d/freebsd-nq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Add libzutil for libzfs or libzpool consumers

Browse Source 
Adds a libzutil for utility functions that are common to libzfs and
libzpool consumers (most of what was in libzfs_import.c).  This
removes the need for utilities to link against both libzpool and
libzfs.

Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Don Brady <don.brady@delphix.com>
Closes #8050
This commit is contained in:
Don Brady 2018-11-05 12:22:33 -07:00 committed by Brian Behlendorf
parent 6644e5bb6e
commit e89f1295d4
39 changed files with 3725 additions and 3396 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
cmd/mount_zfs/mount_zfs.c
View File

@ -31,6 +31,7 @@
#include <sys/mntent.h>
#include <sys/stat.h>
#include <libzfs.h>
#include <libzutil.h>
#include <locale.h>
#include <getopt.h>
#include <fcntl.h>

1
cmd/zdb/Makefile.am
View File

@ -16,5 +16,4 @@ zdb_SOURCES = \
zdb_LDADD = \
$(top_builddir)/lib/libnvpair/libnvpair.la \
$(top_builddir)/lib/libzfs/libzfs.la \
$(top_builddir)/lib/libzpool/libzpool.la

13
cmd/zdb/zdb.c
View File

@ -67,7 +67,9 @@
#include <sys/dsl_crypt.h>
#include <sys/dsl_scan.h>
#include <zfs_comutil.h>
#include <libzfs.h>
#include <libnvpair.h>
#include <libzutil.h>
#include "zdb.h"
@ -106,7 +108,6 @@ typedef void object_viewer_t(objset_t *, uint64_t, void *data, size_t size);
uint64_t *zopt_object = NULL;
static unsigned zopt_objects = 0;
libzfs_handle_t *g_zfs;
uint64_t max_inflight = 1000;
static int leaked_objects = 0;
static range_tree_t *mos_refd_objs;
@ -5996,10 +5997,6 @@ main(int argc, char **argv)
spa_load_verify_dryrun = B_TRUE;
kernel_init(FREAD);
if ((g_zfs = libzfs_init()) == NULL) {
(void) fprintf(stderr, "%s", libzfs_error_init(errno));
return (1);
}
if (dump_all)
verbose = MAX(verbose, 1);
@ -6078,7 +6075,8 @@ main(int argc, char **argv)
args.path = searchdirs;
args.can_be_active = B_TRUE;
error = zpool_tryimport(g_zfs, target_pool, &cfg, &args);
error = zpool_find_config(NULL, target_pool, &cfg, &args,
&libzpool_config_ops);
if (error == 0) {
@ -6228,7 +6226,6 @@ main(int argc, char **argv)
dump_debug_buffer();
libzfs_fini(g_zfs);
kernel_fini();
return (error);

1
cmd/zed/agents/zfs_mod.c
View File

@ -73,6 +73,7 @@
#include <fcntl.h>
#include <libnvpair.h>
#include <libzfs.h>
#include <libzutil.h>
#include <limits.h>
#include <stddef.h>
#include <stdlib.h>

1
cmd/zed/zed_disk_event.c
View File

@ -21,6 +21,7 @@
#include <libnvpair.h>
#include <libudev.h>
#include <libzfs.h>
#include <libzutil.h>
#include <pthread.h>
#include <stdlib.h>
#include <string.h>

1
cmd/zfs/zfs_main.c
View File

@ -66,6 +66,7 @@
#include <libzfs_core.h>
#include <zfs_prop.h>
#include <zfs_deleg.h>
#include <libzutil.h>
#include <libuutil.h>
#ifdef HAVE_IDMAP
#include <aclutils.h>

1
cmd/zhack/Makefile.am
View File

@ -11,5 +11,4 @@ zhack_SOURCES = \
zhack_LDADD = \
$(top_builddir)/lib/libnvpair/libnvpair.la \
$(top_builddir)/lib/libzfs/libzfs.la \
$(top_builddir)/lib/libzpool/libzpool.la

13
cmd/zhack/zhack.c
View File

@ -48,12 +48,11 @@
#include <sys/zio_compress.h>
#include <sys/zfeature.h>
#include <sys/dmu_tx.h>
#include <libzfs.h>
#include <libzutil.h>
extern boolean_t zfeature_checks_disable;
const char cmdname[] = "zhack";
libzfs_handle_t *g_zfs;
static importargs_t g_importargs;
static char *g_pool;
static boolean_t g_readonly;
@ -128,20 +127,17 @@ zhack_import(char *target, boolean_t readonly)
int error;
kernel_init(readonly ? FREAD : (FREAD | FWRITE));
g_zfs = libzfs_init();
ASSERT(g_zfs != NULL);
dmu_objset_register_type(DMU_OST_ZFS, space_delta_cb);
g_readonly = readonly;
g_importargs.unique = B_TRUE;
g_importargs.can_be_active = readonly;
g_pool = strdup(target);
error = zpool_tryimport(g_zfs, target, &config, &g_importargs);
error = zpool_find_config(NULL, target, &config, &g_importargs,
&libzpool_config_ops);
if (error)
fatal(NULL, FTAG, "cannot import '%s': %s", target,
libzfs_error_description(g_zfs));
fatal(NULL, FTAG, "cannot import '%s'", target);
props = NULL;
if (readonly) {
@ -529,7 +525,6 @@ main(int argc, char **argv)
"changes may not be committed to disk\n");
}
libzfs_fini(g_zfs);
kernel_fini();
return (rv);

3
cmd/zinject/Makefile.am
View File

@ -13,5 +13,4 @@ zinject_SOURCES = \
zinject_LDADD = \
$(top_builddir)/lib/libnvpair/libnvpair.la \
$(top_builddir)/lib/libzfs/libzfs.la \
$(top_builddir)/lib/libzpool/libzpool.la
$(top_builddir)/lib/libzfs/libzfs.la

