode_end_t dnode;
+
+ uint64_t vdev_phys_sector;
+
+ int (*userhook)(const char *, const struct zfs_dirhook_info *);
+ struct zfs_dirhook_info *dirinfo;
+
+};
+
+
+
+
+static int
+zlib_decompress(void *s, void *d,
+ uint32_t slen, uint32_t dlen)
+{
+ if (zlib_decompress(s, d, slen, dlen) < 0)
+ return ZFS_ERR_BAD_FS;
+ return ZFS_ERR_NONE;
+}
+
+static decomp_entry_t decomp_table[ZIO_COMPRESS_FUNCTIONS] = {
+ {"inherit", NULL}, /* ZIO_COMPRESS_INHERIT */
+ {"on", lzjb_decompress}, /* ZIO_COMPRESS_ON */
+ {"off", NULL}, /* ZIO_COMPRESS_OFF */
+ {"lzjb", lzjb_decompress}, /* ZIO_COMPRESS_LZJB */
+ {"empty", NULL}, /* ZIO_COMPRESS_EMPTY */
+ {"gzip-1", zlib_decompress}, /* ZIO_COMPRESS_GZIP1 */
+ {"gzip-2", zlib_decompress}, /* ZIO_COMPRESS_GZIP2 */
+ {"gzip-3", zlib_decompress}, /* ZIO_COMPRESS_GZIP3 */
+ {"gzip-4", zlib_decompress}, /* ZIO_COMPRESS_GZIP4 */
+ {"gzip-5", zlib_decompress}, /* ZIO_COMPRESS_GZIP5 */
+ {"gzip-6", zlib_decompress}, /* ZIO_COMPRESS_GZIP6 */
+ {"gzip-7", zlib_decompress}, /* ZIO_COMPRESS_GZIP7 */
+ {"gzip-8", zlib_decompress}, /* ZIO_COMPRESS_GZIP8 */
+ {"gzip-9", zlib_decompress}, /* ZIO_COMPRESS_GZIP9 */
+};
+
+
+
+static int zio_read_data(blkptr_t *bp, zfs_endian_t endian,
+ void *buf, struct zfs_data *data);
+
+static int
+zio_read(blkptr_t *bp, zfs_endian_t endian, void **buf,
+ size_t *size, struct zfs_data *data);
+
+/*
+ * Our own version of log2(). Same thing as highbit()-1.
+ */
+static int
+zfs_log2(uint64_t num)
+{
+ int i = 0;
+
+ while (num > 1) {
+ i++;
+ num = num >> 1;
+ }
+
+ return i;
+}
+
+
+/* Checksum Functions */
+static void
+zio_checksum_off(const void *buf __attribute__ ((unused)),
+ uint64_t size __attribute__ ((unused)),
+ zfs_endian_t endian __attribute__ ((unused)),
+ zio_cksum_t *zcp)
+{
+ ZIO_SET_CHECKSUM(zcp, 0, 0, 0, 0);
+}
+
+/* Checksum Table and Values */
+static zio_checksum_info_t zio_checksum_table[ZIO_CHECKSUM_FUNCTIONS] = {
+ {NULL, 0, 0, "inherit"},
+ {NULL, 0, 0, "on"},
+ {zio_checksum_off, 0, 0, "off"},
+ {zio_checksum_SHA256, 1, 1, "label"},
+ {zio_checksum_SHA256, 1, 1, "gang_header"},
+ {NULL, 0, 0, "zilog"},
+ {fletcher_2_endian, 0, 0, "fletcher2"},
+ {fletcher_4_endian, 1, 0, "fletcher4"},
+ {zio_checksum_SHA256, 1, 0, "SHA256"},
+ {NULL, 0, 0, "zilog2"},
+};
+
+/*
+ * zio_checksum_verify: Provides support for checksum verification.
+ *
+ * Fletcher2, Fletcher4, and SHA256 are supported.
+ *
+ */
+static int
+zio_checksum_verify(zio_cksum_t zc, uint32_t checksum,
+ zfs_endian_t endian, char *buf, int size)
+{
+ zio_eck_t *zec = (zio_eck_t *) (buf + size) - 1;
+ zio_checksum_info_t *ci = &zio_checksum_table[checksum];
+ zio_cksum_t actual_cksum, expected_cksum;
+
+ if (checksum >= ZIO_CHECKSUM_FUNCTIONS || ci->ci_func == NULL) {
+ printf("zfs unknown checksum function %d\n", checksum);
+ return ZFS_ERR_NOT_IMPLEMENTED_YET;
+ }
+
+ if (ci->ci_eck) {
+ expected_cksum = zec->zec_cksum;
+ zec->zec_cksum = zc;
+ ci->ci_func(buf, size, endian, &actual_cksum);
+ zec->zec_cksum = expected_cksum;
+ zc = expected_cksum;
+ } else {
+ ci->ci_func(buf, size, endian, &actual_cksum);
+ }
+
+ if ((actual_cksum.zc_word[0] != zc.zc_word[0])
+ || (actual_cksum.zc_word[1] != zc.zc_word[1])
+ || (actual_cksum.zc_word[2] != zc.zc_word[2])
+ || (actual_cksum.zc_word[3] != zc.zc_word[3])) {
+ return ZFS_ERR_BAD_FS;
+ }
+
+ return ZFS_ERR_NONE;
+}
+
+/*
+ * vdev_uberblock_compare takes two uberblock structures and returns an integer
+ * indicating the more recent of the two.
+ * Return Value = 1 if ub2 is more recent
+ * Return Value = -1 if ub1 is more recent
+ * The most recent uberblock is determined using its transaction number and
+ * timestamp. The uberblock with the highest transaction number is
+ * considered "newer". If the transaction numbers of the two blocks match, the
+ * timestamps are compared to determine the "newer" of the two.
+ */
+static int
+vdev_uberblock_compare(uberblock_t *ub1, uberblock_t *ub2)
+{
+ zfs_endian_t ub1_endian, ub2_endian;
+ if (zfs_to_cpu64(ub1->ub_magic, LITTLE_ENDIAN) == UBERBLOCK_MAGIC)
+ ub1_endian = LITTLE_ENDIAN;
+ else
+ ub1_endian = BIG_ENDIAN;
+ if (zfs_to_cpu64(ub2->ub_magic, LITTLE_ENDIAN) == UBERBLOCK_MAGIC)
+ ub2_endian = LITTLE_ENDIAN;
+ else
+ ub2_endian = BIG_ENDIAN;
+
+ if (zfs_to_cpu64(ub1->ub_txg, ub1_endian)
+ < zfs_to_cpu64(ub2->ub_txg, ub2_endian))
+ return -1;
+ if (zfs_to_cpu64(ub1->ub_txg, ub1_endian)
+ > zfs_to_cpu64(ub2->ub_txg, ub2_endian))
+ return 1;
+
+ if (zfs_to_cpu64(ub1->ub_timestamp, ub1_endian)
+ < zfs_to_cpu64(ub2->ub_timestamp, ub2_endian))
+ return -1;
+ if (zfs_to_cpu64(ub1->ub_timestamp, ub1_endian)
+ > zfs_to_cpu64(ub2->ub_timestamp, ub2_endian))
+ return 1;
+
+ return 0;
+}
+
+/*
+ * Three pieces of information are needed to verify an uberblock: the magic
+ * number, the version number, and the checksum.
+ *
+ * Currently Implemented: version number, magic number, label txg
+ * Need to Implement: checksum
+ *
+ */
+static int
+uberblock_verify(uberblock_t *uber, int offset, struct zfs_data *data)
+{
+ int err;
+ zfs_endian_t endian = UNKNOWN_ENDIAN;
+ zio_cksum_t zc;
+
+ if (uber->ub_txg < data->label_txg) {
+ debug("ignoring partially written label: uber_txg < label_txg %llu %llu\n",
+ uber->ub_txg, data->label_txg);
+ return ZFS_ERR_BAD_FS;
+ }
+
+ if (zfs_to_cpu64(uber->ub_magic, LITTLE_ENDIAN) == UBERBLOCK_MAGIC
+ && zfs_to_cpu64(uber->ub_version, LITTLE_ENDIAN) > 0
+ && zfs_to_cpu64(uber->ub_version, LITTLE_ENDIAN) <= SPA_VERSION)
+ endian = LITTLE_ENDIAN;
+
+ if (zfs_to_cpu64(uber->ub_magic, BIG_ENDIAN) == UBERBLOCK_MAGIC
+ && zfs_to_cpu64(uber->ub_version, BIG_ENDIAN) > 0
+ && zfs_to_cpu64(uber->ub_version, BIG_ENDIAN) <= SPA_VERSION)
+ endian = BIG_ENDIAN;
+
+ if (endian == UNKNOWN_ENDIAN) {
+ printf("invalid uberblock magic\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ memset(&zc, 0, sizeof(zc));
+ zc.zc_word[0] = cpu_to_zfs64(offset, endian);
+ err = zio_checksum_verify(zc, ZIO_CHECKSUM_LABEL, endian,
+ (char *) uber, UBERBLOCK_SIZE(data->vdev_ashift));
+
+ if (!err) {
+ /* Check that the data pointed by the rootbp is usable. */
+ void *osp = NULL;
+ size_t ospsize;
+ err = zio_read(&uber->ub_rootbp, endian, &osp, &ospsize, data);
+ free(osp);
+
+ if (!err && ospsize < OBJSET_PHYS_SIZE_V14) {
+ printf("uberblock rootbp points to invalid data\n");
+ return ZFS_ERR_BAD_FS;
+ }
+ }
+
+ return err;
+}
+
+/*
+ * Find the best uberblock.
+ * Return:
+ * Success - Pointer to the best uberblock.
+ * Failure - NULL
+ */
+static uberblock_t *find_bestub(char *ub_array, struct zfs_data *data)
+{
+ const uint64_t sector = data->vdev_phys_sector;
+ uberblock_t *ubbest = NULL;
+ uberblock_t *ubnext;
+ unsigned int i, offset, pickedub = 0;
+ int err = ZFS_ERR_NONE;
+
+ const unsigned int UBCOUNT = UBERBLOCK_COUNT(data->vdev_ashift);
+ const uint64_t UBBYTES = UBERBLOCK_SIZE(data->vdev_ashift);
+
+ for (i = 0; i < UBCOUNT; i++) {
+ ubnext = (uberblock_t *) (i * UBBYTES + ub_array);
+ offset = (sector << SPA_MINBLOCKSHIFT) + VDEV_PHYS_SIZE + (i * UBBYTES);
+
+ err = uberblock_verify(ubnext, offset, data);
+ if (err)
+ continue;
+
+ if (ubbest == NULL || vdev_uberblock_compare(ubnext, ubbest) > 0) {
+ ubbest = ubnext;
+ pickedub = i;
+ }
+ }
+
+ if (ubbest)
+ debug("zfs Found best uberblock at idx %d, txg %llu\n",
+ pickedub, (unsigned long long) ubbest->ub_txg);
+
+ return ubbest;
+}
+
+static inline size_t
+get_psize(blkptr_t *bp, zfs_endian_t endian)
+{
+ return (((zfs_to_cpu64((bp)->blk_prop, endian) >> 16) & 0xffff) + 1)
+ << SPA_MINBLOCKSHIFT;
+}
+
+static uint64_t
+dva_get_offset(dva_t *dva, zfs_endian_t endian)
+{
+ return zfs_to_cpu64((dva)->dva_word[1],
+ endian) << SPA_MINBLOCKSHIFT;
+}
+
+/*
+ * Read a block of data based on the gang block address dva,
+ * and put its data in buf.
+ *
+ */
+static int
+zio_read_gang(blkptr_t *bp, zfs_endian_t endian, dva_t *dva, void *buf,
+ struct zfs_data *data)
+{
+ zio_gbh_phys_t *zio_gb;
+ uint64_t offset, sector;
+ unsigned i;
+ int err;
+ zio_cksum_t zc;
+
+ memset(&zc, 0, sizeof(zc));
+
+ zio_gb = malloc(SPA_GANGBLOCKSIZE);
+ if (!zio_gb)
+ return ZFS_ERR_OUT_OF_MEMORY;
+
+ offset = dva_get_offset(dva, endian);
+ sector = DVA_OFFSET_TO_PHYS_SECTOR(offset);
+
+ /* read in the gang block header */
+ err = zfs_devread(sector, 0, SPA_GANGBLOCKSIZE, (char *) zio_gb);
+
+ if (err) {
+ free(zio_gb);
+ return err;
+ }
+
+ /* XXX */
+ /* self checksuming the gang block header */
+ ZIO_SET_CHECKSUM(&zc, DVA_GET_VDEV(dva),
+ dva_get_offset(dva, endian), bp->blk_birth, 0);
+ err = zio_checksum_verify(zc, ZIO_CHECKSUM_GANG_HEADER, endian,
+ (char *) zio_gb, SPA_GANGBLOCKSIZE);
+ if (err) {
+ free(zio_gb);
+ return err;
+ }
+
+ endian = (zfs_to_cpu64(bp->blk_prop, endian) >> 63) & 1;
+
+ for (i = 0; i < SPA_GBH_NBLKPTRS; i++) {
+ if (zio_gb->zg_blkptr[i].blk_birth == 0)
+ continue;
+
+ err = zio_read_data(&zio_gb->zg_blkptr[i], endian, buf, data);
+ if (err) {
+ free(zio_gb);
+ return err;
+ }
+ buf = (char *) buf + get_psize(&zio_gb->zg_blkptr[i], endian);
+ }
+ free(zio_gb);
+ return ZFS_ERR_NONE;
+}
+
+/*
+ * Read in a block of raw data to buf.
+ */
+static int
+zio_read_data(blkptr_t *bp, zfs_endian_t endian, void *buf,
+ struct zfs_data *data)
+{
+ int i, psize;
+ int err = ZFS_ERR_NONE;
+
+ psize = get_psize(bp, endian);
+
+ /* pick a good dva from the block pointer */
+ for (i = 0; i < SPA_DVAS_PER_BP; i++) {
+ uint64_t offset, sector;
+
+ if (bp->blk_dva[i].dva_word[0] == 0 && bp->blk_dva[i].dva_word[1] == 0)
+ continue;
+
+ if ((zfs_to_cpu64(bp->blk_dva[i].dva_word[1], endian)>>63) & 1) {
+ err = zio_read_gang(bp, endian, &bp->blk_dva[i], buf, data);
+ } else {
+ /* read in a data block */
+ offset = dva_get_offset(&bp->blk_dva[i], endian);
+ sector = DVA_OFFSET_TO_PHYS_SECTOR(offset);
+
+ err = zfs_devread(sector, 0, psize, buf);
+ }
+
+ if (!err) {
+ /*Check the underlying checksum before we rule this DVA as "good"*/
+ uint32_t checkalgo = (zfs_to_cpu64((bp)->blk_prop, endian) >> 40) & 0xff;
+
+ err = zio_checksum_verify(bp->blk_cksum, checkalgo, endian, buf, psize);
+ if (!err)
+ return ZFS_ERR_NONE;
+ }
+
+ /* If read failed or checksum bad, reset the error. Hopefully we've got some more DVA's to try.*/
+ }
+
+ if (!err) {
+ printf("couldn't find a valid DVA\n");
+ err = ZFS_ERR_BAD_FS;
+ }
+
+ return err;
+}
+
+/*
+ * Read in a block of data, verify its checksum, decompress if needed,
+ * and put the uncompressed data in buf.
+ */
+static int
+zio_read(blkptr_t *bp, zfs_endian_t endian, void **buf,
+ size_t *size, struct zfs_data *data)
+{
+ size_t lsize, psize;
+ unsigned int comp;
+ char *compbuf = NULL;
+ int err;
+
+ *buf = NULL;
+
+ comp = (zfs_to_cpu64((bp)->blk_prop, endian)>>32) & 0xff;
+ lsize = (BP_IS_HOLE(bp) ? 0 :
+ (((zfs_to_cpu64((bp)->blk_prop, endian) & 0xffff) + 1)
+ << SPA_MINBLOCKSHIFT));
+ psize = get_psize(bp, endian);
+
+ if (size)
+ *size = lsize;
+
+ if (comp >= ZIO_COMPRESS_FUNCTIONS) {
+ printf("compression algorithm %u not supported\n", (unsigned int) comp);
+ return ZFS_ERR_NOT_IMPLEMENTED_YET;
+ }
+
+ if (comp != ZIO_COMPRESS_OFF && decomp_table[comp].decomp_func == NULL) {
+ printf("compression algorithm %s not supported\n", decomp_table[comp].name);
+ return ZFS_ERR_NOT_IMPLEMENTED_YET;
+ }
+
+ if (comp != ZIO_COMPRESS_OFF) {
+ compbuf = malloc(psize);
+ if (!compbuf)
+ return ZFS_ERR_OUT_OF_MEMORY;
+ } else {
+ compbuf = *buf = malloc(lsize);
+ }
+
+ err = zio_read_data(bp, endian, compbuf, data);
+ if (err) {
+ free(compbuf);
+ *buf = NULL;
+ return err;
+ }
+
+ if (comp != ZIO_COMPRESS_OFF) {
+ *buf = malloc(lsize);
+ if (!*buf) {
+ free(compbuf);
+ return ZFS_ERR_OUT_OF_MEMORY;
+ }
+
+ err = decomp_table[comp].decomp_func(compbuf, *buf, psize, lsize);
+ free(compbuf);
+ if (err) {
+ free(*buf);
+ *buf = NULL;
+ return err;
+ }
+ }
+
+ return ZFS_ERR_NONE;
+}
+
+/*
+ * Get the block from a block id.
+ * push the block onto the stack.