126
cmd/zinject/translate.c
View File

@ -25,8 +25,6 @@
#include <libzfs.h>
#include <sys/zfs_context.h>
#include <errno.h>
#include <fcntl.h>
#include <stdarg.h>
@ -49,9 +47,6 @@
#include "zinject.h"
extern void kernel_init(int);
extern void kernel_fini(void);
static int debug;
static void
@ -161,51 +156,32 @@ parse_pathname(const char *inpath, char *dataset, char *relpath,
}
/*
* Convert from a (dataset, path) pair into a (objset, object) pair. Note that
* we grab the object number from the inode number, since looking this up via
* libzpool is a real pain.
* Convert from a dataset to a objset id. Note that
* we grab the object number from the inode number.
*/
/* ARGSUSED */
static int
object_from_path(const char *dataset, const char *path, struct stat64 *statbuf,
zinject_record_t *record)
object_from_path(const char *dataset, uint64_t object, zinject_record_t *record)
{
objset_t *os;
int err;
zfs_handle_t *zhp;
/*
* Before doing any libzpool operations, call sync() to ensure that the
* on-disk state is consistent with the in-core state.
*/
sync();
err = dmu_objset_own(dataset, DMU_OST_ZFS, B_TRUE, B_FALSE, FTAG, &os);
if (err != 0) {
(void) fprintf(stderr, "cannot open dataset '%s': %s\n",
dataset, strerror(err));
if ((zhp = zfs_open(g_zfs, dataset, ZFS_TYPE_DATASET)) == NULL)
return (-1);
}
record->zi_objset = dmu_objset_id(os);
record->zi_object = statbuf->st_ino;
record->zi_objset = zfs_prop_get_int(zhp, ZFS_PROP_OBJSETID);
record->zi_object = object;
dmu_objset_disown(os, B_FALSE, FTAG);
zfs_close(zhp);
return (0);
}
/*
* Calculate the real range based on the type, level, and range given.
* Intialize the range based on the type, level, and range given.
*/
static int
calculate_range(const char *dataset, err_type_t type, int level, char *range,
initialize_range(err_type_t type, int level, char *range,
zinject_record_t *record)
{
objset_t *os = NULL;
dnode_t *dn = NULL;
int err;
int ret = -1;
/*
* Determine the numeric range from the string.
*/
@ -233,7 +209,7 @@ calculate_range(const char *dataset, err_type_t type, int level, char *range,
(void) fprintf(stderr, "invalid range '%s': must be "
"a numeric range of the form 'start[,end]'\n",
range);
goto out;
return (-1);
}
}
@ -253,7 +229,7 @@ calculate_range(const char *dataset, err_type_t type, int level, char *range,
if (range != NULL) {
(void) fprintf(stderr, "range cannot be specified when "
"type is 'dnode'\n");
goto out;
return (-1);
}
record->zi_start = record->zi_object * sizeof (dnode_phys_t);
@ -262,76 +238,9 @@ calculate_range(const char *dataset, err_type_t type, int level, char *range,
break;
}
/*
* Get the dnode associated with object, so we can calculate the block
* size.
*/
if ((err = dmu_objset_own(dataset, DMU_OST_ANY,
B_TRUE, B_FALSE, FTAG, &os)) != 0) {
(void) fprintf(stderr, "cannot open dataset '%s': %s\n",
dataset, strerror(err));
goto out;
}
if (record->zi_object == 0) {
dn = DMU_META_DNODE(os);
} else {
err = dnode_hold(os, record->zi_object, FTAG, &dn);
if (err != 0) {
(void) fprintf(stderr, "failed to hold dnode "
"for object %llu\n",
(u_longlong_t)record->zi_object);
goto out;
}
}
ziprintf("data shift: %d\n", (int)dn->dn_datablkshift);
ziprintf(" ind shift: %d\n", (int)dn->dn_indblkshift);
/*
* Translate range into block IDs.
*/
if (record->zi_start != 0 || record->zi_end != -1ULL) {
record->zi_start >>= dn->dn_datablkshift;
record->zi_end >>= dn->dn_datablkshift;
}
/*
* Check level, and then translate level 0 blkids into ranges
* appropriate for level of indirection.
*/
record->zi_level = level;
if (level > 0) {
ziprintf("level 0 blkid range: [%llu, %llu]\n",
record->zi_start, record->zi_end);
if (level >= dn->dn_nlevels) {
(void) fprintf(stderr, "level %d exceeds max level "
"of object (%d)\n", level, dn->dn_nlevels - 1);
goto out;
}
if (record->zi_start != 0 || record->zi_end != 0) {
int shift = dn->dn_indblkshift - SPA_BLKPTRSHIFT;
for (; level > 0; level--) {
record->zi_start >>= shift;
record->zi_end >>= shift;
}
}
}
ret = 0;
out:
if (dn) {
if (dn != DMU_META_DNODE(os))
dnode_rele(dn, FTAG);
}
if (os)
dmu_objset_disown(os, B_FALSE, FTAG);
return (ret);
return (0);
}
int
@ -343,8 +252,6 @@ translate_record(err_type_t type, const char *object, const char *range,
struct stat64 statbuf;
int ret = -1;
kernel_init(FREAD);
debug = (getenv("ZINJECT_DEBUG") != NULL);
ziprintf("translating: %s\n", object);
@ -396,16 +303,16 @@ translate_record(err_type_t type, const char *object, const char *range,
/*
* Convert (dataset, file) into (objset, object)
*/
if (object_from_path(dataset, path, &statbuf, record) != 0)
if (object_from_path(dataset, statbuf.st_ino, record) != 0)
goto err;
ziprintf("raw objset: %llu\n", record->zi_objset);
ziprintf("raw object: %llu\n", record->zi_object);
/*
* For the given object, calculate the real (type, level, range)
* For the given object, intialize the range in bytes
*/
if (calculate_range(dataset, type, level, (char *)range, record) != 0)
if (initialize_range(type, level, (char *)range, record) != 0)
goto err;
ziprintf(" objset: %llu\n", record->zi_objset);
@ -427,7 +334,6 @@ translate_record(err_type_t type, const char *object, const char *range,
ret = 0;
err:
kernel_fini();
return (ret);
}

2
cmd/zinject/zinject.c
View File

@ -561,6 +561,7 @@ register_handler(const char *pool, int flags, zinject_record_t *record,
if (ioctl(zfs_fd, ZFS_IOC_INJECT_FAULT, &zc) != 0) {
(void) fprintf(stderr, "failed to add handler: %s\n",
errno == EDOM ? "block level exceeds max level of object" :
strerror(errno));
return (1);
}
@ -853,6 +854,7 @@ main(int argc, char **argv)
break;
case 'r':
range = optarg;
flags |= ZINJECT_CALC_RANGE;
break;
case 's':
dur_secs = 1;