+ *
+ */
+static int
+dmu_read(dnode_end_t *dn, uint64_t blkid, void **buf,
+ zfs_endian_t *endian_out, struct zfs_data *data)
+{
+ int idx, level;
+ blkptr_t *bp_array = dn->dn.dn_blkptr;
+ int epbs = dn->dn.dn_indblkshift - SPA_BLKPTRSHIFT;
+ blkptr_t *bp;
+ void *tmpbuf = 0;
+ zfs_endian_t endian;
+ int err = ZFS_ERR_NONE;
+
+ bp = malloc(sizeof(blkptr_t));
+ if (!bp)
+ return ZFS_ERR_OUT_OF_MEMORY;
+
+ endian = dn->endian;
+ for (level = dn->dn.dn_nlevels - 1; level >= 0; level--) {
+ idx = (blkid >> (epbs * level)) & ((1 << epbs) - 1);
+ *bp = bp_array[idx];
+ if (bp_array != dn->dn.dn_blkptr) {
+ free(bp_array);
+ bp_array = 0;
+ }
+
+ if (BP_IS_HOLE(bp)) {
+ size_t size = zfs_to_cpu16(dn->dn.dn_datablkszsec,
+ dn->endian)
+ << SPA_MINBLOCKSHIFT;
+ *buf = malloc(size);
+ if (*buf) {
+ err = ZFS_ERR_OUT_OF_MEMORY;
+ break;
+ }
+ memset(*buf, 0, size);
+ endian = (zfs_to_cpu64(bp->blk_prop, endian) >> 63) & 1;
+ break;
+ }
+ if (level == 0) {
+ err = zio_read(bp, endian, buf, 0, data);
+ endian = (zfs_to_cpu64(bp->blk_prop, endian) >> 63) & 1;
+ break;
+ }
+ err = zio_read(bp, endian, &tmpbuf, 0, data);
+ endian = (zfs_to_cpu64(bp->blk_prop, endian) >> 63) & 1;
+ if (err)
+ break;
+ bp_array = tmpbuf;
+ }
+ if (bp_array != dn->dn.dn_blkptr)
+ free(bp_array);
+ if (endian_out)
+ *endian_out = endian;
+
+ free(bp);
+ return err;
+}
+
+/*
+ * mzap_lookup: Looks up property described by "name" and returns the value
+ * in "value".
+ */
+static int
+mzap_lookup(mzap_phys_t *zapobj, zfs_endian_t endian,
+ int objsize, char *name, uint64_t * value)
+{
+ int i, chunks;
+ mzap_ent_phys_t *mzap_ent = zapobj->mz_chunk;
+
+ chunks = objsize / MZAP_ENT_LEN - 1;
+ for (i = 0; i < chunks; i++) {
+ if (strcmp(mzap_ent[i].mze_name, name) == 0) {
+ *value = zfs_to_cpu64(mzap_ent[i].mze_value, endian);
+ return ZFS_ERR_NONE;
+ }
+ }
+
+ printf("couldn't find '%s'\n", name);
+ return ZFS_ERR_FILE_NOT_FOUND;
+}
+
+static int
+mzap_iterate(mzap_phys_t *zapobj, zfs_endian_t endian, int objsize,
+ int (*hook)(const char *name,
+ uint64_t val,
+ struct zfs_data *data),
+ struct zfs_data *data)
+{
+ int i, chunks;
+ mzap_ent_phys_t *mzap_ent = zapobj->mz_chunk;
+
+ chunks = objsize / MZAP_ENT_LEN - 1;
+ for (i = 0; i < chunks; i++) {
+ if (hook(mzap_ent[i].mze_name,
+ zfs_to_cpu64(mzap_ent[i].mze_value, endian),
+ data))
+ return 1;
+ }
+
+ return 0;
+}
+
+static uint64_t
+zap_hash(uint64_t salt, const char *name)
+{
+ static uint64_t table[256];
+ const uint8_t *cp;
+ uint8_t c;
+ uint64_t crc = salt;
+
+ if (table[128] == 0) {
+ uint64_t *ct;
+ int i, j;
+ for (i = 0; i < 256; i++) {
+ for (ct = table + i, *ct = i, j = 8; j > 0; j--)
+ *ct = (*ct >> 1) ^ (-(*ct & 1) & ZFS_CRC64_POLY);
+ }
+ }
+
+ for (cp = (const uint8_t *) name; (c = *cp) != '\0'; cp++)
+ crc = (crc >> 8) ^ table[(crc ^ c) & 0xFF];
+
+ /*
+ * Only use 28 bits, since we need 4 bits in the cookie for the
+ * collision differentiator. We MUST use the high bits, since
+ * those are the onces that we first pay attention to when
+ * chosing the bucket.
+ */
+ crc &= ~((1ULL << (64 - ZAP_HASHBITS)) - 1);
+
+ return crc;
+}
+
+/*
+ * Only to be used on 8-bit arrays.
+ * array_len is actual len in bytes (not encoded le_value_length).
+ * buf is null-terminated.
+ */
+/* XXX */
+static int
+zap_leaf_array_equal(zap_leaf_phys_t *l, zfs_endian_t endian,
+ int blksft, int chunk, int array_len, const char *buf)
+{
+ int bseen = 0;
+
+ while (bseen < array_len) {
+ struct zap_leaf_array *la = &ZAP_LEAF_CHUNK(l, blksft, chunk).l_array;
+ int toread = MIN(array_len - bseen, ZAP_LEAF_ARRAY_BYTES);
+
+ if (chunk >= ZAP_LEAF_NUMCHUNKS(blksft))
+ return 0;
+
+ if (memcmp(la->la_array, buf + bseen, toread) != 0)
+ break;
+ chunk = zfs_to_cpu16(la->la_next, endian);
+ bseen += toread;
+ }
+ return (bseen == array_len);
+}
+
+/* XXX */
+static int
+zap_leaf_array_get(zap_leaf_phys_t *l, zfs_endian_t endian, int blksft,
+ int chunk, int array_len, char *buf)
+{
+ int bseen = 0;
+
+ while (bseen < array_len) {
+ struct zap_leaf_array *la = &ZAP_LEAF_CHUNK(l, blksft, chunk).l_array;
+ int toread = MIN(array_len - bseen, ZAP_LEAF_ARRAY_BYTES);
+
+ if (chunk >= ZAP_LEAF_NUMCHUNKS(blksft))
+ /* Don't use errno because this error is to be ignored. */
+ return ZFS_ERR_BAD_FS;
+
+ memcpy(buf + bseen, la->la_array, toread);
+ chunk = zfs_to_cpu16(la->la_next, endian);
+ bseen += toread;
+ }
+ return ZFS_ERR_NONE;
+}
+
+
+/*
+ * Given a zap_leaf_phys_t, walk thru the zap leaf chunks to get the
+ * value for the property "name".
+ *
+ */
+/* XXX */
+static int
+zap_leaf_lookup(zap_leaf_phys_t *l, zfs_endian_t endian,
+ int blksft, uint64_t h,
+ const char *name, uint64_t *value)
+{
+ uint16_t chunk;
+ struct zap_leaf_entry *le;
+
+ /* Verify if this is a valid leaf block */
+ if (zfs_to_cpu64(l->l_hdr.lh_block_type, endian) != ZBT_LEAF) {
+ printf("invalid leaf type\n");
+ return ZFS_ERR_BAD_FS;
+ }
+ if (zfs_to_cpu32(l->l_hdr.lh_magic, endian) != ZAP_LEAF_MAGIC) {
+ printf("invalid leaf magic\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ for (chunk = zfs_to_cpu16(l->l_hash[LEAF_HASH(blksft, h)], endian);
+ chunk != CHAIN_END; chunk = le->le_next) {
+
+ if (chunk >= ZAP_LEAF_NUMCHUNKS(blksft)) {
+ printf("invalid chunk number\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ le = ZAP_LEAF_ENTRY(l, blksft, chunk);
+
+ /* Verify the chunk entry */
+ if (le->le_type != ZAP_CHUNK_ENTRY) {
+ printf("invalid chunk entry\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ if (zfs_to_cpu64(le->le_hash, endian) != h)
+ continue;
+
+ if (zap_leaf_array_equal(l, endian, blksft,
+ zfs_to_cpu16(le->le_name_chunk, endian),
+ zfs_to_cpu16(le->le_name_length, endian),
+ name)) {
+ struct zap_leaf_array *la;
+
+ if (le->le_int_size != 8 || le->le_value_length != 1) {
+ printf("invalid leaf chunk entry\n");
+ return ZFS_ERR_BAD_FS;
+ }
+ /* get the uint64_t property value */
+ la = &ZAP_LEAF_CHUNK(l, blksft, le->le_value_chunk).l_array;
+
+ *value = be64_to_cpu(la->la_array64);
+
+ return ZFS_ERR_NONE;
+ }
+ }
+
+ printf("couldn't find '%s'\n", name);
+ return ZFS_ERR_FILE_NOT_FOUND;
+}
+
+
+/* Verify if this is a fat zap header block */
+static int
+zap_verify(zap_phys_t *zap)
+{
+ if (zap->zap_magic != (uint64_t) ZAP_MAGIC) {
+ printf("bad ZAP magic\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ if (zap->zap_flags != 0) {
+ printf("bad ZAP flags\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ if (zap->zap_salt == 0) {
+ printf("bad ZAP salt\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ return ZFS_ERR_NONE;
+}
+
+/*
+ * Fat ZAP lookup
+ *
+ */
+/* XXX */
+static int
+fzap_lookup(dnode_end_t *zap_dnode, zap_phys_t *zap,
+ char *name, uint64_t *value, struct zfs_data *data)
+{
+ void *l;
+ uint64_t hash, idx, blkid;
+ int blksft = zfs_log2(zfs_to_cpu16(zap_dnode->dn.dn_datablkszsec,
+ zap_dnode->endian) << DNODE_SHIFT);
+ int err;
+ zfs_endian_t leafendian;
+
+ err = zap_verify(zap);
+ if (err)
+ return err;
+
+ hash = zap_hash(zap->zap_salt, name);
+
+ /* get block id from index */
+ if (zap->zap_ptrtbl.zt_numblks != 0) {
+ printf("external pointer tables not supported\n");
+ return ZFS_ERR_NOT_IMPLEMENTED_YET;
+ }
+ idx = ZAP_HASH_IDX(hash, zap->zap_ptrtbl.zt_shift);
+ blkid = ((uint64_t *) zap)[idx + (1 << (blksft - 3 - 1))];
+
+ /* Get the leaf block */
+ if ((1U << blksft) < sizeof(zap_leaf_phys_t)) {
+ printf("ZAP leaf is too small\n");
+ return ZFS_ERR_BAD_FS;
+ }
+ err = dmu_read(zap_dnode, blkid, &l, &leafendian, data);
+ if (err)
+ return err;
+
+ err = zap_leaf_lookup(l, leafendian, blksft, hash, name, value);
+ free(l);
+ return err;
+}
+
+/* XXX */
+static int
+fzap_iterate(dnode_end_t *zap_dnode, zap_phys_t *zap,
+ int (*hook)(const char *name,
+ uint64_t val,
+ struct zfs_data *data),
+ struct zfs_data *data)
+{
+ zap_leaf_phys_t *l;
+ void *l_in;
+ uint64_t idx, blkid;
+ uint16_t chunk;
+ int blksft = zfs_log2(zfs_to_cpu16(zap_dnode->dn.dn_datablkszsec,
+ zap_dnode->endian) << DNODE_SHIFT);
+ int err;
+ zfs_endian_t endian;
+
+ if (zap_verify(zap))
+ return 0;
+
+ /* get block id from index */
+ if (zap->zap_ptrtbl.zt_numblks != 0) {
+ printf("external pointer tables not supported\n");
+ return 0;
+ }
+ /* Get the leaf block */
+ if ((1U << blksft) < sizeof(zap_leaf_phys_t)) {
+ printf("ZAP leaf is too small\n");
+ return 0;
+ }
+ for (idx = 0; idx < zap->zap_ptrtbl.zt_numblks; idx++) {
+ blkid = ((uint64_t *) zap)[idx + (1 << (blksft - 3 - 1))];
+
+ err = dmu_read(zap_dnode, blkid, &l_in, &endian, data);
+ l = l_in;
+ if (err)
+ continue;
+
+ /* Verify if this is a valid leaf block */
+ if (zfs_to_cpu64(l->l_hdr.lh_block_type, endian) != ZBT_LEAF) {
+ free(l);
+ continue;
+ }
+ if (zfs_to_cpu32(l->l_hdr.lh_magic, endian) != ZAP_LEAF_MAGIC) {
+ free(l);
+ continue;
+ }
+
+ for (chunk = 0; chunk < ZAP_LEAF_NUMCHUNKS(blksft); chunk++) {
+ char *buf;
+ struct zap_leaf_array *la;
+ struct zap_leaf_entry *le;
+ uint64_t val;
+ le = ZAP_LEAF_ENTRY(l, blksft, chunk);
+
+ /* Verify the chunk entry */
+ if (le->le_type != ZAP_CHUNK_ENTRY)
+ continue;
+
+ buf = malloc(zfs_to_cpu16(le->le_name_length, endian)
+ + 1);
+ if (zap_leaf_array_get(l, endian, blksft, le->le_name_chunk,
+ le->le_name_length, buf)) {
+ free(buf);
+ continue;
+ }
+ buf[le->le_name_length] = 0;
+
+ if (le->le_int_size != 8
+ || zfs_to_cpu16(le->le_value_length, endian) != 1)
+ continue;
+
+ /* get the uint64_t property value */
+ la = &ZAP_LEAF_CHUNK(l, blksft, le->le_value_chunk).l_array;
+ val = be64_to_cpu(la->la_array64);
+ if (hook(buf, val, data))
+ return 1;
+ free(buf);
+ }
+ }
+ return 0;
+}
+
+
+/*
+ * Read in the data of a zap object and find the value for a matching
+ * property name.
+ *
+ */
+static int
+zap_lookup(dnode_end_t *zap_dnode, char *name, uint64_t *val,
+ struct zfs_data *data)
+{
+ uint64_t block_type;
+ int size;
+ void *zapbuf;
+ int err;
+ zfs_endian_t endian;
+
+ /* Read in the first block of the zap object data. */
+ size = zfs_to_cpu16(zap_dnode->dn.dn_datablkszsec,
+ zap_dnode->endian) << SPA_MINBLOCKSHIFT;
+ err = dmu_read(zap_dnode, 0, &zapbuf, &endian, data);
+ if (err)
+ return err;
+ block_type = zfs_to_cpu64(*((uint64_t *) zapbuf), endian);
+
+ if (block_type == ZBT_MICRO) {
+ err = (mzap_lookup(zapbuf, endian, size, name, val));
+ free(zapbuf);
+ return err;
+ } else if (block_type == ZBT_HEADER) {
+ /* this is a fat zap */
+ err = (fzap_lookup(zap_dnode, zapbuf, name, val, data));
+ free(zapbuf);
+ return err;
+ }
+
+ printf("unknown ZAP type\n");
+ return ZFS_ERR_BAD_FS;
+}
+
+static int
+zap_iterate(dnode_end_t *zap_dnode,
+ int (*hook)(const char *name, uint64_t val,
+ struct zfs_data *data),
+ struct zfs_data *data)
+{
+ uint64_t block_type;
+ int size;
+ void *zapbuf;
+ int err;
+ int ret;
+ zfs_endian_t endian;
+
+ /* Read in the first block of the zap object data. */
+ size = zfs_to_cpu16(zap_dnode->dn.dn_datablkszsec, zap_dnode->endian) << SPA_MINBLOCKSHIFT;
+ err = dmu_read(zap_dnode, 0, &zapbuf, &endian, data);
+ if (err)
+ return 0;
+ block_type = zfs_to_cpu64(*((uint64_t *) zapbuf), endian);
+
+ if (block_type == ZBT_MICRO) {
+ ret = mzap_iterate(zapbuf, endian, size, hook, data);
+ free(zapbuf);
+ return ret;
+ } else if (block_type == ZBT_HEADER) {
+ /* this is a fat zap */
+ ret = fzap_iterate(zap_dnode, zapbuf, hook, data);
+ free(zapbuf);
+ return ret;
+ }
+ printf("unknown ZAP type\n");
+ return 0;
+}
+
+
+/*
+ * Get the dnode of an object number from the metadnode of an object set.
+ *
+ * Input
+ * mdn - metadnode to get the object dnode
+ * objnum - object number for the object dnode
+ * buf - data buffer that holds the returning dnode
+ */
+static int
+dnode_get(dnode_end_t *mdn, uint64_t objnum, uint8_t type,
+ dnode_end_t *buf, struct zfs_data *data)
+{
+ uint64_t blkid, blksz; /* the block id this object dnode is in */
+ int epbs; /* shift of number of dnodes in a block */
+ int idx; /* index within a block */
+ void *dnbuf;
+ int err;
+ zfs_endian_t endian;
+
+ blksz = zfs_to_cpu16(mdn->dn.dn_datablkszsec,
+ mdn->endian) << SPA_MINBLOCKSHIFT;
+
+ epbs = zfs_log2(blksz) - DNODE_SHIFT;
+ blkid = objnum >> epbs;
+ idx = objnum & ((1 << epbs) - 1);
+
+ if (data->dnode_buf != NULL && memcmp(data->dnode_mdn, mdn,
+ sizeof(*mdn)) == 0
+ && objnum >= data->dnode_start && objnum < data->dnode_end) {
+ memmove(&(buf->dn), &(data->dnode_buf)[idx], DNODE_SIZE);
+ buf->endian = data->dnode_endian;
+ if (type && buf->dn.dn_type != type) {
+ printf("incorrect dnode type: %02X != %02x\n", buf->dn.dn_type, type);
+ return ZFS_ERR_BAD_FS;
+ }
+ return ZFS_ERR_NONE;
+ }
+
+ err = dmu_read(mdn, blkid, &dnbuf, &endian, data);
+ if (err)
+ return err;
+
+ free(data->dnode_buf);
+ free(data->dnode_mdn);
+ data->dnode_mdn = malloc(sizeof(*mdn));
+ if (!data->dnode_mdn) {
+ data->dnode_buf = 0;
+ } else {
+ memcpy(data->dnode_mdn, mdn, sizeof(*mdn));
+ data->dnode_buf = dnbuf;
+ data->dnode_start = blkid << epbs;
+ data->dnode_end = (blkid + 1) << epbs;
+ data->dnode_endian = endian;
+ }
+
+ memmove(&(buf->dn), (dnode_phys_t *) dnbuf + idx, DNODE_SIZE);
+ buf->endian = endian;
+ if (type && buf->dn.dn_type != type) {
+ printf("incorrect dnode type\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ return ZFS_ERR_NONE;
+}
+
+/*
+ * Get the file dnode for a given file name where mdn is the meta dnode
+ * for this ZFS object set. When found, place the file dnode in dn.
+ * The 'path' argument will be mangled.
+ *
+ */
+static int
+dnode_get_path(dnode_end_t *mdn, const char *path_in, dnode_end_t *dn,
+ struct zfs_data *data)
+{
+ uint64_t objnum, version;
+ char *cname, ch;
+ int err = ZFS_ERR_NONE;
+ char *path, *path_buf;
+ struct dnode_chain {
+ struct dnode_chain *next;
+ dnode_end_t dn;
+ };
+ struct dnode_chain *dnode_path = 0, *dn_new, *root;
+
+ dn_new = malloc(sizeof(*dn_new));
+ if (!dn_new)
+ return ZFS_ERR_OUT_OF_MEMORY;
+ dn_new->next = 0;
+ dnode_path = root = dn_new;
+
+ err = dnode_get(mdn, MASTER_NODE_OBJ, DMU_OT_MASTER_NODE,
+ &(dnode_path->dn), data);
+ if (err) {
+ free(dn_new);
+ return err;
+ }
+
+ err = zap_lookup(&(dnode_path->dn), ZPL_VERSION_STR, &version, data);
+ if (err) {
+ free(dn_new);
+ return err;
+ }
+ if (version > ZPL_VERSION) {
+ free(dn_new);
+ printf("too new ZPL version\n");
+ return ZFS_ERR_NOT_IMPLEMENTED_YET;
+ }
+
+ err = zap_lookup(&(dnode_path->dn), ZFS_ROOT_OBJ, &objnum, data);
+ if (err) {
+ free(dn_new);
+ return err;
+ }
+
+ err = dnode_get(mdn, objnum, 0, &(dnode_path->dn), data);
+ if (err) {
+ free(dn_new);
+ return err;
+ }
+
+ path = path_buf = strdup(path_in);
+ if (!path_buf) {
+ free(dn_new);
+ return ZFS_ERR_OUT_OF_MEMORY;
+ }
+
+ while (1) {
+ /* skip leading slashes */
+ while (*path == '/')
+ path++;
+ if (!*path)
+ break;
+ /* get the next component name */
+ cname = path;
+ while (*path && *path != '/')
+ path++;
+ /* Skip dot. */
+ if (cname + 1 == path && cname[0] == '.')
+ continue;
+ /* Handle double dot. */
+ if (cname + 2 == path && cname[0] == '.' && cname[1] == '.') {
+ if (dn_new->next) {
+ dn_new = dnode_path;
+ dnode_path = dn_new->next;
+ free(dn_new);
+ } else {
+ printf("can't resolve ..\n");
+ err = ZFS_ERR_FILE_NOT_FOUND;
+ break;
+ }
+ continue;
+ }
+
+ ch = *path;
+ *path = 0; /* ensure null termination */
+
+ if (dnode_path->dn.dn.dn_type != DMU_OT_DIRECTORY_CONTENTS) {
+ free(path_buf);
+ printf("not a directory\n");
+ return ZFS_ERR_BAD_FILE_TYPE;
+ }
+ err = zap_lookup(&(dnode_path->dn), cname, &objnum, data);
+ if (err)
+ break;
+
+ dn_new = malloc(sizeof(*dn_new));
+ if (!dn_new) {
+ err = ZFS_ERR_OUT_OF_MEMORY;
+ break;
+ }
+ dn_new->next = dnode_path;
+ dnode_path = dn_new;
+
+ objnum = ZFS_DIRENT_OBJ(objnum);
+ err = dnode_get(mdn, objnum, 0, &(dnode_path->dn), data);
+ if (err)
+ break;
+
+ *path = ch;
+ }
+
+ if (!err)
+ memcpy(dn, &(dnode_path->dn), sizeof(*dn));
+
+ while (dnode_path) {
+ dn_new = dnode_path->next;
+ free(dnode_path);
+ dnode_path = dn_new;
+ }
+ free(path_buf);
+ return err;
+}
+
+
+/*
+ * Given a MOS metadnode, get the metadnode of a given filesystem name (fsname),
+ * e.g. pool/rootfs, or a given object number (obj), e.g. the object number
+ * of pool/rootfs.
+ *
+ * If no fsname and no obj are given, return the DSL_DIR metadnode.
+ * If fsname is given, return its metadnode and its matching object number.
+ * If only obj is given, return the metadnode for this object number.
+ *
+ */
+static int
+get_filesystem_dnode(dnode_end_t *mosmdn, char *fsname,
+ dnode_end_t *mdn, struct zfs_data *data)
+{
+ uint64_t objnum;
+ int err;
+
+ err = dnode_get(mosmdn, DMU_POOL_DIRECTORY_OBJECT,
+ DMU_OT_OBJECT_DIRECTORY, mdn, data);
+ if (err)
+ return err;
+
+ err = zap_lookup(mdn, DMU_POOL_ROOT_DATASET, &objnum, data);
+ if (err)
+ return err;
+
+ err = dnode_get(mosmdn, objnum, DMU_OT_DSL_DIR, mdn, data);
+ if (err)
+ return err;
+
+ while (*fsname) {
+ uint64_t childobj;
+ char *cname, ch;
+
+ while (*fsname == '/')
+ fsname++;
+
+ if (!*fsname || *fsname == '@')
+ break;
+
+ cname = fsname;
+ while (*fsname && !isspace(*fsname) && *fsname != '/')
+ fsname++;
+ ch = *fsname;
+ *fsname = 0;
+
+ childobj = zfs_to_cpu64((((dsl_dir_phys_t *) DN_BONUS(&mdn->dn)))->dd_child_dir_zapobj, mdn->endian);
+ err = dnode_get(mosmdn, childobj,
+ DMU_OT_DSL_DIR_CHILD_MAP, mdn, data);
+ if (err)
+ return err;
+
+ err = zap_lookup(mdn, cname, &objnum, data);
+ if (err)
+ return err;
+
+ err = dnode_get(mosmdn, objnum, DMU_OT_DSL_DIR, mdn, data);
+ if (err)
+ return err;
+
+ *fsname = ch;
+ }
+ return ZFS_ERR_NONE;
+}
+
+static int
+make_mdn(dnode_end_t *mdn, struct zfs_data *data)
+{
+ void *osp;
+ blkptr_t *bp;
+ size_t ospsize;
+ int err;
+
+ bp = &(((dsl_dataset_phys_t *) DN_BONUS(&mdn->dn))->ds_bp);
+ err = zio_read(bp, mdn->endian, &osp, &ospsize, data);
+ if (err)
+ return err;
+ if (ospsize < OBJSET_PHYS_SIZE_V14) {
+ free(osp);
+ printf("too small osp\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ mdn->endian = (zfs_to_cpu64(bp->blk_prop, mdn->endian)>>63) & 1;
+ memmove((char *) &(mdn->dn),
+ (char *) &((objset_phys_t *) osp)->os_meta_dnode, DNODE_SIZE);
+ free(osp);
+ return ZFS_ERR_NONE;
+}
+
+static int
+dnode_get_fullpath(const char *fullpath, dnode_end_t *mdn,
+ uint64_t *mdnobj, dnode_end_t *dn, int *isfs,
+ struct zfs_data *data)
+{
+ char *fsname, *snapname;
+ const char *ptr_at, *filename;
+ uint64_t headobj;
+ int err;
+
+ ptr_at = strchr(fullpath, '@');
+ if (!ptr_at) {
+ *isfs = 1;
+ filename = 0;
+ snapname = 0;
+ fsname = strdup(fullpath);
+ } else {
+ const char *ptr_slash = strchr(ptr_at, '/');
+
+ *isfs = 0;
+ fsname = malloc(ptr_at - fullpath + 1);
+ if (!fsname)
+ return ZFS_ERR_OUT_OF_MEMORY;
+ memcpy(fsname, fullpath, ptr_at - fullpath);
+ fsname[ptr_at - fullpath] = 0;
+ if (ptr_at[1] && ptr_at[1] != '/') {
+ snapname = malloc(ptr_slash - ptr_at);
+ if (!snapname) {
+ free(fsname);
+ return ZFS_ERR_OUT_OF_MEMORY;
+ }
+ memcpy(snapname, ptr_at + 1, ptr_slash - ptr_at - 1);
+ snapname[ptr_slash - ptr_at - 1] = 0;
+ } else {
+ snapname = 0;
+ }
+ if (ptr_slash)
+ filename = ptr_slash;
+ else
+ filename = "/";
+ printf("zfs fsname = '%s' snapname='%s' filename = '%s'\n",
+ fsname, snapname, filename);
+ }
+
+
+ err = get_filesystem_dnode(&(data->mos), fsname, dn, data);
+
+ if (err) {
+ free(fsname);
+ free(snapname);
+ return err;
+ }
+
+ headobj = zfs_to_cpu64(((dsl_dir_phys_t *) DN_BONUS(&dn->dn))->dd_head_dataset_obj, dn->endian);
+
+ err = dnode_get(&(data->mos), headobj, DMU_OT_DSL_DATASET, mdn, data);
+ if (err) {
+ free(fsname);
+ free(snapname);
+ return err;
+ }
+
+ if (snapname) {
+ uint64_t snapobj;
+
+ snapobj = zfs_to_cpu64(((dsl_dataset_phys_t *) DN_BONUS(&mdn->dn))->ds_snapnames_zapobj, mdn->endian);
+
+ err = dnode_get(&(data->mos), snapobj,
+ DMU_OT_DSL_DS_SNAP_MAP, mdn, data);
+ if (!err)
+ err = zap_lookup(mdn, snapname, &headobj, data);
+ if (!err)
+ err = dnode_get(&(data->mos), headobj, DMU_OT_DSL_DATASET, mdn, data);
+ if (err) {
+ free(fsname);
+ free(snapname);
+ return err;
+ }
+ }
+
+ if (mdnobj)
+ *mdnobj = headobj;
+
+ make_mdn(mdn, data);
+
+ if (*isfs) {
+ free(fsname);
+ free(snapname);
+ return ZFS_ERR_NONE;
+ }
+ err = dnode_get_path(mdn, filename, dn, data);
+ free(fsname);
+ free(snapname);
+ return err;
+}
+
+/*
+ * For a given XDR packed nvlist, verify the first 4 bytes and move on.
+ *
+ * An XDR packed nvlist is encoded as (comments from nvs_xdr_create) :
+ *
+ * encoding method/host endian (4 bytes)
+ * nvl_version (4 bytes)
+ * nvl_nvflag (4 bytes)
+ * encoded nvpairs:
+ * encoded size of the nvpair (4 bytes)
+ * decoded size of the nvpair (4 bytes)
+ * name string size (4 bytes)
+ * name string data (sizeof(NV_ALIGN4(string))
+ * data type (4 bytes)
+ * # of elements in the nvpair (4 bytes)
+ * data
+ * 2 zero's for the last nvpair
+ * (end of the entire list) (8 bytes)
+ *
+ */
+
+static int
+nvlist_find_value(char *nvlist, char *name, int valtype, char **val,
+ size_t *size_out, size_t *nelm_out)
+{
+ int name_len, type, encode_size;
+ char *nvpair, *nvp_name;
+
+ /* Verify if the 1st and 2nd byte in the nvlist are valid. */
+ /* NOTE: independently of what endianness header announces all
+ subsequent values are big-endian. */
+ if (nvlist[0] != NV_ENCODE_XDR || (nvlist[1] != NV_LITTLE_ENDIAN
+ && nvlist[1] != NV_BIG_ENDIAN)) {
+ printf("zfs incorrect nvlist header\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ /* skip the header, nvl_version, and nvl_nvflag */
+ nvlist = nvlist + 4 * 3;
+ /*
+ * Loop thru the nvpair list
+ * The XDR representation of an integer is in big-endian byte order.
+ */
+ while ((encode_size = be32_to_cpu(*(uint32_t *) nvlist))) {
+ int nelm;
+
+ nvpair = nvlist + 4 * 2; /* skip the encode/decode size */
+
+ name_len = be32_to_cpu(*(uint32_t *) nvpair);
+ nvpair += 4;
+
+ nvp_name = nvpair;
+ nvpair = nvpair + ((name_len + 3) & ~3); /* align */
+
+ type = be32_to_cpu(*(uint32_t *) nvpair);
+ nvpair += 4;
+
+ nelm = be32_to_cpu(*(uint32_t *) nvpair);
+ if (nelm < 1) {
+ printf("empty nvpair\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ nvpair += 4;
+
+ if ((strncmp(nvp_name, name, name_len) == 0) && type == valtype) {
+ *val = nvpair;
+ *size_out = encode_size;
+ if (nelm_out)
+ *nelm_out = nelm;
+ return 1;
+ }
+
+ nvlist += encode_size; /* goto the next nvpair */
+ }
+ return 0;
+}
+
+int
+zfs_nvlist_lookup_uint64(char *nvlist, char *name, uint64_t *out)
+{
+ char *nvpair;
+ size_t size;
+ int found;
+
+ found = nvlist_find_value(nvlist, name, DATA_TYPE_UINT64, &nvpair, &size, 0);
+ if (!found)
+ return 0;
+ if (size < sizeof(uint64_t)) {
+ printf("invalid uint64\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ *out = be64_to_cpu(*(uint64_t *) nvpair);
+ return 1;
+}
+
+char *
+zfs_nvlist_lookup_string(char *nvlist, char *name)
+{
+ char *nvpair;
+ char *ret;
+ size_t slen;
+ size_t size;
+ int found;
+
+ found = nvlist_find_value(nvlist, name, DATA_TYPE_STRING, &nvpair, &size, 0);
+ if (!found)
+ return 0;
+ if (size < 4) {
+ printf("invalid string\n");
+ return 0;
+ }
+ slen = be32_to_cpu(*(uint32_t *) nvpair);
+ if (slen > size - 4)
+ slen = size - 4;
+ ret = malloc(slen + 1);
+ if (!ret)
+ return 0;
+ memcpy(ret, nvpair + 4, slen);
+ ret[slen] = 0;
+ return ret;
+}
+
+char *
+zfs_nvlist_lookup_nvlist(char *nvlist, char *name)
+{
+ char *nvpair;
+ char *ret;
+ size_t size;
+ int found;
+
+ found = nvlist_find_value(nvlist, name, DATA_TYPE_NVLIST, &nvpair,
+ &size, 0);
+ if (!found)
+ return 0;
+ ret = calloc(1, size + 3 * sizeof(uint32_t));
+ if (!ret)
+ return 0;
+ memcpy(ret, nvlist, sizeof(uint32_t));
+
+ memcpy(ret + sizeof(uint32_t), nvpair, size);
+ return ret;
+}
+
+int
+zfs_nvlist_lookup_nvlist_array_get_nelm(char *nvlist, char *name)
+{
+ char *nvpair;
+ size_t nelm, size;
+ int found;
+
+ found = nvlist_find_value(nvlist, name, DATA_TYPE_NVLIST, &nvpair,
+ &size, &nelm);
+ if (!found)
+ return -1;
+ return nelm;
+}
+
+char *
+zfs_nvlist_lookup_nvlist_array(char *nvlist, char *name,
+ size_t index)
+{
+ char *nvpair, *nvpairptr;
+ int found;
+ char *ret;
+ size_t size;
+ unsigned i;
+ size_t nelm;
+
+ found = nvlist_find_value(nvlist, name, DATA_TYPE_NVLIST, &nvpair,
+ &size, &nelm);
+ if (!found)
+ return 0;
+ if (index >= nelm) {
+ printf("trying to lookup past nvlist array\n");
+ return 0;
+ }
+
+ nvpairptr = nvpair;
+
+ for (i = 0; i < index; i++) {
+ uint32_t encode_size;
+
+ /* skip the header, nvl_version, and nvl_nvflag */
+ nvpairptr = nvpairptr + 4 * 2;
+
+ while (nvpairptr < nvpair + size
+ && (encode_size = be32_to_cpu(*(uint32_t *) nvpairptr)))
+ nvlist += encode_size; /* goto the next nvpair */
+
+ nvlist = nvlist + 4 * 2; /* skip the ending 2 zeros - 8 bytes */
+ }
+
+ if (nvpairptr >= nvpair + size
+ || nvpairptr + be32_to_cpu(*(uint32_t *) (nvpairptr + 4 * 2))
+ >= nvpair + size) {
+ printf("incorrect nvlist array\n");
+ return 0;
+ }
+
+ ret = calloc(1, be32_to_cpu(*(uint32_t *) (nvpairptr + 4 * 2))
+ + 3 * sizeof(uint32_t));
+ if (!ret)
+ return 0;
+ memcpy(ret, nvlist, sizeof(uint32_t));
+
+ memcpy(ret + sizeof(uint32_t), nvpairptr, size);
+ return ret;
+}
+
+static int
+int_zfs_fetch_nvlist(struct zfs_data *data, char **nvlist)
+{
+ int err;
+
+ *nvlist = malloc(VDEV_PHYS_SIZE);
+ /* Read in the vdev name-value pair list (112K). */
+ err = zfs_devread(data->vdev_phys_sector, 0, VDEV_PHYS_SIZE, *nvlist);
+ if (err) {
+ free(*nvlist);
+ *nvlist = 0;
+ return err;
+ }
+ return ZFS_ERR_NONE;
+}
+
+/*
+ * Check the disk label information and retrieve needed vdev name-value pairs.