1
cmd/zpool/zpool_iter.c
View File

@ -36,6 +36,7 @@
#include <thread_pool.h>
#include <libzfs.h>
#include <libzutil.h>
#include <sys/zfs_context.h>
#include <sys/wait.h>

45
cmd/zpool/zpool_main.c
View File

@ -64,6 +64,7 @@
#include <math.h>
#include <libzfs.h>
#include <libzutil.h>
#include "zpool_util.h"
#include "zfs_comutil.h"
@ -2533,6 +2534,40 @@ do_import(nvlist_t *config, const char *newname, const char *mntopts,
return (ret);
}
typedef struct target_exists_args {
const char *poolname;
uint64_t poolguid;
} target_exists_args_t;
static int
name_or_guid_exists(zpool_handle_t *zhp, void *data)
{
target_exists_args_t *args = data;
nvlist_t *config = zpool_get_config(zhp, NULL);
int found = 0;
if (config == NULL)
return (0);
if (args->poolname != NULL) {
char *pool_name;
verify(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME,
&pool_name) == 0);
if (strcmp(pool_name, args->poolname) == 0)
found = 1;
} else {
uint64_t pool_guid;
verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID,
&pool_guid) == 0);
if (pool_guid == args->poolguid)
found = 1;
}
zpool_close(zhp);
return (found);
}
/*
* zpool checkpoint <pool>
* checkpoint --discard <pool>
@ -2685,6 +2720,7 @@ zpool_do_import(int argc, char **argv)
boolean_t do_rewind = B_FALSE;
boolean_t xtreme_rewind = B_FALSE;
boolean_t do_scan = B_FALSE;
boolean_t pool_exists = B_FALSE;
uint64_t pool_state, txg = -1ULL;
char *cachefile = NULL;
importargs_t idata = { 0 };
@ -2892,7 +2928,8 @@ zpool_do_import(int argc, char **argv)
/*
* User specified a name or guid. Ensure it's unique.
*/
idata.unique = B_TRUE;
target_exists_args_t search = {searchname, searchguid};
pool_exists = zpool_iter(g_zfs, name_or_guid_exists, &search);
}
/*
@ -2928,9 +2965,9 @@ zpool_do_import(int argc, char **argv)
idata.scan = do_scan;
idata.policy = policy;
pools = zpool_search_import(g_zfs, &idata);
pools = zpool_search_import(g_zfs, &idata, &libzfs_config_ops);
if (pools != NULL && idata.exists &&
if (pools != NULL && pool_exists &&
(argc == 1 || strcmp(argv[0], argv[1]) == 0)) {
(void) fprintf(stderr, gettext("cannot import '%s': "
"a pool with that name already exists\n"),
@ -2939,7 +2976,7 @@ zpool_do_import(int argc, char **argv)
"<pool | id> <newpool>' to give it a new name\n"),
"zpool import");
err = 1;
} else if (pools == NULL && idata.exists) {
} else if (pools == NULL && pool_exists) {
(void) fprintf(stderr, gettext("cannot import '%s': "
"a pool with that name is already created/imported,\n"),
argv[0]);

1
cmd/zpool/zpool_vdev.c
View File

@ -69,6 +69,7 @@
#include <fcntl.h>
#include <libintl.h>
#include <libnvpair.h>
#include <libzutil.h>
#include <limits.h>
#include <sys/spa.h>
#include <scsi/scsi.h>

1
cmd/ztest/Makefile.am
View File

@ -20,7 +20,6 @@ ztest_SOURCES = \
ztest_LDADD = \
$(top_builddir)/lib/libnvpair/libnvpair.la \
$(top_builddir)/lib/libzfs/libzfs.la \
$(top_builddir)/lib/libzpool/libzpool.la
ztest_LDADD += -lm

8
cmd/ztest/ztest.c
View File

@ -128,7 +128,7 @@
#include <sys/fs/zfs.h>
#include <zfs_fletcher.h>
#include <libnvpair.h>
#include <libzfs.h>
#include <libzutil.h>
#include <sys/crypto/icp.h>
#ifdef __GLIBC__
#include <execinfo.h> /* for backtrace() */
@ -7065,7 +7065,6 @@ make_random_props(void)
static void
ztest_import(ztest_shared_t *zs)
{
libzfs_handle_t *hdl;
importargs_t args = { 0 };
spa_t *spa;
nvlist_t *cfg = NULL;
@ -7080,14 +7079,14 @@ ztest_import(ztest_shared_t *zs)
VERIFY0(pthread_rwlock_init(&ztest_name_lock, NULL));
kernel_init(FREAD | FWRITE);
hdl = libzfs_init();
searchdirs[0] = ztest_opts.zo_dir;
args.paths = nsearch;
args.path = searchdirs;
args.can_be_active = B_FALSE;
error = zpool_tryimport(hdl, name, &cfg, &args);
error = zpool_find_config(NULL, name, &cfg, &args,
&libzpool_config_ops);
if (error)
(void) fatal(0, "No pools found\n");
@ -7097,7 +7096,6 @@ ztest_import(ztest_shared_t *zs)
1ULL << spa->spa_root_vdev->vdev_child[0]->vdev_ms_shift;
spa_close(spa, FTAG);
libzfs_fini(hdl);
kernel_fini();
if (!ztest_opts.zo_mmp_test) {

1
configure.ac
View File

@ -93,6 +93,7 @@ AC_CONFIG_FILES([
lib/libefi/Makefile
lib/libicp/Makefile
lib/libnvpair/Makefile
lib/libzutil/Makefile
lib/libtpool/Makefile
lib/libunicode/Makefile
lib/libuutil/Makefile

1
include/Makefile.am
View File

@ -16,6 +16,7 @@ USER_H = \
$(top_srcdir)/include/libzfs.h \
$(top_srcdir)/include/libzfs_core.h \
$(top_srcdir)/include/libzfs_impl.h \
$(top_srcdir)/include/libzutil.h \
$(top_srcdir)/include/thread_pool.h
EXTRA_DIST = $(COMMON_H) $(USER_H)