+ *
+ */
+static int
+check_pool_label(struct zfs_data *data)
+{
+ uint64_t pool_state;
+ char *nvlist; /* for the pool */
+ char *vdevnvlist; /* for the vdev */
+ uint64_t diskguid;
+ uint64_t version;
+ int found;
+ int err;
+
+ err = int_zfs_fetch_nvlist(data, &nvlist);
+ if (err)
+ return err;
+
+ found = zfs_nvlist_lookup_uint64(nvlist, ZPOOL_CONFIG_POOL_STATE,
+ &pool_state);
+ if (!found) {
+ free(nvlist);
+ printf("zfs pool state not found\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ if (pool_state == POOL_STATE_DESTROYED) {
+ free(nvlist);
+ printf("zpool is marked as destroyed\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ data->label_txg = 0;
+ found = zfs_nvlist_lookup_uint64(nvlist, ZPOOL_CONFIG_POOL_TXG,
+ &data->label_txg);
+ if (!found) {
+ free(nvlist);
+ printf("zfs pool txg not found\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ /* not an active device */
+ if (data->label_txg == 0) {
+ free(nvlist);
+ printf("zpool is not active\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ found = zfs_nvlist_lookup_uint64(nvlist, ZPOOL_CONFIG_VERSION,
+ &version);
+ if (!found) {
+ free(nvlist);
+ printf("zpool config version not found\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ if (version > SPA_VERSION) {
+ free(nvlist);
+ printf("SPA version too new %llu > %llu\n",
+ (unsigned long long) version,
+ (unsigned long long) SPA_VERSION);
+ return ZFS_ERR_NOT_IMPLEMENTED_YET;
+ }
+
+ vdevnvlist = zfs_nvlist_lookup_nvlist(nvlist, ZPOOL_CONFIG_VDEV_TREE);
+ if (!vdevnvlist) {
+ free(nvlist);
+ printf("ZFS config vdev tree not found\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ found = zfs_nvlist_lookup_uint64(vdevnvlist, ZPOOL_CONFIG_ASHIFT,
+ &data->vdev_ashift);
+ free(vdevnvlist);
+ if (!found) {
+ free(nvlist);
+ printf("ZPOOL config ashift not found\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ found = zfs_nvlist_lookup_uint64(nvlist, ZPOOL_CONFIG_GUID, &diskguid);
+ if (!found) {
+ free(nvlist);
+ printf("ZPOOL config guid not found\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ found = zfs_nvlist_lookup_uint64(nvlist, ZPOOL_CONFIG_POOL_GUID, &data->pool_guid);
+ if (!found) {
+ free(nvlist);
+ printf("ZPOOL config pool guid not found\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ free(nvlist);
+
+ printf("ZFS Pool GUID: %llu (%016llx) Label: GUID: %llu (%016llx), txg: %llu, SPA v%llu, ashift: %llu\n",
+ (unsigned long long) data->pool_guid,
+ (unsigned long long) data->pool_guid,
+ (unsigned long long) diskguid,
+ (unsigned long long) diskguid,
+ (unsigned long long) data->label_txg,
+ (unsigned long long) version,
+ (unsigned long long) data->vdev_ashift);
+
+ return ZFS_ERR_NONE;
+}
+
+/*
+ * vdev_label_start returns the physical disk offset (in bytes) of
+ * label "l".
+ */
+static uint64_t vdev_label_start(uint64_t psize, int l)
+{
+ return (l * sizeof(vdev_label_t) + (l < VDEV_LABELS / 2 ?
+ 0 : psize -
+ VDEV_LABELS * sizeof(vdev_label_t)));
+}
+
+void
+zfs_unmount(struct zfs_data *data)
+{
+ free(data->dnode_buf);
+ free(data->dnode_mdn);
+ free(data->file_buf);
+ free(data);
+}
+
+/*
+ * zfs_mount() locates a valid uberblock of the root pool and read in its MOS
+ * to the memory address MOS.
+ *
+ */
+struct zfs_data *
+zfs_mount(device_t dev)
+{
+ struct zfs_data *data = 0;
+ int label = 0, bestlabel = -1;
+ char *ub_array;
+ uberblock_t *ubbest;
+ uberblock_t *ubcur = NULL;
+ void *osp = 0;
+ size_t ospsize;
+ int err;
+
+ data = malloc(sizeof(*data));
+ if (!data)
+ return 0;
+ memset(data, 0, sizeof(*data));
+
+ ub_array = malloc(VDEV_UBERBLOCK_RING);
+ if (!ub_array) {
+ zfs_unmount(data);
+ return 0;
+ }
+
+ ubbest = malloc(sizeof(*ubbest));
+ if (!ubbest) {
+ zfs_unmount(data);
+ return 0;
+ }
+ memset(ubbest, 0, sizeof(*ubbest));
+
+ /*
+ * some eltorito stacks don't give us a size and
+ * we end up setting the size to MAXUINT, further
+ * some of these devices stop working once a single
+ * read past the end has been issued. Checking
+ * for a maximum part_length and skipping the backup
+ * labels at the end of the slice/partition/device
+ * avoids breaking down on such devices.
+ */
+ const int vdevnum =
+ dev->part_length == 0 ?
+ VDEV_LABELS / 2 : VDEV_LABELS;
+
+ /* Size in bytes of the device (disk or partition) aligned to label size*/
+ uint64_t device_size =
+ dev->part_length << SECTOR_BITS;
+
+ const uint64_t alignedbytes =
+ P2ALIGN(device_size, (uint64_t) sizeof(vdev_label_t));
+
+ for (label = 0; label < vdevnum; label++) {
+ uint64_t labelstartbytes = vdev_label_start(alignedbytes, label);
+ uint64_t labelstart = labelstartbytes >> SECTOR_BITS;
+
+ debug("zfs reading label %d at sector %llu (byte %llu)\n",
+ label, (unsigned long long) labelstart,
+ (unsigned long long) labelstartbytes);
+
+ data->vdev_phys_sector = labelstart +
+ ((VDEV_SKIP_SIZE + VDEV_BOOT_HEADER_SIZE) >> SECTOR_BITS);
+
+ err = check_pool_label(data);
+ if (err) {
+ printf("zfs error checking label %d\n", label);
+ continue;
+ }
+
+ /* Read in the uberblock ring (128K). */
+ err = zfs_devread(data->vdev_phys_sector +
+ (VDEV_PHYS_SIZE >> SECTOR_BITS),
+ 0, VDEV_UBERBLOCK_RING, ub_array);
+ if (err) {
+ printf("zfs error reading uberblock ring for label %d\n", label);
+ continue;
+ }
+
+ ubcur = find_bestub(ub_array, data);
+ if (!ubcur) {
+ printf("zfs No good uberblocks found in label %d\n", label);
+ continue;
+ }
+
+ if (vdev_uberblock_compare(ubcur, ubbest) > 0) {
+ /* Looks like the block is good, so use it.*/
+ memcpy(ubbest, ubcur, sizeof(*ubbest));
+ bestlabel = label;
+ debug("zfs Current best uberblock found in label %d\n", label);
+ }
+ }
+ free(ub_array);
+
+ /* We zero'd the structure to begin with. If we never assigned to it,
+ magic will still be zero. */
+ if (!ubbest->ub_magic) {
+ printf("couldn't find a valid ZFS label\n");
+ zfs_unmount(data);
+ free(ubbest);
+ return 0;
+ }
+
+ debug("zfs ubbest %p in label %d\n", ubbest, bestlabel);
+
+ zfs_endian_t ub_endian =
+ zfs_to_cpu64(ubbest->ub_magic, LITTLE_ENDIAN) == UBERBLOCK_MAGIC
+ ? LITTLE_ENDIAN : BIG_ENDIAN;
+
+ debug("zfs endian set to %s\n", !ub_endian ? "big" : "little");
+
+ err = zio_read(&ubbest->ub_rootbp, ub_endian, &osp, &ospsize, data);
+
+ if (err) {
+ printf("couldn't zio_read object directory\n");
+ zfs_unmount(data);
+ free(ubbest);
+ return 0;
+ }
+
+ if (ospsize < OBJSET_PHYS_SIZE_V14) {
+ printf("osp too small\n");
+ zfs_unmount(data);
+ free(osp);
+ free(ubbest);
+ return 0;
+ }
+
+ /* Got the MOS. Save it at the memory addr MOS. */
+ memmove(&(data->mos.dn), &((objset_phys_t *) osp)->os_meta_dnode, DNODE_SIZE);
+ data->mos.endian =
+ (zfs_to_cpu64(ubbest->ub_rootbp.blk_prop, ub_endian) >> 63) & 1;
+ memmove(&(data->current_uberblock), ubbest, sizeof(uberblock_t));
+
+ free(osp);
+ free(ubbest);
+
+ return data;
+}
+
+int
+zfs_fetch_nvlist(device_t dev, char **nvlist)
+{
+ struct zfs_data *zfs;
+ int err;
+
+ zfs = zfs_mount(dev);
+ if (!zfs)
+ return ZFS_ERR_BAD_FS;
+ err = int_zfs_fetch_nvlist(zfs, nvlist);
+ zfs_unmount(zfs);
+ return err;
+}
+
+static int
+zfs_label(device_t device, char **label)
+{
+ char *nvlist;
+ int err;
+ struct zfs_data *data;
+
+ data = zfs_mount(device);
+ if (!data)
+ return ZFS_ERR_BAD_FS;
+
+ err = int_zfs_fetch_nvlist(data, &nvlist);
+ if (err) {
+ zfs_unmount(data);
+ return err;
+ }
+
+ *label = zfs_nvlist_lookup_string(nvlist, ZPOOL_CONFIG_POOL_NAME);
+ free(nvlist);
+ zfs_unmount(data);
+ return ZFS_ERR_NONE;
+}
+
+static int
+zfs_uuid(device_t device, char **uuid)
+{
+ struct zfs_data *data;
+
+ data = zfs_mount(device);
+ if (!data)
+ return ZFS_ERR_BAD_FS;
+
+ *uuid = malloc(17); /* %016llx + nil */
+ if (!*uuid)
+ return ZFS_ERR_OUT_OF_MEMORY;
+
+ /* *uuid = xasprintf ("%016llx", (long long unsigned) data->pool_guid);*/
+ snprintf(*uuid, 17, "%016llx", (long long unsigned) data->pool_guid);
+ zfs_unmount(data);
+
+ return ZFS_ERR_NONE;
+}
+
+/*
+ * zfs_open() locates a file in the rootpool by following the
+ * MOS and places the dnode of the file in the memory address DNODE.
+ */
+int
+zfs_open(struct zfs_file *file, const char *fsfilename)
+{
+ struct zfs_data *data;
+ int err;
+ int isfs;
+
+ data = zfs_mount(file->device);
+ if (!data)
+ return ZFS_ERR_BAD_FS;
+
+ err = dnode_get_fullpath(fsfilename, &(data->mdn), 0,
+ &(data->dnode), &isfs, data);
+ if (err) {
+ zfs_unmount(data);
+ return err;
+ }
+
+ if (isfs) {
+ zfs_unmount(data);
+ printf("Missing @ or / separator\n");
+ return ZFS_ERR_FILE_NOT_FOUND;
+ }
+
+ /* We found the dnode for this file. Verify if it is a plain file. */
+ if (data->dnode.dn.dn_type != DMU_OT_PLAIN_FILE_CONTENTS) {
+ zfs_unmount(data);
+ printf("not a file\n");
+ return ZFS_ERR_BAD_FILE_TYPE;
+ }
+
+ /* get the file size and set the file position to 0 */
+
+ /*
+ * For DMU_OT_SA we will need to locate the SIZE attribute
+ * attribute, which could be either in the bonus buffer
+ * or the "spill" block.
+ */
+ if (data->dnode.dn.dn_bonustype == DMU_OT_SA) {
+ void *sahdrp;
+ int hdrsize;
+
+ if (data->dnode.dn.dn_bonuslen != 0) {
+ sahdrp = (sa_hdr_phys_t *) DN_BONUS(&data->dnode.dn);
+ } else if (data->dnode.dn.dn_flags & DNODE_FLAG_SPILL_BLKPTR) {
+ blkptr_t *bp = &data->dnode.dn.dn_spill;
+
+ err = zio_read(bp, data->dnode.endian, &sahdrp, NULL, data);
+ if (err)
+ return err;
+ } else {
+ printf("filesystem is corrupt :(\n");
+ return ZFS_ERR_BAD_FS;
+ }
+
+ hdrsize = SA_HDR_SIZE(((sa_hdr_phys_t *) sahdrp));
+ file->size = *(uint64_t *) ((char *) sahdrp + hdrsize + SA_SIZE_OFFSET);
+ } else {
+ file->size = zfs_to_cpu64(((znode_phys_t *) DN_BONUS(&data->dnode.dn))->zp_size, data->dnode.endian);
+ }
+
+ file->data = data;
+ file->offset = 0;
+
+ return ZFS_ERR_NONE;
+}
+
+uint64_t
+zfs_read(zfs_file_t file, char *buf, uint64_t len)
+{
+ struct zfs_data *data = (struct zfs_data *) file->data;
+ int blksz, movesize;
+ uint64_t length;
+ int64_t red;
+ int err;
+
+ if (data->file_buf == NULL) {
+ data->file_buf = malloc(SPA_MAXBLOCKSIZE);
+ if (!data->file_buf)
+ return -1;
+ data->file_start = data->file_end = 0;
+ }
+
+ /*
+ * If offset is in memory, move it into the buffer provided and return.
+ */
+ if (file->offset >= data->file_start
+ && file->offset + len <= data->file_end) {
+ memmove(buf, data->file_buf + file->offset - data->file_start,
+ len);
+ return len;
+ }
+
+ blksz = zfs_to_cpu16(data->dnode.dn.dn_datablkszsec,
+ data->dnode.endian) << SPA_MINBLOCKSHIFT;
+
+ /*
+ * Entire Dnode is too big to fit into the space available. We
+ * will need to read it in chunks. This could be optimized to
+ * read in as large a chunk as there is space available, but for
+ * now, this only reads in one data block at a time.
+ */
+ length = len;
+ red = 0;
+ while (length) {
+ void *t;
+ /*
+ * Find requested blkid and the offset within that block.
+ */
+ uint64_t blkid = (file->offset + red) / blksz;
+ free(data->file_buf);
+ data->file_buf = 0;
+
+ err = dmu_read(&(data->dnode), blkid, &t,
+ 0, data);
+ data->file_buf = t;
+ if (err)
+ return -1;
+
+ data->file_start = blkid * blksz;
+ data->file_end = data->file_start + blksz;
+
+ movesize = MIN(length, data->file_end - (int) file->offset - red);
+
+ memmove(buf, data->file_buf + file->offset + red
+ - data->file_start, movesize);
+ buf += movesize;
+ length -= movesize;
+ red += movesize;
+ }
+
+ return len;
+}
+
+int
+zfs_close(zfs_file_t file)
+{
+ zfs_unmount((struct zfs_data *) file->data);
+ return ZFS_ERR_NONE;
+}
+
+int
+zfs_getmdnobj(device_t dev, const char *fsfilename,
+ uint64_t *mdnobj)
+{
+ struct zfs_data *data;
+ int err;
+ int isfs;
+
+ data = zfs_mount(dev);
+ if (!data)
+ return ZFS_ERR_BAD_FS;
+
+ err = dnode_get_fullpath(fsfilename, &(data->mdn), mdnobj,
+ &(data->dnode), &isfs, data);
+ zfs_unmount(data);
+ return err;
+}
+
+static void
+fill_fs_info(struct zfs_dirhook_info *info,
+ dnode_end_t mdn, struct zfs_data *data)
+{
+ int err;
+ dnode_end_t dn;
+ uint64_t objnum;
+ uint64_t headobj;
+
+ memset(info, 0, sizeof(*info));
+
+ info->dir = 1;
+
+ if (mdn.dn.dn_type == DMU_OT_DSL_DIR) {
+ headobj = zfs_to_cpu64(((dsl_dir_phys_t *) DN_BONUS(&mdn.dn))->dd_head_dataset_obj, mdn.endian);
+
+ err = dnode_get(&(data->mos), headobj, DMU_OT_DSL_DATASET, &mdn, data);
+ if (err) {
+ printf("zfs failed here 1\n");
+ return;
+ }
+ }
+ make_mdn(&mdn, data);
+ err = dnode_get(&mdn, MASTER_NODE_OBJ, DMU_OT_MASTER_NODE,
+ &dn, data);
+ if (err) {
+ printf("zfs failed here 2\n");
+ return;
+ }
+
+ err = zap_lookup(&dn, ZFS_ROOT_OBJ, &objnum, data);
+ if (err) {
+ printf("zfs failed here 3\n");
+ return;
+ }
+
+ err = dnode_get(&mdn, objnum, 0, &dn, data);
+ if (err) {
+ printf("zfs failed here 4\n");
+ return;
+ }
+
+ info->mtimeset = 1;
+ info->mtime = zfs_to_cpu64(((znode_phys_t *) DN_BONUS(&dn.dn))->zp_mtime[0], dn.endian);
+
+ return;
+}
+
+static int iterate_zap(const char *name, uint64_t val, struct zfs_data *data)
+{
+ struct zfs_dirhook_info info;
+ dnode_end_t dn;
+
+ memset(&info, 0, sizeof(info));
+
+ dnode_get(&(data->mdn), val, 0, &dn, data);
+ info.mtimeset = 1;
+ info.mtime = zfs_to_cpu64(((znode_phys_t *) DN_BONUS(&dn.dn))->zp_mtime[0], dn.endian);
+ info.dir = (dn.dn.dn_type == DMU_OT_DIRECTORY_CONTENTS);
+ debug("zfs type=%d, name=%s\n",
+ (int)dn.dn.dn_type, (char *)name);
+ if (!data->userhook)
+ return 0;
+ return data->userhook(name, &info);
+}
+
+static int iterate_zap_fs(const char *name, uint64_t val, struct zfs_data *data)
+{
+ struct zfs_dirhook_info info;
+ dnode_end_t mdn;
+ int err;
+ err = dnode_get(&(data->mos), val, 0, &mdn, data);
+ if (err)
+ return 0;
+ if (mdn.dn.dn_type != DMU_OT_DSL_DIR)