102
include/libzfs.h
View File

@ -54,25 +54,6 @@ extern "C" {
#define ZFS_MAXPROPLEN MAXPATHLEN
#define ZPOOL_MAXPROPLEN MAXPATHLEN
/*
* Default device paths
*/
#define DISK_ROOT "/dev"
#define UDISK_ROOT "/dev/disk"
#define ZVOL_ROOT "/dev/zvol"
/*
* Default wait time for a device name to be created.
*/
#define DISK_LABEL_WAIT (30 * 1000) /* 30 seconds */
#define IMPORT_ORDER_PREFERRED_1 1
#define IMPORT_ORDER_PREFERRED_2 2
#define IMPORT_ORDER_SCAN_OFFSET 10
#define IMPORT_ORDER_DEFAULT 100
#define DEFAULT_IMPORT_PATH_SIZE 9
extern char *zpool_default_import_path[DEFAULT_IMPORT_PATH_SIZE];
/*
* libzfs errors
*/
@ -298,15 +279,9 @@ extern nvlist_t *zpool_find_vdev(zpool_handle_t *, const char *, boolean_t *,
boolean_t *, boolean_t *);
extern nvlist_t *zpool_find_vdev_by_physpath(zpool_handle_t *, const char *,
boolean_t *, boolean_t *, boolean_t *);
extern int zpool_label_disk_wait(char *, int);
extern int zpool_label_disk(libzfs_handle_t *, zpool_handle_t *, char *);
extern uint64_t zpool_vdev_path_to_guid(zpool_handle_t *zhp, const char *path);
int zfs_dev_is_dm(char *dev_name);
int zfs_dev_is_whole_disk(char *dev_name);
char *zfs_get_underlying_path(char *dev_name);
char *zfs_get_enclosure_sysfs_path(char *dev_name);
const char *zpool_get_state_str(zpool_handle_t *);
/*
@ -386,7 +361,6 @@ extern zpool_status_t zpool_get_status(zpool_handle_t *, char **,
zpool_errata_t *);
extern zpool_status_t zpool_import_status(nvlist_t *, char **,
zpool_errata_t *);
extern void zpool_dump_ddt(const ddt_stat_t *dds, const ddt_histogram_t *ddh);
/*
* Statistics and configuration functions.
@ -407,32 +381,6 @@ extern int zpool_import_props(libzfs_handle_t *, nvlist_t *, const char *,
nvlist_t *, int);
extern void zpool_print_unsup_feat(nvlist_t *config);
/*
* Search for pools to import
*/
typedef struct importargs {
char **path; /* a list of paths to search */
int paths; /* number of paths to search */
char *poolname; /* name of a pool to find */
uint64_t guid; /* guid of a pool to find */
char *cachefile; /* cachefile to use for import */
int can_be_active : 1; /* can the pool be active? */
int unique : 1; /* does 'poolname' already exist? */
int exists : 1; /* set on return if pool already exists */
int scan : 1; /* prefer scanning to libblkid cache */
nvlist_t *policy; /* load policy (max txg, rewind, etc.) */
} importargs_t;
extern nvlist_t *zpool_search_import(libzfs_handle_t *, importargs_t *);
extern int zpool_tryimport(libzfs_handle_t *hdl, char *target,
nvlist_t **configp, importargs_t *args);
/* legacy pool search routines */
extern nvlist_t *zpool_find_import(libzfs_handle_t *, int, char **);
extern nvlist_t *zpool_find_import_cached(libzfs_handle_t *, const char *,
char *, uint64_t);
/*
* Miscellaneous pool functions
*/
@ -451,8 +399,6 @@ extern char *zpool_vdev_name(libzfs_handle_t *, zpool_handle_t *, nvlist_t *,
int name_flags);
extern int zpool_upgrade(zpool_handle_t *, uint64_t);
extern int zpool_get_history(zpool_handle_t *, nvlist_t **);
extern int zpool_history_unpack(char *, uint64_t, uint64_t *,
nvlist_t ***, uint_t *);
extern int zpool_events_next(libzfs_handle_t *, nvlist_t **, int *, unsigned,
int);
extern int zpool_events_clear(libzfs_handle_t *, int *);
@ -780,10 +726,6 @@ extern boolean_t zfs_dataset_exists(libzfs_handle_t *, const char *,
zfs_type_t);
extern int zfs_spa_version(zfs_handle_t *, int *);
extern boolean_t zfs_bookmark_exists(const char *path);
extern int zfs_append_partition(char *path, size_t max_len);
extern int zfs_resolve_shortname(const char *name, char *path, size_t pathlen);
extern int zfs_strcmp_pathname(char *name, char *cmp_name, int wholedisk);
extern int zfs_path_order(char *path, int *order);
/*
* Mount support functions.
@ -819,33 +761,6 @@ extern int zfs_unshareall(zfs_handle_t *);
extern int zfs_deleg_share_nfs(libzfs_handle_t *, char *, char *, char *,
void *, void *, int, zfs_share_op_t);
/*
* Formats for iostat numbers. Examples: "12K", "30ms", "4B", "2321234", "-".
*
* ZFS_NICENUM_1024: Print kilo, mega, tera, peta, exa..
* ZFS_NICENUM_BYTES: Print single bytes ("13B"), kilo, mega, tera...
* ZFS_NICENUM_TIME: Print nanosecs, microsecs, millisecs, seconds...
* ZFS_NICENUM_RAW: Print the raw number without any formatting
* ZFS_NICENUM_RAWTIME: Same as RAW, but print dashes ('-') for zero.
*/
enum zfs_nicenum_format {
ZFS_NICENUM_1024 = 0,
ZFS_NICENUM_BYTES = 1,
ZFS_NICENUM_TIME = 2,
ZFS_NICENUM_RAW = 3,
ZFS_NICENUM_RAWTIME = 4
};
/*
* Utility function to convert a number to a human-readable form.
*/
extern void zfs_nicenum(uint64_t, char *, size_t);
extern void zfs_nicenum_format(uint64_t num, char *buf, size_t buflen,
enum zfs_nicenum_format type);
extern void zfs_nicetime(uint64_t, char *, size_t);
extern void zfs_nicebytes(uint64_t, char *, size_t);
extern int zfs_nicestrtonum(libzfs_handle_t *, const char *, uint64_t *);
/*
@ -874,7 +789,6 @@ extern int zpool_in_use(libzfs_handle_t *, int, pool_state_t *, char **,
/*
* Label manipulation.
*/
extern int zpool_read_label(int, nvlist_t **, int *);
extern int zpool_clear_label(int);
/*
@ -893,22 +807,6 @@ int zfs_smb_acl_rename(libzfs_handle_t *, char *, char *, char *, char *);
extern int zpool_enable_datasets(zpool_handle_t *, const char *, int);
extern int zpool_disable_datasets(zpool_handle_t *, boolean_t);
/*
* Support for Linux libudev derived persistent device strings
*/
extern boolean_t is_mpath_whole_disk(const char *);
extern void update_vdev_config_dev_strs(nvlist_t *);
extern char *zfs_strip_partition(char *);
extern char *zfs_strip_partition_path(char *);
#ifdef HAVE_LIBUDEV
struct udev_device;
extern boolean_t udev_is_mpath(struct udev_device *dev);
extern int zfs_device_get_devid(struct udev_device *, char *, size_t);
extern int zfs_device_get_physical(struct udev_device *, char *, size_t);
#endif
extern int zfs_remap_indirects(libzfs_handle_t *hdl, const char *);
#ifdef __cplusplus

150
include/libzutil.h Normal file
View File

@ -0,0 +1,150 @@
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2018 by Delphix. All rights reserved.
*/
#ifndef _LIBZUTIL_H
#define _LIBZUTIL_H
#include <sys/nvpair.h>
#include <sys/fs/zfs.h>
#ifdef __cplusplus
extern "C" {
#endif
/*
* Default wait time for a device name to be created.
*/
#define DISK_LABEL_WAIT (30 * 1000) /* 30 seconds */
/*
* Pool Config Operations
*
* These are specific to the library libzfs or libzpool instance.
*/
typedef nvlist_t *refresh_config_func_t(void *, nvlist_t *);
typedef int pool_active_func_t(void *, const char *, uint64_t, boolean_t *);
typedef const struct pool_config_ops {
refresh_config_func_t *pco_refresh_config;
pool_active_func_t *pco_pool_active;
} pool_config_ops_t;
/*
* An instance of pool_config_ops_t is expected in the caller's binary.
*/
extern const pool_config_ops_t libzfs_config_ops;
extern const pool_config_ops_t libzpool_config_ops;
typedef struct importargs {
char **path; /* a list of paths to search */
int paths; /* number of paths to search */
const char *poolname; /* name of a pool to find */
uint64_t guid; /* guid of a pool to find */
const char *cachefile; /* cachefile to use for import */
boolean_t can_be_active; /* can the pool be active? */
boolean_t scan; /* prefer scanning to libblkid cache */
nvlist_t *policy; /* load policy (max txg, rewind, etc.) */
} importargs_t;
extern nvlist_t *zpool_search_import(void *, importargs_t *,
const pool_config_ops_t *);
extern int zpool_find_config(void *, const char *, nvlist_t **, importargs_t *,
const pool_config_ops_t *);
extern const char * const * zpool_default_search_paths(size_t *count);
extern int zpool_read_label(int, nvlist_t **, int *);
extern int zpool_label_disk_wait(const char *, int);
struct udev_device;
extern int zfs_device_get_devid(struct udev_device *, char *, size_t);
extern int zfs_device_get_physical(struct udev_device *, char *, size_t);
extern void update_vdev_config_dev_strs(nvlist_t *);
/*
* Default device paths
*/
#define DISK_ROOT "/dev"
#define UDISK_ROOT "/dev/disk"
#define ZVOL_ROOT "/dev/zvol"
extern int zfs_append_partition(char *path, size_t max_len);
extern int zfs_resolve_shortname(const char *name, char *path, size_t pathlen);
extern char *zfs_strip_partition(char *);
extern char *zfs_strip_partition_path(char *);
extern int zfs_strcmp_pathname(const char *, const char *, int);
extern int zfs_dev_is_dm(const char *);
extern int zfs_dev_is_whole_disk(const char *);
extern char *zfs_get_underlying_path(const char *);
extern char *zfs_get_enclosure_sysfs_path(const char *);
#ifdef HAVE_LIBUDEV
extern boolean_t is_mpath_whole_disk(const char *);
#else
#define is_mpath_whole_disk(path) (B_FALSE);
#endif
/*
* Formats for iostat numbers. Examples: "12K", "30ms", "4B", "2321234", "-".
*
* ZFS_NICENUM_1024: Print kilo, mega, tera, peta, exa..
* ZFS_NICENUM_BYTES: Print single bytes ("13B"), kilo, mega, tera...
* ZFS_NICENUM_TIME: Print nanosecs, microsecs, millisecs, seconds...
* ZFS_NICENUM_RAW: Print the raw number without any formatting
* ZFS_NICENUM_RAWTIME: Same as RAW, but print dashes ('-') for zero.
*/
enum zfs_nicenum_format {
ZFS_NICENUM_1024 = 0,
ZFS_NICENUM_BYTES = 1,
ZFS_NICENUM_TIME = 2,
ZFS_NICENUM_RAW = 3,
ZFS_NICENUM_RAWTIME = 4
};
/*
* Convert a number to a human-readable form.
*/
extern void zfs_nicebytes(uint64_t, char *, size_t);
extern void zfs_nicenum(uint64_t, char *, size_t);
extern void zfs_nicenum_format(uint64_t, char *, size_t,
enum zfs_nicenum_format);
extern void zfs_nicetime(uint64_t, char *, size_t);
#define nicenum(num, buf, size) zfs_nicenum(num, buf, size)
extern void zpool_dump_ddt(const ddt_stat_t *, const ddt_histogram_t *);
extern int zpool_history_unpack(char *, uint64_t, uint64_t *, nvlist_t ***,
uint_t *);
#ifdef __cplusplus
}
#endif
#endif /* _LIBZUTIL_H */