+ return 0;
+
+ fill_fs_info(&info, mdn, data);
+
+ if (!data->userhook)
+ return 0;
+ return data->userhook(name, &info);
+}
+
+static int iterate_zap_snap(const char *name, uint64_t val, struct zfs_data *data)
+{
+ struct zfs_dirhook_info info;
+ char *name2;
+ int ret = 0;
+ dnode_end_t mdn;
+ int err;
+
+ err = dnode_get(&(data->mos), val, 0, &mdn, data);
+ if (err)
+ return 0;
+
+ if (mdn.dn.dn_type != DMU_OT_DSL_DATASET)
+ return 0;
+
+ fill_fs_info(&info, mdn, data);
+
+ name2 = malloc(strlen(name) + 2);
+ name2[0] = '@';
+ memcpy(name2 + 1, name, strlen(name) + 1);
+ if (data->userhook)
+ ret = data->userhook(name2, &info);
+ free(name2);
+ return ret;
+}
+
+int
+zfs_ls(device_t device, const char *path,
+ int (*hook)(const char *, const struct zfs_dirhook_info *))
+{
+ struct zfs_data *data;
+ int err;
+ int isfs;
+#if 0
+ char *label = NULL;
+
+ zfs_label(device, &label);
+ if (label)
+ printf("ZPOOL label '%s'\n",
+ label);
+#endif
+
+ data = zfs_mount(device);
+ if (!data)
+ return ZFS_ERR_BAD_FS;
+
+ data->userhook = hook;
+
+ err = dnode_get_fullpath(path, &(data->mdn), 0, &(data->dnode), &isfs, data);
+ if (err) {
+ zfs_unmount(data);
+ return err;
+ }
+ if (isfs) {
+ uint64_t childobj, headobj;
+ uint64_t snapobj;
+ dnode_end_t dn;
+ struct zfs_dirhook_info info;
+
+ fill_fs_info(&info, data->dnode, data);
+ hook("@", &info);
+
+ childobj = zfs_to_cpu64(((dsl_dir_phys_t *) DN_BONUS(&data->dnode.dn))->dd_child_dir_zapobj, data->dnode.endian);
+ headobj = zfs_to_cpu64(((dsl_dir_phys_t *) DN_BONUS(&data->dnode.dn))->dd_head_dataset_obj, data->dnode.endian);
+ err = dnode_get(&(data->mos), childobj,
+ DMU_OT_DSL_DIR_CHILD_MAP, &dn, data);
+ if (err) {
+ zfs_unmount(data);
+ return err;
+ }
+
+
+ zap_iterate(&dn, iterate_zap_fs, data);
+
+ err = dnode_get(&(data->mos), headobj, DMU_OT_DSL_DATASET, &dn, data);
+ if (err) {
+ zfs_unmount(data);
+ return err;
+ }
+
+ snapobj = zfs_to_cpu64(((dsl_dataset_phys_t *) DN_BONUS(&dn.dn))->ds_snapnames_zapobj, dn.endian);
+
+ err = dnode_get(&(data->mos), snapobj,
+ DMU_OT_DSL_DS_SNAP_MAP, &dn, data);
+ if (err) {
+ zfs_unmount(data);
+ return err;
+ }
+
+ zap_iterate(&dn, iterate_zap_snap, data);
+ } else {
+ if (data->dnode.dn.dn_type != DMU_OT_DIRECTORY_CONTENTS) {
+ zfs_unmount(data);
+ printf("not a directory\n");
+ return ZFS_ERR_BAD_FILE_TYPE;
+ }
+ zap_iterate(&(data->dnode), iterate_zap, data);
+ }
+ zfs_unmount(data);
+ return ZFS_ERR_NONE;
+}
diff --git a/fs/zfs/zfs_fletcher.c b/fs/zfs/zfs_fletcher.c
new file mode 100644
index 00000000000..08f708ca968
--- /dev/null
+++ b/fs/zfs/zfs_fletcher.c
@@ -0,0 +1,87 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2007 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#include <common.h>
+#include <malloc.h>
+#include <linux/stat.h>
+#include <linux/time.h>
+#include <linux/ctype.h>
+#include <asm/byteorder.h>
+#include "zfs_common.h"
+
+#include <zfs/zfs.h>
+#include <zfs/zio.h>
+#include <zfs/dnode.h>
+#include <zfs/uberblock_impl.h>
+#include <zfs/vdev_impl.h>
+#include <zfs/zio_checksum.h>
+#include <zfs/zap_impl.h>
+#include <zfs/zap_leaf.h>
+#include <zfs/zfs_znode.h>
+#include <zfs/dmu.h>
+#include <zfs/dmu_objset.h>
+#include <zfs/dsl_dir.h>
+#include <zfs/dsl_dataset.h>
+
+void
+fletcher_2_endian(const void *buf, uint64_t size,
+ zfs_endian_t endian,
+ zio_cksum_t *zcp)
+{
+ const uint64_t *ip = buf;
+ const uint64_t *ipend = ip + (size / sizeof(uint64_t));
+ uint64_t a0, b0, a1, b1;
+
+ for (a0 = b0 = a1 = b1 = 0; ip < ipend; ip += 2) {
+ a0 += zfs_to_cpu64(ip[0], endian);
+ a1 += zfs_to_cpu64(ip[1], endian);
+ b0 += a0;
+ b1 += a1;
+ }
+
+ zcp->zc_word[0] = cpu_to_zfs64(a0, endian);
+ zcp->zc_word[1] = cpu_to_zfs64(a1, endian);
+ zcp->zc_word[2] = cpu_to_zfs64(b0, endian);
+ zcp->zc_word[3] = cpu_to_zfs64(b1, endian);
+}
+
+void
+fletcher_4_endian(const void *buf, uint64_t size, zfs_endian_t endian,
+ zio_cksum_t *zcp)
+{
+ const uint32_t *ip = buf;
+ const uint32_t *ipend = ip + (size / sizeof(uint32_t));
+ uint64_t a, b, c, d;
+
+ for (a = b = c = d = 0; ip < ipend; ip++) {
+ a += zfs_to_cpu32(ip[0], endian);
+ b += a;
+ c += b;
+ d += c;
+ }
+
+ zcp->zc_word[0] = cpu_to_zfs64(a, endian);
+ zcp->zc_word[1] = cpu_to_zfs64(b, endian);
+ zcp->zc_word[2] = cpu_to_zfs64(c, endian);
+ zcp->zc_word[3] = cpu_to_zfs64(d, endian);
+}
diff --git a/fs/zfs/zfs_lzjb.c b/fs/zfs/zfs_lzjb.c
new file mode 100644
index 00000000000..b22d7e1ddbf
--- /dev/null
+++ b/fs/zfs/zfs_lzjb.c
@@ -0,0 +1,97 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2007 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#include <common.h>
+#include <malloc.h>
+#include <linux/stat.h>
+#include <linux/time.h>
+#include <linux/ctype.h>
+#include <asm/byteorder.h>
+#include "zfs_common.h"
+
+#include <zfs/zfs.h>
+#include <zfs/zio.h>
+#include <zfs/dnode.h>
+#include <zfs/uberblock_impl.h>
+#include <zfs/vdev_impl.h>
+#include <zfs/zio_checksum.h>
+#include <zfs/zap_impl.h>
+#include <zfs/zap_leaf.h>
+#include <zfs/zfs_znode.h>
+#include <zfs/dmu.h>
+#include <zfs/dmu_objset.h>
+#include <zfs/dsl_dir.h>
+#include <zfs/dsl_dataset.h>
+
+#define MATCH_BITS 6
+#define MATCH_MIN 3
+#define OFFSET_MASK ((1 << (16 - MATCH_BITS)) - 1)
+
+/*
+ * Decompression Entry - lzjb
+ */
+#ifndef NBBY
+#define NBBY 8
+#endif
+
+int
+lzjb_decompress(void *s_start, void *d_start, uint32_t s_len,
+ uint32_t d_len)
+{
+ uint8_t *src = s_start;
+ uint8_t *dst = d_start;
+ uint8_t *d_end = (uint8_t *) d_start + d_len;
+ uint8_t *s_end = (uint8_t *) s_start + s_len;
+ uint8_t *cpy, copymap = 0;
+ int copymask = 1 << (NBBY - 1);
+
+ while (dst < d_end && src < s_end) {
+ if ((copymask <<= 1) == (1 << NBBY)) {
+ copymask = 1;
+ copymap = *src++;
+ }
+ if (src >= s_end) {
+ printf("lzjb decompression failed\n");
+ return ZFS_ERR_BAD_FS;
+ }
+ if (copymap & copymask) {
+ int mlen = (src[0] >> (NBBY - MATCH_BITS)) + MATCH_MIN;
+ int offset = ((src[0] << NBBY) | src[1]) & OFFSET_MASK;
+ src += 2;
+ cpy = dst - offset;
+ if (src > s_end || cpy < (uint8_t *) d_start) {
+ printf("lzjb decompression failed\n");
+ return ZFS_ERR_BAD_FS;
+ }
+ while (--mlen >= 0 && dst < d_end)
+ *dst++ = *cpy++;
+ } else {
+ *dst++ = *src++;
+ }
+ }
+ if (dst < d_end) {
+ printf("lzjb decompression failed\n");
+ return ZFS_ERR_BAD_FS;
+ }
+ return ZFS_ERR_NONE;
+}
diff --git a/fs/zfs/zfs_sha256.c b/fs/zfs/zfs_sha256.c
new file mode 100644
index 00000000000..f1a4d9769c8
--- /dev/null
+++ b/fs/zfs/zfs_sha256.c
@@ -0,0 +1,148 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2007 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#include <common.h>
+#include <malloc.h>
+#include <linux/stat.h>
+#include <linux/time.h>
+#include <linux/ctype.h>
+#include <asm/byteorder.h>
+#include "zfs_common.h"
+
+#include <zfs/zfs.h>
+#include <zfs/zio.h>
+#include <zfs/dnode.h>
+#include <zfs/uberblock_impl.h>
+#include <zfs/vdev_impl.h>
+#include <zfs/zio_checksum.h>
+#include <zfs/zap_impl.h>
+#include <zfs/zap_leaf.h>
+#include <zfs/zfs_znode.h>
+#include <zfs/dmu.h>
+#include <zfs/dmu_objset.h>
+#include <zfs/dsl_dir.h>
+#include <zfs/dsl_dataset.h>
+
+/*
+ * SHA-256 checksum, as specified in FIPS 180-2, available at:
+ * http://csrc.nist.gov/cryptval
+ *
+ * This is a very compact implementation of SHA-256.
+ * It is designed to be simple and portable, not to be fast.
+ */
+
+/*
+ * The literal definitions according to FIPS180-2 would be:
+ *
+ * Ch(x, y, z) (((x) & (y)) ^ ((~(x)) & (z)))
+ * Maj(x, y, z) (((x) & (y)) | ((x) & (z)) | ((y) & (z)))
+ *
+ * We use logical equivalents which require one less op.
+ */
+#define Ch(x, y, z) ((z) ^ ((x) & ((y) ^ (z))))
+#define Maj(x, y, z) (((x) & (y)) ^ ((z) & ((x) ^ (y))))
+#define Rot32(x, s) (((x) >> s) | ((x) << (32 - s)))
+#define SIGMA0(x) (Rot32(x, 2) ^ Rot32(x, 13) ^ Rot32(x, 22))
+#define SIGMA1(x) (Rot32(x, 6) ^ Rot32(x, 11) ^ Rot32(x, 25))
+#define sigma0(x) (Rot32(x, 7) ^ Rot32(x, 18) ^ ((x) >> 3))
+#define sigma1(x) (Rot32(x, 17) ^ Rot32(x, 19) ^ ((x) >> 10))
+
+static const uint32_t SHA256_K[64] = {
+ 0x428a2f98, 0x71374491, 0xb5c0fbcf, 0xe9b5dba5,
+ 0x3956c25b, 0x59f111f1, 0x923f82a4, 0xab1c5ed5,
+ 0xd807aa98, 0x12835b01, 0x243185be, 0x550c7dc3,
+ 0x72be5d74, 0x80deb1fe, 0x9bdc06a7, 0xc19bf174,
+ 0xe49b69c1, 0xefbe4786, 0x0fc19dc6, 0x240ca1cc,
+ 0x2de92c6f, 0x4a7484aa, 0x5cb0a9dc, 0x76f988da,
+ 0x983e5152, 0xa831c66d, 0xb00327c8, 0xbf597fc7,
+ 0xc6e00bf3, 0xd5a79147, 0x06ca6351, 0x14292967,
+ 0x27b70a85, 0x2e1b2138, 0x4d2c6dfc, 0x53380d13,
+ 0x650a7354, 0x766a0abb, 0x81c2c92e, 0x92722c85,
+ 0xa2bfe8a1, 0xa81a664b, 0xc24b8b70, 0xc76c51a3,
+ 0xd192e819, 0xd6990624, 0xf40e3585, 0x106aa070,
+ 0x19a4c116, 0x1e376c08, 0x2748774c, 0x34b0bcb5,
+ 0x391c0cb3, 0x4ed8aa4a, 0x5b9cca4f, 0x682e6ff3,
+ 0x748f82ee, 0x78a5636f, 0x84c87814, 0x8cc70208,
+ 0x90befffa, 0xa4506ceb, 0xbef9a3f7, 0xc67178f2
+};
+
+static void
+SHA256Transform(uint32_t *H, const uint8_t *cp)
+{
+ uint32_t a, b, c, d, e, f, g, h, t, T1, T2, W[64];
+
+ for (t = 0; t < 16; t++, cp += 4)
+ W[t] = (cp[0] << 24) | (cp[1] << 16) | (cp[2] << 8) | cp[3];
+
+ for (t = 16; t < 64; t++)
+ W[t] = sigma1(W[t - 2]) + W[t - 7] +
+ sigma0(W[t - 15]) + W[t - 16];
+
+ a = H[0]; b = H[1]; c = H[2]; d = H[3];
+ e = H[4]; f = H[5]; g = H[6]; h = H[7];
+
+ for (t = 0; t < 64; t++) {
+ T1 = h + SIGMA1(e) + Ch(e, f, g) + SHA256_K[t] + W[t];
+ T2 = SIGMA0(a) + Maj(a, b, c);
+ h = g; g = f; f = e; e = d + T1;
+ d = c; c = b; b = a; a = T1 + T2;
+ }
+
+ H[0] += a; H[1] += b; H[2] += c; H[3] += d;
+ H[4] += e; H[5] += f; H[6] += g; H[7] += h;
+}
+
+void
+zio_checksum_SHA256(const void *buf, uint64_t size,
+ zfs_endian_t endian, zio_cksum_t *zcp)
+{
+ uint32_t H[8] = { 0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a,
+ 0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19 };
+ uint8_t pad[128];
+ unsigned padsize = size & 63;
+ unsigned i;
+
+ for (i = 0; i < size - padsize; i += 64)
+ SHA256Transform(H, (uint8_t *)buf + i);
+
+ for (i = 0; i < padsize; i++)
+ pad[i] = ((uint8_t *)buf)[i];
+
+ for (pad[padsize++] = 0x80; (padsize & 63) != 56; padsize++)
+ pad[padsize] = 0;
+
+ for (i = 0; i < 8; i++)
+ pad[padsize++] = (size << 3) >> (56 - 8 * i);
+
+ for (i = 0; i < padsize; i += 64)
+ SHA256Transform(H, pad + i);
+
+ zcp->zc_word[0] = cpu_to_zfs64((uint64_t)H[0] << 32 | H[1],
+ endian);
+ zcp->zc_word[1] = cpu_to_zfs64((uint64_t)H[2] << 32 | H[3],
+ endian);
+ zcp->zc_word[2] = cpu_to_zfs64((uint64_t)H[4] << 32 | H[5],
+ endian);
+ zcp->zc_word[3] = cpu_to_zfs64((uint64_t)H[6] << 32 | H[7],
+ endian);
+}
diff --git a/include/config_cmd_all.h b/include/config_cmd_all.h
index 73c05582021..f434cd08910 100644
--- a/include/config_cmd_all.h
+++ b/include/config_cmd_all.h
@@ -87,5 +87,6 @@
#define CONFIG_CMD_UNZIP /* unzip from memory to memory */
#define CONFIG_CMD_USB /* USB Support */
#define CONFIG_CMD_XIMG /* Load part of Multi Image */
+#define CONFIG_CMD_ZFS /* ZFS Support */
#endif /* _CONFIG_CMD_ALL_H */
diff --git a/include/zfs/dmu.h b/include/zfs/dmu.h
new file mode 100644
index 00000000000..7faa7088b55
--- /dev/null
+++ b/include/zfs/dmu.h
@@ -0,0 +1,120 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2010 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _SYS_DMU_H
+#define _SYS_DMU_H
+
+/*
+ * This file describes the interface that the DMU provides for its
+ * consumers.
+ *
+ * The DMU also interacts with the SPA. That interface is described in
+ * dmu_spa.h.
+ */
+typedef enum dmu_object_type {
+ DMU_OT_NONE,
+ /* general: */
+ DMU_OT_OBJECT_DIRECTORY, /* ZAP */
+ DMU_OT_OBJECT_ARRAY, /* UINT64 */
+ DMU_OT_PACKED_NVLIST, /* UINT8 (XDR by nvlist_pack/unpack) */
+ DMU_OT_PACKED_NVLIST_SIZE, /* UINT64 */
+ DMU_OT_BPLIST, /* UINT64 */
+ DMU_OT_BPLIST_HDR, /* UINT64 */
+ /* spa: */
+ DMU_OT_SPACE_MAP_HEADER, /* UINT64 */
+ DMU_OT_SPACE_MAP, /* UINT64 */
+ /* zil: */
+ DMU_OT_INTENT_LOG, /* UINT64 */
+ /* dmu: */
+ DMU_OT_DNODE, /* DNODE */
+ DMU_OT_OBJSET, /* OBJSET */
+ /* dsl: */
+ DMU_OT_DSL_DIR, /* UINT64 */
+ DMU_OT_DSL_DIR_CHILD_MAP, /* ZAP */
+ DMU_OT_DSL_DS_SNAP_MAP, /* ZAP */
+ DMU_OT_DSL_PROPS, /* ZAP */
+ DMU_OT_DSL_DATASET, /* UINT64 */
+ /* zpl: */
+ DMU_OT_ZNODE, /* ZNODE */
+ DMU_OT_OLDACL, /* OLD ACL */
+ DMU_OT_PLAIN_FILE_CONTENTS, /* UINT8 */
+ DMU_OT_DIRECTORY_CONTENTS, /* ZAP */
+ DMU_OT_MASTER_NODE, /* ZAP */
+ DMU_OT_UNLINKED_SET, /* ZAP */
+ /* zvol: */
+ DMU_OT_ZVOL, /* UINT8 */
+ DMU_OT_ZVOL_PROP, /* ZAP */
+ /* other; for testing only! */
+ DMU_OT_PLAIN_OTHER, /* UINT8 */
+ DMU_OT_UINT64_OTHER, /* UINT64 */
+ DMU_OT_ZAP_OTHER, /* ZAP */
+ /* new object types: */
+ DMU_OT_ERROR_LOG, /* ZAP */
+ DMU_OT_SPA_HISTORY, /* UINT8 */
+ DMU_OT_SPA_HISTORY_OFFSETS, /* spa_his_phys_t */
+ DMU_OT_POOL_PROPS, /* ZAP */
+ DMU_OT_DSL_PERMS, /* ZAP */
+ DMU_OT_ACL, /* ACL */
+ DMU_OT_SYSACL, /* SYSACL */
+ DMU_OT_FUID, /* FUID table (Packed NVLIST UINT8) */
+ DMU_OT_FUID_SIZE, /* FUID table size UINT64 */
+ DMU_OT_NEXT_CLONES, /* ZAP */
+ DMU_OT_SCRUB_QUEUE, /* ZAP */
+ DMU_OT_USERGROUP_USED, /* ZAP */
+ DMU_OT_USERGROUP_QUOTA, /* ZAP */
+ DMU_OT_USERREFS, /* ZAP */
+ DMU_OT_DDT_ZAP, /* ZAP */
+ DMU_OT_DDT_STATS, /* ZAP */
+ DMU_OT_SA, /* System attr */
+ DMU_OT_SA_MASTER_NODE, /* ZAP */
+ DMU_OT_SA_ATTR_REGISTRATION, /* ZAP */
+ DMU_OT_SA_ATTR_LAYOUTS, /* ZAP */
+ DMU_OT_NUMTYPES
+} dmu_object_type_t;
+
+typedef enum dmu_objset_type {
+ DMU_OST_NONE,
+ DMU_OST_META,
+ DMU_OST_ZFS,
+ DMU_OST_ZVOL,
+ DMU_OST_OTHER, /* For testing only! */
+ DMU_OST_ANY, /* Be careful! */
+ DMU_OST_NUMTYPES
+} dmu_objset_type_t;
+
+/*
+ * The names of zap entries in the DIRECTORY_OBJECT of the MOS.