1
include/sys/zfs_context.h
View File

@ -663,7 +663,6 @@ extern void random_init(void);
extern void random_fini(void);
struct spa;
extern void nicenum(uint64_t num, char *buf, size_t);
extern void show_pool_stats(struct spa *);
extern int set_global_var(char *arg);

1
include/sys/zfs_ioctl.h
View File

@ -372,6 +372,7 @@ typedef struct zinject_record {
#define ZINJECT_NULL 0x1
#define ZINJECT_FLUSH_ARC 0x2
#define ZINJECT_UNLOAD_SPA 0x4
#define ZINJECT_CALC_RANGE 0x8
#define ZEVENT_NONE 0x0
#define ZEVENT_NONBLOCK 0x1

2
lib/Makefile.am
View File

@ -1,6 +1,6 @@
# NB: GNU Automake Manual, Chapter 8.3.5: Libtool Convenience Libraries
# These six libraries are intermediary build components.
SUBDIRS = libavl libefi libicp libshare libspl libtpool libunicode
SUBDIRS = libavl libefi libicp libshare libspl libtpool libzutil libunicode
# These four libraries, which are installed as the final build product,
# incorporate the six convenience libraries given above.

7
lib/libzfs/Makefile.am
View File

@ -55,14 +55,13 @@ nodist_libzfs_la_SOURCES = \
$(KERNEL_C)
libzfs_la_LIBADD = \
$(top_builddir)/lib/libefi/libefi.la \
$(top_builddir)/lib/libnvpair/libnvpair.la \
$(top_builddir)/lib/libshare/libshare.la \
$(top_builddir)/lib/libtpool/libtpool.la \
$(top_builddir)/lib/libuutil/libuutil.la \
$(top_builddir)/lib/libzfs_core/libzfs_core.la
$(top_builddir)/lib/libzfs_core/libzfs_core.la \
$(top_builddir)/lib/libzutil/libzutil.la
libzfs_la_LIBADD += -lm $(LIBBLKID) $(LIBUDEV) $(LIBSSL)
libzfs_la_LIBADD += -lm $(LIBSSL)
libzfs_la_LDFLAGS = -version-info 2:0:0
EXTRA_DIST = $(libzfs_pc_DATA) $(USER_C)

2
lib/libzfs/libzfs_dataset.c
View File

@ -36,7 +36,6 @@
#include <ctype.h>
#include <errno.h>
#include <libintl.h>
#include <math.h>
#include <stdio.h>
#include <stdlib.h>
#include <strings.h>
@ -61,6 +60,7 @@
#include <sys/zap.h>
#include <sys/dsl_crypt.h>
#include <libzfs.h>
#include <libzutil.h>
#include "zfs_namecheck.h"
#include "zfs_prop.h"

2259
lib/libzfs/libzfs_import.c
View File

File diff suppressed because it is too large Load Diff

1
lib/libzfs/libzfs_iter.c
View File

@ -33,6 +33,7 @@
#include <stddef.h>
#include <libintl.h>
#include <libzfs.h>
#include <libzutil.h>
#include <sys/mntent.h>
#include "libzfs_impl.h"