+ */
+#define DMU_POOL_DIRECTORY_OBJECT 1
+#define DMU_POOL_CONFIG "config"
+#define DMU_POOL_ROOT_DATASET "root_dataset"
+#define DMU_POOL_SYNC_BPLIST "sync_bplist"
+#define DMU_POOL_ERRLOG_SCRUB "errlog_scrub"
+#define DMU_POOL_ERRLOG_LAST "errlog_last"
+#define DMU_POOL_SPARES "spares"
+#define DMU_POOL_DEFLATE "deflate"
+#define DMU_POOL_HISTORY "history"
+#define DMU_POOL_PROPS "pool_props"
+#define DMU_POOL_L2CACHE "l2cache"
+
+#endif /* _SYS_DMU_H */
diff --git a/include/zfs/dmu_objset.h b/include/zfs/dmu_objset.h
new file mode 100644
index 00000000000..dbd67ad00af
--- /dev/null
+++ b/include/zfs/dmu_objset.h
@@ -0,0 +1,43 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2009 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _SYS_DMU_OBJSET_H
+#define _SYS_DMU_OBJSET_H
+
+#include <zfs/zil.h>
+
+#define OBJSET_PHYS_SIZE 2048
+#define OBJSET_PHYS_SIZE_V14 1024
+
+typedef struct objset_phys {
+ dnode_phys_t os_meta_dnode;
+ zil_header_t os_zil_header;
+ uint64_t os_type;
+ uint64_t os_flags;
+ char os_pad[OBJSET_PHYS_SIZE - sizeof(dnode_phys_t)*3 -
+ sizeof(zil_header_t) - sizeof(uint64_t)*2];
+ dnode_phys_t os_userused_dnode;
+ dnode_phys_t os_groupused_dnode;
+} objset_phys_t;
+
+#endif /* _SYS_DMU_OBJSET_H */
diff --git a/include/zfs/dnode.h b/include/zfs/dnode.h
new file mode 100644
index 00000000000..09b45626b81
--- /dev/null
+++ b/include/zfs/dnode.h
@@ -0,0 +1,81 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2010 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _SYS_DNODE_H
+#define _SYS_DNODE_H
+
+#include <zfs/spa.h>
+
+/*
+ * Fixed constants.
+ */
+#define DNODE_SHIFT 9 /* 512 bytes */
+#define DN_MIN_INDBLKSHIFT 10 /* 1k */
+#define DN_MAX_INDBLKSHIFT 14 /* 16k */
+#define DNODE_BLOCK_SHIFT 14 /* 16k */
+#define DNODE_CORE_SIZE 64 /* 64 bytes for dnode sans blkptrs */
+#define DN_MAX_OBJECT_SHIFT 48 /* 256 trillion (zfs_fid_t limit) */
+#define DN_MAX_OFFSET_SHIFT 64 /* 2^64 bytes in a dnode */
+
+/*
+ * Derived constants.
+ */
+#define DNODE_SIZE (1 << DNODE_SHIFT)
+#define DN_MAX_NBLKPTR ((DNODE_SIZE - DNODE_CORE_SIZE) >> SPA_BLKPTRSHIFT)
+#define DN_MAX_BONUSLEN (DNODE_SIZE - DNODE_CORE_SIZE - (1 << SPA_BLKPTRSHIFT))
+#define DN_MAX_OBJECT (1ULL << DN_MAX_OBJECT_SHIFT)
+
+#define DNODES_PER_BLOCK_SHIFT (DNODE_BLOCK_SHIFT - DNODE_SHIFT)
+#define DNODES_PER_BLOCK (1ULL << DNODES_PER_BLOCK_SHIFT)
+#define DNODES_PER_LEVEL_SHIFT (DN_MAX_INDBLKSHIFT - SPA_BLKPTRSHIFT)
+
+#define DNODE_FLAG_SPILL_BLKPTR (1<<2)
+
+#define DN_BONUS(dnp) ((void *)((dnp)->dn_bonus + \
+ (((dnp)->dn_nblkptr - 1) * sizeof(blkptr_t))))
+
+typedef struct dnode_phys {
+ uint8_t dn_type; /* dmu_object_type_t */
+ uint8_t dn_indblkshift; /* ln2(indirect block size) */
+ uint8_t dn_nlevels; /* 1=dn_blkptr->data blocks */
+ uint8_t dn_nblkptr; /* length of dn_blkptr */
+ uint8_t dn_bonustype; /* type of data in bonus buffer */
+ uint8_t dn_checksum; /* ZIO_CHECKSUM type */
+ uint8_t dn_compress; /* ZIO_COMPRESS type */
+ uint8_t dn_flags; /* DNODE_FLAG_* */
+ uint16_t dn_datablkszsec; /* data block size in 512b sectors */
+ uint16_t dn_bonuslen; /* length of dn_bonus */
+ uint8_t dn_pad2[4];
+
+ /* accounting is protected by dn_dirty_mtx */
+ uint64_t dn_maxblkid; /* largest allocated block ID */
+ uint64_t dn_used; /* bytes (or sectors) of disk space */
+
+ uint64_t dn_pad3[4];
+
+ blkptr_t dn_blkptr[1];
+ uint8_t dn_bonus[DN_MAX_BONUSLEN - sizeof(blkptr_t)];
+ blkptr_t dn_spill;
+} dnode_phys_t;
+
+#endif /* _SYS_DNODE_H */
diff --git a/include/zfs/dsl_dataset.h b/include/zfs/dsl_dataset.h
new file mode 100644
index 00000000000..ab0ee22a364
--- /dev/null
+++ b/include/zfs/dsl_dataset.h
@@ -0,0 +1,53 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2007 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _SYS_DSL_DATASET_H
+#define _SYS_DSL_DATASET_H
+
+typedef struct dsl_dataset_phys {
+ uint64_t ds_dir_obj;
+ uint64_t ds_prev_snap_obj;
+ uint64_t ds_prev_snap_txg;
+ uint64_t ds_next_snap_obj;
+ uint64_t ds_snapnames_zapobj; /* zap obj of snaps; ==0 for snaps */
+ uint64_t ds_num_children; /* clone/snap children; ==0 for head */
+ uint64_t ds_creation_time; /* seconds since 1970 */
+ uint64_t ds_creation_txg;
+ uint64_t ds_deadlist_obj;
+ uint64_t ds_used_bytes;
+ uint64_t ds_compressed_bytes;
+ uint64_t ds_uncompressed_bytes;
+ uint64_t ds_unique_bytes; /* only relevant to snapshots */
+ /*
+ * The ds_fsid_guid is a 56-bit ID that can change to avoid
+ * collisions. The ds_guid is a 64-bit ID that will never
+ * change, so there is a small probability that it will collide.
+ */
+ uint64_t ds_fsid_guid;
+ uint64_t ds_guid;
+ uint64_t ds_flags;
+ blkptr_t ds_bp;
+ uint64_t ds_pad[8]; /* pad out to 320 bytes for good measure */
+} dsl_dataset_phys_t;
+
+#endif /* _SYS_DSL_DATASET_H */
diff --git a/include/zfs/dsl_dir.h b/include/zfs/dsl_dir.h
new file mode 100644
index 00000000000..54d66633666
--- /dev/null
+++ b/include/zfs/dsl_dir.h
@@ -0,0 +1,49 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2007 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _SYS_DSL_DIR_H
+#define _SYS_DSL_DIR_H
+
+typedef struct dsl_dir_phys {
+ uint64_t dd_creation_time; /* not actually used */
+ uint64_t dd_head_dataset_obj;
+ uint64_t dd_parent_obj;
+ uint64_t dd_clone_parent_obj;
+ uint64_t dd_child_dir_zapobj;
+ /*
+ * how much space our children are accounting for; for leaf
+ * datasets, == physical space used by fs + snaps
+ */
+ uint64_t dd_used_bytes;
+ uint64_t dd_compressed_bytes;
+ uint64_t dd_uncompressed_bytes;
+ /* Administrative quota setting */
+ uint64_t dd_quota;
+ /* Administrative reservation setting */
+ uint64_t dd_reserved;
+ uint64_t dd_props_zapobj;
+ uint64_t dd_deleg_zapobj; /* dataset permissions */
+ uint64_t dd_pad[20]; /* pad out to 256 bytes for good measure */
+} dsl_dir_phys_t;
+
+#endif /* _SYS_DSL_DIR_H */
diff --git a/include/zfs/sa_impl.h b/include/zfs/sa_impl.h
new file mode 100644
index 00000000000..4d935586c5a
--- /dev/null
+++ b/include/zfs/sa_impl.h
@@ -0,0 +1,35 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2010 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+#ifndef _SYS_SA_IMPL_H
+#define _SYS_SA_IMPL_H
+
+typedef struct sa_hdr_phys {
+ uint32_t sa_magic;
+ uint16_t sa_layout_info;
+ uint16_t sa_lengths[1];
+} sa_hdr_phys_t;
+
+#define SA_HDR_SIZE(hdr) BF32_GET_SB(hdr->sa_layout_info, 10, 16, 3, 0)
+#define SA_SIZE_OFFSET 0x8
+
+#endif /* _SYS_SA_IMPL_H */
diff --git a/include/zfs/spa.h b/include/zfs/spa.h
new file mode 100644
index 00000000000..360cf897d65
--- /dev/null
+++ b/include/zfs/spa.h
@@ -0,0 +1,292 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright (c) 2008, 2011, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#ifndef ZFS_SPA_HEADER
+#define ZFS_SPA_HEADER 1
+
+
+/*
+ * General-purpose 32-bit and 64-bit bitfield encodings.
+ */
+#define BF32_DECODE(x, low, len) P2PHASE((x) >> (low), 1U << (len))
+#define BF64_DECODE(x, low, len) P2PHASE((x) >> (low), 1ULL << (len))
+#define BF32_ENCODE(x, low, len) (P2PHASE((x), 1U << (len)) << (low))
+#define BF64_ENCODE(x, low, len) (P2PHASE((x), 1ULL << (len)) << (low))
+
+#define BF32_GET(x, low, len) BF32_DECODE(x, low, len)
+#define BF64_GET(x, low, len) BF64_DECODE(x, low, len)
+
+#define BF32_SET(x, low, len, val) \
+ ((x) ^= BF32_ENCODE((x >> low) ^ (val), low, len))
+#define BF64_SET(x, low, len, val) \
+ ((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len))
+
+#define BF32_GET_SB(x, low, len, shift, bias) \
+ ((BF32_GET(x, low, len) + (bias)) << (shift))
+#define BF64_GET_SB(x, low, len, shift, bias) \
+ ((BF64_GET(x, low, len) + (bias)) << (shift))
+
+#define BF32_SET_SB(x, low, len, shift, bias, val) \
+ BF32_SET(x, low, len, ((val) >> (shift)) - (bias))
+#define BF64_SET_SB(x, low, len, shift, bias, val) \
+ BF64_SET(x, low, len, ((val) >> (shift)) - (bias))
+
+/*
+ * We currently support nine block sizes, from 512 bytes to 128K.
+ * We could go higher, but the benefits are near-zero and the cost
+ * of COWing a giant block to modify one byte would become excessive.
+ */
+#define SPA_MINBLOCKSHIFT 9
+#define SPA_MAXBLOCKSHIFT 17
+#define SPA_MINBLOCKSIZE (1ULL << SPA_MINBLOCKSHIFT)
+#define SPA_MAXBLOCKSIZE (1ULL << SPA_MAXBLOCKSHIFT)
+
+#define SPA_BLOCKSIZES (SPA_MAXBLOCKSHIFT - SPA_MINBLOCKSHIFT + 1)
+
+/*
+ * Size of block to hold the configuration data (a packed nvlist)
+ */
+#define SPA_CONFIG_BLOCKSIZE (1 << 14)
+
+/*
+ * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB.
+ * The ASIZE encoding should be at least 64 times larger (6 more bits)
+ * to support up to 4-way RAID-Z mirror mode with worst-case gang block
+ * overhead, three DVAs per bp, plus one more bit in case we do anything
+ * else that expands the ASIZE.
+ */
+#define SPA_LSIZEBITS 16 /* LSIZE up to 32M (2^16 * 512) */
+#define SPA_PSIZEBITS 16 /* PSIZE up to 32M (2^16 * 512) */
+#define SPA_ASIZEBITS 24 /* ASIZE up to 64 times larger */
+
+/*
+ * All SPA data is represented by 128-bit data virtual addresses (DVAs).
+ * The members of the dva_t should be considered opaque outside the SPA.
+ */
+typedef struct dva {
+ uint64_t dva_word[2];
+} dva_t;
+
+/*
+ * Each block has a 256-bit checksum -- strong enough for cryptographic hashes.
+ */
+typedef struct zio_cksum {
+ uint64_t zc_word[4];
+} zio_cksum_t;
+
+/*
+ * Each block is described by its DVAs, time of birth, checksum, etc.
+ * The word-by-word, bit-by-bit layout of the blkptr is as follows:
+ *
+ * 64 56 48 40 32 24 16 8 0
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * 0 | vdev1 | GRID | ASIZE |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * 1 |G| offset1 |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * 2 | vdev2 | GRID | ASIZE |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * 3 |G| offset2 |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * 4 | vdev3 | GRID | ASIZE |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * 5 |G| offset3 |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * 6 |BDX|lvl| type | cksum | comp | PSIZE | LSIZE |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * 7 | padding |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * 8 | padding |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * 9 | physical birth txg |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * a | logical birth txg |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * b | fill count |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * c | checksum[0] |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * d | checksum[1] |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * e | checksum[2] |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ * f | checksum[3] |
+ * +-------+-------+-------+-------+-------+-------+-------+-------+
+ *
+ * Legend:
+ *
+ * vdev virtual device ID
+ * offset offset into virtual device
+ * LSIZE logical size
+ * PSIZE physical size (after compression)
+ * ASIZE allocated size (including RAID-Z parity and gang block headers)
+ * GRID RAID-Z layout information (reserved for future use)
+ * cksum checksum function
+ * comp compression function
+ * G gang block indicator
+ * B byteorder (endianness)
+ * D dedup
+ * X unused
+ * lvl level of indirection
+ * type DMU object type
+ * phys birth txg of block allocation; zero if same as logical birth txg
+ * log. birth transaction group in which the block was logically born
+ * fill count number of non-zero blocks under this bp
+ * checksum[4] 256-bit checksum of the data this bp describes
+ */
+#define SPA_BLKPTRSHIFT 7 /* blkptr_t is 128 bytes */
+#define SPA_DVAS_PER_BP 3 /* Number of DVAs in a bp */
+
+typedef struct blkptr {
+ dva_t blk_dva[SPA_DVAS_PER_BP]; /* Data Virtual Addresses */
+ uint64_t blk_prop; /* size, compression, type, etc */
+ uint64_t blk_pad[2]; /* Extra space for the future */
+ uint64_t blk_phys_birth; /* txg when block was allocated */
+ uint64_t blk_birth; /* transaction group at birth */
+ uint64_t blk_fill; /* fill count */
+ zio_cksum_t blk_cksum; /* 256-bit checksum */
+} blkptr_t;
+
+/*
+ * Macros to get and set fields in a bp or DVA.