403
lib/libzfs/libzfs_pool.c
View File

@ -29,10 +29,8 @@
* Copyright (c) 2017, Intel Corporation.
*/
#include <ctype.h>
#include <errno.h>
#include <devid.h>
#include <fcntl.h>
#include <libintl.h>
#include <stdio.h>
#include <stdlib.h>
@ -47,6 +45,7 @@
#include <sys/zfs_ioctl.h>
#include <sys/vdev_disk.h>
#include <dlfcn.h>
#include <libzutil.h>
#include "zfs_namecheck.h"
#include "zfs_prop.h"
@ -3697,80 +3696,6 @@ set_path(zpool_handle_t *zhp, nvlist_t *nv, const char *path)
}
#endif /* sun */
/*
* Remove partition suffix from a vdev path. Partition suffixes may take three
* forms: "-partX", "pX", or "X", where X is a string of digits. The second
* case only occurs when the suffix is preceded by a digit, i.e. "md0p0" The
* third case only occurs when preceded by a string matching the regular
* expression "^([hsv]|xv)d[a-z]+", i.e. a scsi, ide, virtio or xen disk.
*
* caller must free the returned string
*/
char *
zfs_strip_partition(char *path)
{
char *tmp = strdup(path);
char *part = NULL, *d = NULL;
if (!tmp)
return (NULL);
if ((part = strstr(tmp, "-part")) && part != tmp) {
d = part + 5;
} else if ((part = strrchr(tmp, 'p')) &&
part > tmp + 1 && isdigit(*(part-1))) {
d = part + 1;
} else if ((tmp[0] == 'h' || tmp[0] == 's' || tmp[0] == 'v') &&
tmp[1] == 'd') {
for (d = &tmp[2]; isalpha(*d); part = ++d) { }
} else if (strncmp("xvd", tmp, 3) == 0) {
for (d = &tmp[3]; isalpha(*d); part = ++d) { }
}
if (part && d && *d != '\0') {
for (; isdigit(*d); d++) { }
if (*d == '\0')
*part = '\0';
}
return (tmp);
}
/*
* Same as zfs_strip_partition, but allows "/dev/" to be in the pathname
*
* path: /dev/sda1
* returns: /dev/sda
*
* Returned string must be freed.
*/
char *
zfs_strip_partition_path(char *path)
{
char *newpath = strdup(path);
char *sd_offset;
char *new_sd;
if (!newpath)
return (NULL);
/* Point to "sda1" part of "/dev/sda1" */
sd_offset = strrchr(newpath, '/') + 1;
/* Get our new name "sda" */
new_sd = zfs_strip_partition(sd_offset);
if (!new_sd) {
free(newpath);
return (NULL);
}
/* Paste the "sda" where "sda1" was */
strlcpy(sd_offset, new_sd, strlen(sd_offset) + 1);
/* Free temporary "sda" */
free(new_sd);
return (newpath);
}
#define PATH_BUF_LEN 64
/*
@ -4133,54 +4058,6 @@ get_history(zpool_handle_t *zhp, char *buf, uint64_t *off, uint64_t *len)
return (0);
}
/*
* Process the buffer of nvlists, unpacking and storing each nvlist record
* into 'records'. 'leftover' is set to the number of bytes that weren't
* processed as there wasn't a complete record.
*/
int
zpool_history_unpack(char *buf, uint64_t bytes_read, uint64_t *leftover,
nvlist_t ***records, uint_t *numrecords)
{
uint64_t reclen;
nvlist_t *nv;
int i;
void *tmp;
while (bytes_read > sizeof (reclen)) {
/* get length of packed record (stored as little endian) */
for (i = 0, reclen = 0; i < sizeof (reclen); i++)
reclen += (uint64_t)(((uchar_t *)buf)[i]) << (8*i);
if (bytes_read < sizeof (reclen) + reclen)
break;
/* unpack record */
if (nvlist_unpack(buf + sizeof (reclen), reclen, &nv, 0) != 0)
return (ENOMEM);
bytes_read -= sizeof (reclen) + reclen;
buf += sizeof (reclen) + reclen;
/* add record to nvlist array */
(*numrecords)++;
if (ISP2(*numrecords + 1)) {
tmp = realloc(*records,
*numrecords * 2 * sizeof (nvlist_t *));
if (tmp == NULL) {
nvlist_free(nv);
(*numrecords)--;
return (ENOMEM);
}
*records = tmp;
}
(*records)[*numrecords - 1] = nv;
}
*leftover = bytes_read;
return (0);
}
/*
* Retrieve the command history of a pool.
*/
@ -4669,281 +4546,3 @@ zpool_label_disk(libzfs_handle_t *hdl, zpool_handle_t *zhp, char *name)
return (0);
}
/*
* Allocate and return the underlying device name for a device mapper device.
* If a device mapper device maps to multiple devices, return the first device.
*
* For example, dm_name = "/dev/dm-0" could return "/dev/sda". Symlinks to a
* DM device (like /dev/disk/by-vdev/A0) are also allowed.
*
* Returns device name, or NULL on error or no match. If dm_name is not a DM
* device then return NULL.
*
* NOTE: The returned name string must be *freed*.
*/
char *
dm_get_underlying_path(char *dm_name)
{
DIR *dp = NULL;
struct dirent *ep;
char *realp;
char *tmp = NULL;
char *path = NULL;
char *dev_str;
int size;
if (dm_name == NULL)
return (NULL);
/* dm name may be a symlink (like /dev/disk/by-vdev/A0) */
realp = realpath(dm_name, NULL);
if (realp == NULL)
return (NULL);
/*
* If they preface 'dev' with a path (like "/dev") then strip it off.
* We just want the 'dm-N' part.
*/
tmp = strrchr(realp, '/');
if (tmp != NULL)
dev_str = tmp + 1; /* +1 since we want the chr after '/' */
else
dev_str = tmp;
size = asprintf(&tmp, "/sys/block/%s/slaves/", dev_str);
if (size == -1 || !tmp)
goto end;
dp = opendir(tmp);
if (dp == NULL)
goto end;
/* Return first sd* entry in /sys/block/dm-N/slaves/ */
while ((ep = readdir(dp))) {
if (ep->d_type != DT_DIR) { /* skip "." and ".." dirs */
size = asprintf(&path, "/dev/%s", ep->d_name);
break;
}
}
end:
if (dp != NULL)
closedir(dp);
free(tmp);
free(realp);
return (path);
}
/*
* Return 1 if device is a device mapper or multipath device.
* Return 0 if not.
*/
int
zfs_dev_is_dm(char *dev_name)
{
char *tmp;
tmp = dm_get_underlying_path(dev_name);
if (tmp == NULL)
return (0);
free(tmp);
return (1);
}
/*
* By "whole disk" we mean an entire physical disk (something we can
* label, toggle the write cache on, etc.) as opposed to the full
* capacity of a pseudo-device such as lofi or did. We act as if we
* are labeling the disk, which should be a pretty good test of whether
* it's a viable device or not. Returns B_TRUE if it is and B_FALSE if
* it isn't.
*/
int
zfs_dev_is_whole_disk(char *dev_name)
{
struct dk_gpt *label;
int fd;
if ((fd = open(dev_name, O_RDONLY | O_DIRECT)) < 0)
return (0);
if (efi_alloc_and_init(fd, EFI_NUMPAR, &label) != 0) {
(void) close(fd);
return (0);
}
efi_free(label);
(void) close(fd);
return (1);
}
/*
* Lookup the underlying device for a device name
*
* Often you'll have a symlink to a device, a partition device,
* or a multipath device, and want to look up the underlying device.
* This function returns the underlying device name. If the device
* name is already the underlying device, then just return the same
* name. If the device is a DM device with multiple underlying devices
* then return the first one.
*
* For example:
*
* 1. /dev/disk/by-id/ata-QEMU_HARDDISK_QM00001 -> ../../sda
* dev_name: /dev/disk/by-id/ata-QEMU_HARDDISK_QM00001
* returns: /dev/sda
*
* 2. /dev/mapper/mpatha (made up of /dev/sda and /dev/sdb)
* dev_name: /dev/mapper/mpatha
* returns: /dev/sda (first device)
*
* 3. /dev/sda (already the underlying device)
* dev_name: /dev/sda
* returns: /dev/sda
*
* 4. /dev/dm-3 (mapped to /dev/sda)
* dev_name: /dev/dm-3
* returns: /dev/sda
*
* 5. /dev/disk/by-id/scsi-0QEMU_drive-scsi0-0-0-0-part9 -> ../../sdb9
* dev_name: /dev/disk/by-id/scsi-0QEMU_drive-scsi0-0-0-0-part9
* returns: /dev/sdb
*
* 6. /dev/disk/by-uuid/5df030cf-3cd9-46e4-8e99-3ccb462a4e9a -> ../dev/sda2
* dev_name: /dev/disk/by-uuid/5df030cf-3cd9-46e4-8e99-3ccb462a4e9a
* returns: /dev/sda
*
* Returns underlying device name, or NULL on error or no match.
*
* NOTE: The returned name string must be *freed*.
*/
char *
zfs_get_underlying_path(char *dev_name)
{
char *name = NULL;
char *tmp;
if (dev_name == NULL)
return (NULL);
tmp = dm_get_underlying_path(dev_name);
/* dev_name not a DM device, so just un-symlinkize it */
if (tmp == NULL)
tmp = realpath(dev_name, NULL);
if (tmp != NULL) {
name = zfs_strip_partition_path(tmp);
free(tmp);
}
return (name);
}
/*
* Given a dev name like "sda", return the full enclosure sysfs path to
* the disk. You can also pass in the name with "/dev" prepended
* to it (like /dev/sda).
*
* For example, disk "sda" in enclosure slot 1:
* dev: "sda"
* returns: "/sys/class/enclosure/1:0:3:0/Slot 1"
*
* 'dev' must be a non-devicemapper device.
*
* Returned string must be freed.
*/
char *
zfs_get_enclosure_sysfs_path(char *dev_name)
{
DIR *dp = NULL;
struct dirent *ep;
char buf[MAXPATHLEN];
char *tmp1 = NULL;
char *tmp2 = NULL;
char *tmp3 = NULL;
char *path = NULL;
size_t size;
int tmpsize;
if (dev_name == NULL)
return (NULL);
/* If they preface 'dev' with a path (like "/dev") then strip it off */
tmp1 = strrchr(dev_name, '/');
if (tmp1 != NULL)
dev_name = tmp1 + 1; /* +1 since we want the chr after '/' */
tmpsize = asprintf(&tmp1, "/sys/block/%s/device", dev_name);
if (tmpsize == -1 || tmp1 == NULL) {
tmp1 = NULL;
goto end;
}
dp = opendir(tmp1);
if (dp == NULL) {
tmp1 = NULL; /* To make free() at the end a NOP */
goto end;
}
/*
* Look though all sysfs entries in /sys/block/<dev>/device for
* the enclosure symlink.
*/
while ((ep = readdir(dp))) {
/* Ignore everything that's not our enclosure_device link */
if (strstr(ep->d_name, "enclosure_device") == NULL)
continue;
if (asprintf(&tmp2, "%s/%s", tmp1, ep->d_name) == -1 ||
tmp2 == NULL)
break;
size = readlink(tmp2, buf, sizeof (buf));
/* Did readlink fail or crop the link name? */
if (size == -1 || size >= sizeof (buf)) {
free(tmp2);
tmp2 = NULL; /* To make free() at the end a NOP */
break;
}
/*
* We got a valid link. readlink() doesn't terminate strings
* so we have to do it.
*/
buf[size] = '\0';
/*
* Our link will look like:
*
* "../../../../port-11:1:2/..STUFF../enclosure/1:0:3:0/SLOT 1"
*
* We want to grab the "enclosure/1:0:3:0/SLOT 1" part
*/
tmp3 = strstr(buf, "enclosure");
if (tmp3 == NULL)
break;
if (asprintf(&path, "/sys/class/%s", tmp3) == -1) {
/* If asprintf() fails, 'path' is undefined */
path = NULL;
break;
}
if (path == NULL)
break;
}
end:
free(tmp2);
free(tmp1);
if (dp != NULL)
closedir(dp);
return (path);
}