+ */
+#define DVA_GET_ASIZE(dva) \
+ BF64_GET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0)
+#define DVA_SET_ASIZE(dva, x) \
+ BF64_SET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0, x)
+
+#define DVA_GET_GRID(dva) BF64_GET((dva)->dva_word[0], 24, 8)
+#define DVA_SET_GRID(dva, x) BF64_SET((dva)->dva_word[0], 24, 8, x)
+
+#define DVA_GET_VDEV(dva) BF64_GET((dva)->dva_word[0], 32, 32)
+#define DVA_SET_VDEV(dva, x) BF64_SET((dva)->dva_word[0], 32, 32, x)
+
+#define DVA_GET_GANG(dva) BF64_GET((dva)->dva_word[1], 63, 1)
+#define DVA_SET_GANG(dva, x) BF64_SET((dva)->dva_word[1], 63, 1, x)
+
+#define BP_GET_LSIZE(bp) \
+ BF64_GET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1)
+#define BP_SET_LSIZE(bp, x) \
+ BF64_SET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1, x)
+
+#define BP_GET_COMPRESS(bp) BF64_GET((bp)->blk_prop, 32, 8)
+#define BP_SET_COMPRESS(bp, x) BF64_SET((bp)->blk_prop, 32, 8, x)
+
+#define BP_GET_CHECKSUM(bp) BF64_GET((bp)->blk_prop, 40, 8)
+#define BP_SET_CHECKSUM(bp, x) BF64_SET((bp)->blk_prop, 40, 8, x)
+
+#define BP_GET_TYPE(bp) BF64_GET((bp)->blk_prop, 48, 8)
+#define BP_SET_TYPE(bp, x) BF64_SET((bp)->blk_prop, 48, 8, x)
+
+#define BP_GET_LEVEL(bp) BF64_GET((bp)->blk_prop, 56, 5)
+#define BP_SET_LEVEL(bp, x) BF64_SET((bp)->blk_prop, 56, 5, x)
+
+#define BP_GET_PROP_BIT_61(bp) BF64_GET((bp)->blk_prop, 61, 1)
+#define BP_SET_PROP_BIT_61(bp, x) BF64_SET((bp)->blk_prop, 61, 1, x)
+
+#define BP_GET_DEDUP(bp) BF64_GET((bp)->blk_prop, 62, 1)
+#define BP_SET_DEDUP(bp, x) BF64_SET((bp)->blk_prop, 62, 1, x)
+
+#define BP_GET_BYTEORDER(bp) (0 - BF64_GET((bp)->blk_prop, 63, 1))
+#define BP_SET_BYTEORDER(bp, x) BF64_SET((bp)->blk_prop, 63, 1, x)
+
+#define BP_PHYSICAL_BIRTH(bp) \
+ ((bp)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth)
+
+#define BP_SET_BIRTH(bp, logical, physical) \
+ { \
+ (bp)->blk_birth = (logical); \
+ (bp)->blk_phys_birth = ((logical) == (physical) ? 0 : (physical)); \
+ }
+
+#define BP_GET_ASIZE(bp) \
+ (DVA_GET_ASIZE(&(bp)->blk_dva[0]) + DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
+ DVA_GET_ASIZE(&(bp)->blk_dva[2]))
+
+#define BP_GET_UCSIZE(bp) \
+ ((BP_GET_LEVEL(bp) > 0 || dmu_ot[BP_GET_TYPE(bp)].ot_metadata) ? \
+ BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp));
+
+#define BP_GET_NDVAS(bp) \
+ (!!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \
+ !!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
+ !!DVA_GET_ASIZE(&(bp)->blk_dva[2]))
+
+#define BP_COUNT_GANG(bp) \
+ (DVA_GET_GANG(&(bp)->blk_dva[0]) + \
+ DVA_GET_GANG(&(bp)->blk_dva[1]) + \
+ DVA_GET_GANG(&(bp)->blk_dva[2]))
+
+#define DVA_EQUAL(dva1, dva2) \
+ ((dva1)->dva_word[1] == (dva2)->dva_word[1] && \
+ (dva1)->dva_word[0] == (dva2)->dva_word[0])
+
+#define BP_EQUAL(bp1, bp2) \
+ (BP_PHYSICAL_BIRTH(bp1) == BP_PHYSICAL_BIRTH(bp2) && \
+ DVA_EQUAL(&(bp1)->blk_dva[0], &(bp2)->blk_dva[0]) && \
+ DVA_EQUAL(&(bp1)->blk_dva[1], &(bp2)->blk_dva[1]) && \
+ DVA_EQUAL(&(bp1)->blk_dva[2], &(bp2)->blk_dva[2]))
+
+#define ZIO_CHECKSUM_EQUAL(zc1, zc2) \
+ (0 == (((zc1).zc_word[0] - (zc2).zc_word[0]) | \
+ ((zc1).zc_word[1] - (zc2).zc_word[1]) | \
+ ((zc1).zc_word[2] - (zc2).zc_word[2]) | \
+ ((zc1).zc_word[3] - (zc2).zc_word[3])))
+
+#define DVA_IS_VALID(dva) (DVA_GET_ASIZE(dva) != 0)
+
+#define ZIO_SET_CHECKSUM(zcp, w0, w1, w2, w3) \
+ { \
+ (zcp)->zc_word[0] = w0; \
+ (zcp)->zc_word[1] = w1; \
+ (zcp)->zc_word[2] = w2; \
+ (zcp)->zc_word[3] = w3; \
+ }
+
+#define BP_IDENTITY(bp) (&(bp)->blk_dva[0])
+#define BP_IS_GANG(bp) DVA_GET_GANG(BP_IDENTITY(bp))
+#define BP_IS_HOLE(bp) ((bp)->blk_birth == 0)
+
+/* BP_IS_RAIDZ(bp) assumes no block compression */
+#define BP_IS_RAIDZ(bp) (DVA_GET_ASIZE(&(bp)->blk_dva[0]) > \
+ BP_GET_PSIZE(bp))
+
+#define BP_ZERO(bp) \
+ { \
+ (bp)->blk_dva[0].dva_word[0] = 0; \
+ (bp)->blk_dva[0].dva_word[1] = 0; \
+ (bp)->blk_dva[1].dva_word[0] = 0; \
+ (bp)->blk_dva[1].dva_word[1] = 0; \
+ (bp)->blk_dva[2].dva_word[0] = 0; \
+ (bp)->blk_dva[2].dva_word[1] = 0; \
+ (bp)->blk_prop = 0; \
+ (bp)->blk_pad[0] = 0; \
+ (bp)->blk_pad[1] = 0; \
+ (bp)->blk_phys_birth = 0; \
+ (bp)->blk_birth = 0; \
+ (bp)->blk_fill = 0; \
+ ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0); \
+ }
+
+#define BP_SPRINTF_LEN 320
+
+#endif /* ! ZFS_SPA_HEADER */
diff --git a/include/zfs/uberblock_impl.h b/include/zfs/uberblock_impl.h
new file mode 100644
index 00000000000..806da9585c8
--- /dev/null
+++ b/include/zfs/uberblock_impl.h
@@ -0,0 +1,57 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2007 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _SYS_UBERBLOCK_IMPL_H
+#define _SYS_UBERBLOCK_IMPL_H
+
+#define UBMAX(a, b) ((a) > (b) ? (a) : (b))
+
+/*
+ * The uberblock version is incremented whenever an incompatible on-disk
+ * format change is made to the SPA, DMU, or ZAP.
+ *
+ * Note: the first two fields should never be moved. When a storage pool
+ * is opened, the uberblock must be read off the disk before the version
+ * can be checked. If the ub_version field is moved, we may not detect
+ * version mismatch. If the ub_magic field is moved, applications that
+ * expect the magic number in the first word won't work.
+ */
+#define UBERBLOCK_MAGIC 0x00bab10c /* oo-ba-bloc! */
+#define UBERBLOCK_SHIFT 10 /* up to 1K */
+
+typedef struct uberblock {
+ uint64_t ub_magic; /* UBERBLOCK_MAGIC */
+ uint64_t ub_version; /* ZFS_VERSION */
+ uint64_t ub_txg; /* txg of last sync */
+ uint64_t ub_guid_sum; /* sum of all vdev guids */
+ uint64_t ub_timestamp; /* UTC time of last sync */
+ blkptr_t ub_rootbp; /* MOS objset_phys_t */
+} uberblock_t;
+
+#define VDEV_UBERBLOCK_SHIFT(as) UBMAX(as, UBERBLOCK_SHIFT)
+#define UBERBLOCK_SIZE(as) (1ULL << VDEV_UBERBLOCK_SHIFT(as))
+
+/* Number of uberblocks that can fit in the ring at a given ashift */
+#define UBERBLOCK_COUNT(as) (VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(as))
+
+#endif /* _SYS_UBERBLOCK_IMPL_H */
diff --git a/include/zfs/vdev_impl.h b/include/zfs/vdev_impl.h
new file mode 100644
index 00000000000..3aa646ea8c8
--- /dev/null
+++ b/include/zfs/vdev_impl.h
@@ -0,0 +1,70 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2010 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _SYS_VDEV_IMPL_H
+#define _SYS_VDEV_IMPL_H
+
+#define VDEV_SKIP_SIZE (8 << 10)
+#define VDEV_BOOT_HEADER_SIZE (8 << 10)
+#define VDEV_PHYS_SIZE (112 << 10)
+#define VDEV_UBERBLOCK_RING (128 << 10)
+
+/* ZFS boot block */
+#define VDEV_BOOT_MAGIC 0x2f5b007b10cULL
+#define VDEV_BOOT_VERSION 1 /* version number */
+
+typedef struct vdev_boot_header {
+ uint64_t vb_magic; /* VDEV_BOOT_MAGIC */
+ uint64_t vb_version; /* VDEV_BOOT_VERSION */
+ uint64_t vb_offset; /* start offset (bytes) */
+ uint64_t vb_size; /* size (bytes) */
+ char vb_pad[VDEV_BOOT_HEADER_SIZE - 4 * sizeof(uint64_t)];
+} vdev_boot_header_t;
+
+typedef struct vdev_phys {
+ char vp_nvlist[VDEV_PHYS_SIZE - sizeof(zio_eck_t)];
+ zio_eck_t vp_zbt;
+} vdev_phys_t;
+
+typedef struct vdev_label {
+ char vl_pad[VDEV_SKIP_SIZE]; /* 8K */
+ vdev_boot_header_t vl_boot_header; /* 8K */
+ vdev_phys_t vl_vdev_phys; /* 112K */
+ char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */
+} vdev_label_t; /* 256K total */
+
+/*
+ * Size and offset of embedded boot loader region on each label.
+ * The total size of the first two labels plus the boot area is 4MB.
+ */
+#define VDEV_BOOT_OFFSET (2 * sizeof(vdev_label_t))
+#define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */
+
+/*
+ * Size of label regions at the start and end of each leaf device.
+ */
+#define VDEV_LABEL_START_SIZE (2 * sizeof(vdev_label_t) + VDEV_BOOT_SIZE)
+#define VDEV_LABEL_END_SIZE (2 * sizeof(vdev_label_t))
+#define VDEV_LABELS 4
+
+#endif /* _SYS_VDEV_IMPL_H */
diff --git a/include/zfs/zap_impl.h b/include/zfs/zap_impl.h
new file mode 100644
index 00000000000..5d24ab06fe5
--- /dev/null
+++ b/include/zfs/zap_impl.h
@@ -0,0 +1,111 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright (c) 2008, 2011, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#ifndef _SYS_ZAP_IMPL_H
+#define _SYS_ZAP_IMPL_H
+
+#define ZAP_MAGIC 0x2F52AB2ABULL
+
+#define ZAP_HASHBITS 28
+#define MZAP_ENT_LEN 64
+#define MZAP_NAME_LEN (MZAP_ENT_LEN - 8 - 4 - 2)
+#define MZAP_MAX_BLKSHIFT SPA_MAXBLOCKSHIFT
+#define MZAP_MAX_BLKSZ (1 << MZAP_MAX_BLKSHIFT)
+
+typedef struct mzap_ent_phys {
+ uint64_t mze_value;
+ uint32_t mze_cd;
+ uint16_t mze_pad; /* in case we want to chain them someday */
+ char mze_name[MZAP_NAME_LEN];
+} mzap_ent_phys_t;
+
+typedef struct mzap_phys {
+ uint64_t mz_block_type; /* ZBT_MICRO */
+ uint64_t mz_salt;
+ uint64_t mz_pad[6];
+ mzap_ent_phys_t mz_chunk[1];
+ /* actually variable size depending on block size */
+} mzap_phys_t;
+
+/*
+ * The (fat) zap is stored in one object. It is an array of
+ * 1<<FZAP_BLOCK_SHIFT byte blocks. The layout looks like one of:
+ *
+ * ptrtbl fits in first block:
+ * [zap_phys_t zap_ptrtbl_shift < 6] [zap_leaf_t] ...
+ *
+ * ptrtbl too big for first block:
+ * [zap_phys_t zap_ptrtbl_shift >= 6] [zap_leaf_t] [ptrtbl] ...
+ *
+ */
+
+#define ZBT_LEAF ((1ULL << 63) + 0)
+#define ZBT_HEADER ((1ULL << 63) + 1)
+#define ZBT_MICRO ((1ULL << 63) + 3)
+/* any other values are ptrtbl blocks */
+
+/*
+ * the embedded pointer table takes up half a block:
+ * block size / entry size (2^3) / 2
+ */
+#define ZAP_EMBEDDED_PTRTBL_SHIFT(zap) (FZAP_BLOCK_SHIFT(zap) - 3 - 1)
+
+/*
+ * The embedded pointer table starts half-way through the block. Since
+ * the pointer table itself is half the block, it starts at (64-bit)
+ * word number (1<<ZAP_EMBEDDED_PTRTBL_SHIFT(zap)).
+ */
+#define ZAP_EMBEDDED_PTRTBL_ENT(zap, idx) \
+ ((uint64_t *)(zap)->zap_f.zap_phys) \
+ [(idx) + (1<<ZAP_EMBEDDED_PTRTBL_SHIFT(zap))]
+
+/*
+ * TAKE NOTE:
+ * If zap_phys_t is modified, zap_byteswap() must be modified.
+ */
+typedef struct zap_phys {
+ uint64_t zap_block_type; /* ZBT_HEADER */
+ uint64_t zap_magic; /* ZAP_MAGIC */
+
+ struct zap_table_phys {
+ uint64_t zt_blk; /* starting block number */
+ uint64_t zt_numblks; /* number of blocks */
+ uint64_t zt_shift; /* bits to index it */
+ uint64_t zt_nextblk; /* next (larger) copy start block */
+ uint64_t zt_blks_copied; /* number source blocks copied */
+ } zap_ptrtbl;
+
+ uint64_t zap_freeblk; /* the next free block */
+ uint64_t zap_num_leafs; /* number of leafs */
+ uint64_t zap_num_entries; /* number of entries */
+ uint64_t zap_salt; /* salt to stir into hash function */
+ uint64_t zap_normflags; /* flags for u8_textprep_str() */
+ uint64_t zap_flags; /* zap_flag_t */
+ /*
+ * This structure is followed by padding, and then the embedded
+ * pointer table. The embedded pointer table takes up second
+ * half of the block. It is accessed using the
+ * ZAP_EMBEDDED_PTRTBL_ENT() macro.
+ */
+} zap_phys_t;
+
+#endif /* _SYS_ZAP_IMPL_H */
diff --git a/include/zfs/zap_leaf.h b/include/zfs/zap_leaf.h
new file mode 100644
index 00000000000..ed7ff62d37f
--- /dev/null
+++ b/include/zfs/zap_leaf.h
@@ -0,0 +1,103 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2007 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _SYS_ZAP_LEAF_H
+#define _SYS_ZAP_LEAF_H
+
+#define ZAP_LEAF_MAGIC 0x2AB1EAF
+
+/* chunk size = 24 bytes */
+#define ZAP_LEAF_CHUNKSIZE 24
+
+/*
+ * The amount of space within the chunk available for the array is:
+ * chunk size - space for type (1) - space for next pointer (2)
+ */
+#define ZAP_LEAF_ARRAY_BYTES (ZAP_LEAF_CHUNKSIZE - 3)
+
+typedef enum zap_chunk_type {
+ ZAP_CHUNK_FREE = 253,
+ ZAP_CHUNK_ENTRY = 252,
+ ZAP_CHUNK_ARRAY = 251,
+ ZAP_CHUNK_TYPE_MAX = 250
+} zap_chunk_type_t;
+
+/*
+ * TAKE NOTE:
+ * If zap_leaf_phys_t is modified, zap_leaf_byteswap() must be modified.
+ */
+typedef struct zap_leaf_phys {
+ struct zap_leaf_header {
+ uint64_t lh_block_type; /* ZBT_LEAF */
+ uint64_t lh_pad1;
+ uint64_t lh_prefix; /* hash prefix of this leaf */
+ uint32_t lh_magic; /* ZAP_LEAF_MAGIC */
+ uint16_t lh_nfree; /* number free chunks */
+ uint16_t lh_nentries; /* number of entries */
+ uint16_t lh_prefix_len; /* num bits used to id this */
+
+ /* above is accessable to zap, below is zap_leaf private */
+
+ uint16_t lh_freelist; /* chunk head of free list */
+ uint8_t lh_pad2[12];
+ } l_hdr; /* 2 24-byte chunks */
+
+ /*
+ * The header is followed by a hash table with
+ * ZAP_LEAF_HASH_NUMENTRIES(zap) entries. The hash table is
+ * followed by an array of ZAP_LEAF_NUMCHUNKS(zap)
+ * zap_leaf_chunk structures. These structures are accessed
+ * with the ZAP_LEAF_CHUNK() macro.
+ */
+
+ uint16_t l_hash[1];
+} zap_leaf_phys_t;
+
+typedef union zap_leaf_chunk {
+ struct zap_leaf_entry {
+ uint8_t le_type; /* always ZAP_CHUNK_ENTRY */
+ uint8_t le_int_size; /* size of ints */
+ uint16_t le_next; /* next entry in hash chain */
+ uint16_t le_name_chunk; /* first chunk of the name */
+ uint16_t le_name_length; /* bytes in name, incl null */
+ uint16_t le_value_chunk; /* first chunk of the value */
+ uint16_t le_value_length; /* value length in ints */
+ uint32_t le_cd; /* collision differentiator */
+ uint64_t le_hash; /* hash value of the name */
+ } l_entry;
+ struct zap_leaf_array {
+ uint8_t la_type; /* always ZAP_CHUNK_ARRAY */
+ union {
+ uint8_t la_array[ZAP_LEAF_ARRAY_BYTES];
+ uint64_t la_array64;
+ } __attribute__ ((packed));
+ uint16_t la_next; /* next blk or CHAIN_END */
+ } l_array;
+ struct zap_leaf_free {
+ uint8_t lf_type; /* always ZAP_CHUNK_FREE */
+ uint8_t lf_pad[ZAP_LEAF_ARRAY_BYTES];
+ uint16_t lf_next; /* next in free list, or CHAIN_END */
+ } l_free;
+} zap_leaf_chunk_t;
+
+#endif /* _SYS_ZAP_LEAF_H */
diff --git a/include/zfs/zfs.h b/include/zfs/zfs.h
new file mode 100644
index 00000000000..319ee6abc68
--- /dev/null
+++ b/include/zfs/zfs.h
@@ -0,0 +1,122 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright (c) 2007, 2011, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#ifndef GRUB_ZFS_HEADER
+#define GRUB_ZFS_HEADER 1
+
+
+/*
+ * On-disk version number.
+ */
+#define SPA_VERSION 28ULL
+
+/*
+ * The following are configuration names used in the nvlist describing a pool's
+ * configuration.
+ */
+#define ZPOOL_CONFIG_VERSION "version"
+#define ZPOOL_CONFIG_POOL_NAME "name"
+#define ZPOOL_CONFIG_POOL_STATE "state"
+#define ZPOOL_CONFIG_POOL_TXG "txg"
+#define ZPOOL_CONFIG_POOL_GUID "pool_guid"
+#define ZPOOL_CONFIG_CREATE_TXG "create_txg"
+#define ZPOOL_CONFIG_TOP_GUID "top_guid"
+#define ZPOOL_CONFIG_VDEV_TREE "vdev_tree"
+#define ZPOOL_CONFIG_TYPE "type"
+#define ZPOOL_CONFIG_CHILDREN "children"
+#define ZPOOL_CONFIG_ID "id"
+#define ZPOOL_CONFIG_GUID "guid"
+#define ZPOOL_CONFIG_PATH "path"
+#define ZPOOL_CONFIG_DEVID "devid"
+#define ZPOOL_CONFIG_METASLAB_ARRAY "metaslab_array"
+#define ZPOOL_CONFIG_METASLAB_SHIFT "metaslab_shift"
+#define ZPOOL_CONFIG_ASHIFT "ashift"
+#define ZPOOL_CONFIG_ASIZE "asize"
+#define ZPOOL_CONFIG_DTL "DTL"
+#define ZPOOL_CONFIG_STATS "stats"
+#define ZPOOL_CONFIG_WHOLE_DISK "whole_disk"
+#define ZPOOL_CONFIG_ERRCOUNT "error_count"
+#define ZPOOL_CONFIG_NOT_PRESENT "not_present"
+#define ZPOOL_CONFIG_SPARES "spares"
+#define ZPOOL_CONFIG_IS_SPARE "is_spare"
+#define ZPOOL_CONFIG_NPARITY "nparity"
+#define ZPOOL_CONFIG_PHYS_PATH "phys_path"
+#define ZPOOL_CONFIG_L2CACHE "l2cache"
+#define ZPOOL_CONFIG_HOLE_ARRAY "hole_array"
+#define ZPOOL_CONFIG_VDEV_CHILDREN "vdev_children"
+#define ZPOOL_CONFIG_IS_HOLE "is_hole"
+#define ZPOOL_CONFIG_DDT_HISTOGRAM "ddt_histogram"
+#define ZPOOL_CONFIG_DDT_OBJ_STATS "ddt_object_stats"
+#define ZPOOL_CONFIG_DDT_STATS "ddt_stats"
+/*
+ * The persistent vdev state is stored as separate values rather than a single
+ * 'vdev_state' entry. This is because a device can be in multiple states, such
+ * as offline and degraded.
+ */
+#define ZPOOL_CONFIG_OFFLINE "offline"
+#define ZPOOL_CONFIG_FAULTED "faulted"
+#define ZPOOL_CONFIG_DEGRADED "degraded"
+#define ZPOOL_CONFIG_REMOVED "removed"
+
+#define VDEV_TYPE_ROOT "root"
+#define VDEV_TYPE_MIRROR "mirror"
+#define VDEV_TYPE_REPLACING "replacing"
+#define VDEV_TYPE_RAIDZ "raidz"
+#define VDEV_TYPE_DISK "disk"
+#define VDEV_TYPE_FILE "file"
+#define VDEV_TYPE_MISSING "missing"
+#define VDEV_TYPE_HOLE "hole"
+#define VDEV_TYPE_SPARE "spare"
+#define VDEV_TYPE_L2CACHE "l2cache"
+
+/*
+ * pool state. The following states are written to disk as part of the normal
+ * SPA lifecycle: ACTIVE, EXPORTED, DESTROYED, SPARE, L2CACHE. The remaining
+ * states are software abstractions used at various levels to communicate pool
+ * state.
+ */
+typedef enum pool_state {
+ POOL_STATE_ACTIVE = 0, /* In active use */
+ POOL_STATE_EXPORTED, /* Explicitly exported */
+ POOL_STATE_DESTROYED, /* Explicitly destroyed */
+ POOL_STATE_SPARE, /* Reserved for hot spare use */
+ POOL_STATE_L2CACHE, /* Level 2 ARC device */
+ POOL_STATE_UNINITIALIZED, /* Internal spa_t state */
+ POOL_STATE_UNAVAIL, /* Internal libzfs state */
+ POOL_STATE_POTENTIALLY_ACTIVE /* Internal libzfs state */
+} pool_state_t;
+
+struct zfs_data;
+
+int zfs_fetch_nvlist(device_t dev, char **nvlist);
+int zfs_getmdnobj(device_t dev, const char *fsfilename,
+ uint64_t *mdnobj);
+
+char *zfs_nvlist_lookup_string(char *nvlist, char *name);
+char *zfs_nvlist_lookup_nvlist(char *nvlist, char *name);
+int zfs_nvlist_lookup_uint64(char *nvlist, char *name,
+ uint64_t *out);
+char *zfs_nvlist_lookup_nvlist_array(char *nvlist, char *name,
+ size_t index);
+int zfs_nvlist_lookup_nvlist_array_get_nelm(char *nvlist, char *name);
+
+#endif /* ! GRUB_ZFS_HEADER */
diff --git a/include/zfs/zfs_acl.h b/include/zfs/zfs_acl.h
new file mode 100644
index 00000000000..1b3e14ffae9
--- /dev/null
+++ b/include/zfs/zfs_acl.h
@@ -0,0 +1,55 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2007 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _SYS_FS_ZFS_ACL_H
+#define _SYS_FS_ZFS_ACL_H
+
+typedef struct zfs_oldace {
+ uint32_t z_fuid; /* "who" */
+ uint32_t z_access_mask; /* access mask */
+ uint16_t z_flags; /* flags, i.e inheritance */
+ uint16_t z_type; /* type of entry allow/deny */
+} zfs_oldace_t;
+
+#define ACE_SLOT_CNT 6
+
+typedef struct zfs_znode_acl_v0 {
+ uint64_t z_acl_extern_obj; /* ext acl pieces */
+ uint32_t z_acl_count; /* Number of ACEs */
+ uint16_t z_acl_version; /* acl version */
+ uint16_t z_acl_pad; /* pad */
+ zfs_oldace_t z_ace_data[ACE_SLOT_CNT]; /* 6 standard ACEs */
+} zfs_znode_acl_v0_t;
+
+#define ZFS_ACE_SPACE (sizeof(zfs_oldace_t) * ACE_SLOT_CNT)
+
+typedef struct zfs_znode_acl {
+ uint64_t z_acl_extern_obj; /* ext acl pieces */
+ uint32_t z_acl_size; /* Number of bytes in ACL */
+ uint16_t z_acl_version; /* acl version */
+ uint16_t z_acl_count; /* ace count */
+ uint8_t z_ace_data[ZFS_ACE_SPACE]; /* space for embedded ACEs */
+} zfs_znode_acl_t;
+
+
+#endif /* _SYS_FS_ZFS_ACL_H */
diff --git a/include/zfs/zfs_znode.h b/include/zfs/zfs_znode.h
new file mode 100644
index 00000000000..2267acafe57
--- /dev/null
+++ b/include/zfs/zfs_znode.h
@@ -0,0 +1,71 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2010 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _SYS_FS_ZFS_ZNODE_H
+#define _SYS_FS_ZFS_ZNODE_H
+
+#include <zfs/zfs_acl.h>
+
+#define MASTER_NODE_OBJ 1
+#define ZFS_ROOT_OBJ "ROOT"
+#define ZPL_VERSION_STR "VERSION"
+#define ZFS_SA_ATTRS "SA_ATTRS"
+
+#define ZPL_VERSION 5ULL
+
+#define ZFS_DIRENT_OBJ(de) BF64_GET(de, 0, 48)
+
+/*
+ * This is the persistent portion of the znode. It is stored
+ * in the "bonus buffer" of the file. Short symbolic links
+ * are also stored in the bonus buffer.
+ */
+typedef struct znode_phys {
+ uint64_t zp_atime[2]; /* 0 - last file access time */
+ uint64_t zp_mtime[2]; /* 16 - last file modification time */
+ uint64_t zp_ctime[2]; /* 32 - last file change time */
+ uint64_t zp_crtime[2]; /* 48 - creation time */
+ uint64_t zp_gen; /* 64 - generation (txg of creation) */
+ uint64_t zp_mode; /* 72 - file mode bits */
+ uint64_t zp_size; /* 80 - size of file */
+ uint64_t zp_parent; /* 88 - directory parent (`..') */
+ uint64_t zp_links; /* 96 - number of links to file */
+ uint64_t zp_xattr; /* 104 - DMU object for xattrs */
+ uint64_t zp_rdev; /* 112 - dev_t for VBLK & VCHR files */
+ uint64_t zp_flags; /* 120 - persistent flags */
+ uint64_t zp_uid; /* 128 - file owner */
+ uint64_t zp_gid; /* 136 - owning group */
+ uint64_t zp_pad[4]; /* 144 - future */
+ zfs_znode_acl_t zp_acl; /* 176 - 263 ACL */
+ /*
+ * Data may pad out any remaining bytes in the znode buffer, eg:
+ *
+ * |<---------------------- dnode_phys (512) ------------------------>|
+ * |<-- dnode (192) --->|<----------- "bonus" buffer (320) ---------->|
+ * |<---- znode (264) ---->|<---- data (56) ---->|
+ *
+ * At present, we only use this space to store symbolic links.
+ */
+} znode_phys_t;
+
+#endif /* _SYS_FS_ZFS_ZNODE_H */
diff --git a/include/zfs/zil.h b/include/zfs/zil.h
new file mode 100644
index 00000000000..036573aa7f4
--- /dev/null
+++ b/include/zfs/zil.h
@@ -0,0 +1,57 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2009 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _SYS_ZIL_H
+#define _SYS_ZIL_H
+
+/*
+ * Intent log format:
+ *
+ * Each objset has its own intent log. The log header (zil_header_t)
+ * for objset N's intent log is kept in the Nth object of the SPA's
+ * intent_log objset. The log header points to a chain of log blocks,
+ * each of which contains log records (i.e., transactions) followed by
+ * a log block trailer (zil_trailer_t). The format of a log record
+ * depends on the record (or transaction) type, but all records begin
+ * with a common structure that defines the type, length, and txg.
+ */
+
+/*
+ * Intent log header - this on disk structure holds fields to manage
+ * the log. All fields are 64 bit to easily handle cross architectures.
+ */
+typedef struct zil_header {
+ uint64_t zh_claim_txg; /* txg in which log blocks were claimed */
+ uint64_t zh_replay_seq; /* highest replayed sequence number */
+ blkptr_t zh_log; /* log chain */
+ uint64_t zh_claim_seq; /* highest claimed sequence number */
+ uint64_t zh_flags; /* header flags */
+ uint64_t zh_pad[4];
+} zil_header_t;
+
+/*
+ * zh_flags bit settings
+ */
+#define ZIL_REPLAY_NEEDED 0x1 /* replay needed - internal only */
+
+#endif /* _SYS_ZIL_H */
diff --git a/include/zfs/zio.h b/include/zfs/zio.h
new file mode 100644
index 00000000000..4adb14cb04c
--- /dev/null
+++ b/include/zfs/zio.h
@@ -0,0 +1,92 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#ifndef _ZIO_H
+#define _ZIO_H
+
+#include <zfs/spa.h>
+
+#define ZEC_MAGIC 0x210da7ab10c7a11ULL /* zio data bloc tail */
+
+typedef struct zio_eck {
+ uint64_t zec_magic; /* for validation, endianness */
+ zio_cksum_t zec_cksum; /* 256-bit checksum */
+} zio_eck_t;
+
+/*
+ * Gang block headers are self-checksumming and contain an array
+ * of block pointers.
+ */
+#define SPA_GANGBLOCKSIZE SPA_MINBLOCKSIZE
+#define SPA_GBH_NBLKPTRS ((SPA_GANGBLOCKSIZE - \
+ sizeof(zio_eck_t)) / sizeof(blkptr_t))
+#define SPA_GBH_FILLER ((SPA_GANGBLOCKSIZE - \
+ sizeof(zio_eck_t) - \
+ (SPA_GBH_NBLKPTRS * sizeof(blkptr_t))) /\
+ sizeof(uint64_t))
+
+#define ZIO_GET_IOSIZE(zio) \
+ (BP_IS_GANG((zio)->io_bp) ? \
+ SPA_GANGBLOCKSIZE : BP_GET_PSIZE((zio)->io_bp))
+
+typedef struct zio_gbh {
+ blkptr_t zg_blkptr[SPA_GBH_NBLKPTRS];
+ uint64_t zg_filler[SPA_GBH_FILLER];
+ zio_eck_t zg_tail;
+} zio_gbh_phys_t;
+
+enum zio_checksum {
+ ZIO_CHECKSUM_INHERIT = 0,
+ ZIO_CHECKSUM_ON,
+ ZIO_CHECKSUM_OFF,
+ ZIO_CHECKSUM_LABEL,
+ ZIO_CHECKSUM_GANG_HEADER,
+ ZIO_CHECKSUM_ZILOG,
+ ZIO_CHECKSUM_FLETCHER_2,
+ ZIO_CHECKSUM_FLETCHER_4,
+ ZIO_CHECKSUM_SHA256,
+ ZIO_CHECKSUM_ZILOG2,
+ ZIO_CHECKSUM_FUNCTIONS
+};
+
+#define ZIO_CHECKSUM_ON_VALUE ZIO_CHECKSUM_FLETCHER_2
+#define ZIO_CHECKSUM_DEFAULT ZIO_CHECKSUM_ON
+
+enum zio_compress {
+ ZIO_COMPRESS_INHERIT = 0,
+ ZIO_COMPRESS_ON,
+ ZIO_COMPRESS_OFF,
+ ZIO_COMPRESS_LZJB,
+ ZIO_COMPRESS_EMPTY,
+ ZIO_COMPRESS_GZIP1,
+ ZIO_COMPRESS_GZIP2,
+ ZIO_COMPRESS_GZIP3,
+ ZIO_COMPRESS_GZIP4,
+ ZIO_COMPRESS_GZIP5,
+ ZIO_COMPRESS_GZIP6,
+ ZIO_COMPRESS_GZIP7,
+ ZIO_COMPRESS_GZIP8,
+ ZIO_COMPRESS_GZIP9,
+ ZIO_COMPRESS_FUNCTIONS
+};
+
+#endif /* _ZIO_H */
diff --git a/include/zfs/zio_checksum.h b/include/zfs/zio_checksum.h
new file mode 100644
index 00000000000..0d5fce66536
--- /dev/null
+++ b/include/zfs/zio_checksum.h
@@ -0,0 +1,50 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999,2000,2001,2002,2003,2004 Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+/*
+ * Copyright 2010 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _SYS_ZIO_CHECKSUM_H
+#define _SYS_ZIO_CHECKSUM_H
+
+/*
+ * Signature for checksum functions.
+ */
+typedef void zio_checksum_t(const void *data, uint64_t size,
+ zfs_endian_t endian, zio_cksum_t *zcp);
+
+/*
+ * Information about each checksum function.
+ */
+typedef struct zio_checksum_info {
+ zio_checksum_t *ci_func; /* checksum function for each byteorder */
+ int ci_correctable; /* number of correctable bits */
+ int ci_eck; /* uses zio embedded checksum? */
+ char *ci_name; /* descriptive name */
+} zio_checksum_info_t;
+
+extern void zio_checksum_SHA256(const void *, uint64_t,
+ zfs_endian_t endian, zio_cksum_t *);
+extern void fletcher_2_endian(const void *, uint64_t, zfs_endian_t endian,
+ zio_cksum_t *);
+extern void fletcher_4_endian(const void *, uint64_t, zfs_endian_t endian,
+ zio_cksum_t *);
+
+#endif /* _SYS_ZIO_CHECKSUM_H */
diff --git a/include/zfs_common.h b/include/zfs_common.h
new file mode 100644
index 00000000000..04e73d0cd2a
--- /dev/null
+++ b/include/zfs_common.h
@@ -0,0 +1,109 @@
+/*
+ * ZFS filesystem port for Uboot by
+ * Jorgen Lundman <lundman at lundman.net>
+ *
+ * zfsfs support
+ * made from existing GRUB Sources by Sun, GNU and others.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#ifndef __ZFS_COMMON__
+#define __ZFS_COMMON__
+
+#define SECTOR_SIZE 0x200
+#define SECTOR_BITS 9
+
+
+typedef enum zfs_endian {
+ UNKNOWN_ENDIAN = -2,
+ LITTLE_ENDIAN = -1,
+ BIG_ENDIAN = 0
+} zfs_endian_t;
+
+
+/* Endian macros. */
+#define zfs_to_cpu16(x, a) (((a) == BIG_ENDIAN) ? be16_to_cpu(x) \
+ : le16_to_cpu(x))
+#define cpu_to_zfs16(x, a) (((a) == BIG_ENDIAN) ? cpu_to_be16(x) \
+ : cpu_to_le16(x))
+
+#define zfs_to_cpu32(x, a) (((a) == BIG_ENDIAN) ? be32_to_cpu(x) \
+ : le32_to_cpu(x))
+#define cpu_to_zfs32(x, a) (((a) == BIG_ENDIAN) ? cpu_to_be32(x) \
+ : cpu_to_le32(x))
+
+#define zfs_to_cpu64(x, a) (((a) == BIG_ENDIAN) ? be64_to_cpu(x) \
+ : le64_to_cpu(x))
+#define cpu_to_zfs64(x, a) (((a) == BIG_ENDIAN) ? cpu_to_be64(x) \
+ : cpu_to_le64(x))
+
+
+enum zfs_errors {
+ ZFS_ERR_NONE = 0,
+ ZFS_ERR_NOT_IMPLEMENTED_YET = -1,
+ ZFS_ERR_BAD_FS = -2,
+ ZFS_ERR_OUT_OF_MEMORY = -3,
+ ZFS_ERR_FILE_NOT_FOUND = -4,
+ ZFS_ERR_BAD_FILE_TYPE = -5,
+ ZFS_ERR_OUT_OF_RANGE = -6,
+};
+
+struct zfs_filesystem {
+
+ /* Block Device Descriptor */
+ block_dev_desc_t *dev_desc;
+};
+
+
+extern block_dev_desc_t *zfs_dev_desc;
+
+struct device_s {
+ uint64_t part_length;
+};
+typedef struct device_s *device_t;
+
+struct zfs_file {
+ device_t device;
+ uint64_t size;
+ void *data;
+ uint64_t offset;
+};
+
+typedef struct zfs_file *zfs_file_t;
+
+struct zfs_dirhook_info {
+ int dir;
+ int mtimeset;
+ time_t mtime;
+ time_t mtime2;
+};
+
+
+
+
+struct zfs_filesystem *zfsget_fs(void);
+int init_fs(block_dev_desc_t *dev_desc);
+void deinit_fs(block_dev_desc_t *dev_desc);
+int zfs_open(zfs_file_t, const char *filename);
+uint64_t zfs_read(zfs_file_t, char *buf, uint64_t len);
+struct zfs_data *zfs_mount(device_t);
+int zfs_close(zfs_file_t);
+int zfs_ls(device_t dev, const char *path,
+ int (*hook) (const char *, const struct zfs_dirhook_info *));
+int zfs_devread(int sector, int byte_offset, int byte_len, char *buf);
+int zfs_set_blk_dev(block_dev_desc_t *rbdd, int part);
+void zfs_unmount(struct zfs_data *data);
+int lzjb_decompress(void *, void *, uint32_t, uint32_t);
+#endif