1
lib/libzfs/libzfs_sendrecv.c
View File

@ -54,6 +54,7 @@
#include <libzfs.h>
#include <libzfs_core.h>
#include <libzutil.h>
#include "zfs_namecheck.h"
#include "zfs_prop.h"

66
lib/libzfs/libzfs_status.c
View File

@ -42,6 +42,7 @@
*/
#include <libzfs.h>
#include <libzutil.h>
#include <string.h>
#include <unistd.h>
#include <sys/systeminfo.h>
@ -425,68 +426,3 @@ zpool_import_status(nvlist_t *config, char **msgid, zpool_errata_t *errata)
return (ret);
}
static void
dump_ddt_stat(const ddt_stat_t *dds, int h)
{
char refcnt[6];
char blocks[6], lsize[6], psize[6], dsize[6];
char ref_blocks[6], ref_lsize[6], ref_psize[6], ref_dsize[6];
if (dds == NULL || dds->dds_blocks == 0)
return;
if (h == -1)
(void) strcpy(refcnt, "Total");
else
zfs_nicenum(1ULL << h, refcnt, sizeof (refcnt));
zfs_nicenum(dds->dds_blocks, blocks, sizeof (blocks));
zfs_nicebytes(dds->dds_lsize, lsize, sizeof (lsize));
zfs_nicebytes(dds->dds_psize, psize, sizeof (psize));
zfs_nicebytes(dds->dds_dsize, dsize, sizeof (dsize));
zfs_nicenum(dds->dds_ref_blocks, ref_blocks, sizeof (ref_blocks));
zfs_nicebytes(dds->dds_ref_lsize, ref_lsize, sizeof (ref_lsize));
zfs_nicebytes(dds->dds_ref_psize, ref_psize, sizeof (ref_psize));
zfs_nicebytes(dds->dds_ref_dsize, ref_dsize, sizeof (ref_dsize));
(void) printf("%6s %6s %5s %5s %5s %6s %5s %5s %5s\n",
refcnt,
blocks, lsize, psize, dsize,
ref_blocks, ref_lsize, ref_psize, ref_dsize);
}
/*
* Print the DDT histogram and the column totals.
*/
void
zpool_dump_ddt(const ddt_stat_t *dds_total, const ddt_histogram_t *ddh)
{
int h;
(void) printf("\n");
(void) printf("bucket "
" allocated "
" referenced \n");
(void) printf("______ "
"______________________________ "
"______________________________\n");
(void) printf("%6s %6s %5s %5s %5s %6s %5s %5s %5s\n",
"refcnt",
"blocks", "LSIZE", "PSIZE", "DSIZE",
"blocks", "LSIZE", "PSIZE", "DSIZE");
(void) printf("%6s %6s %5s %5s %5s %6s %5s %5s %5s\n",
"------",
"------", "-----", "-----", "-----",
"------", "-----", "-----", "-----");
for (h = 0; h < 64; h++)
dump_ddt_stat(&ddh->ddh_stat[h], h);
dump_ddt_stat(dds_total, -1);
(void) printf("\n");
}

335
lib/libzfs/libzfs_util.c
View File

@ -39,7 +39,6 @@
#include <stdlib.h>
#include <strings.h>
#include <unistd.h>
#include <ctype.h>
#include <math.h>
#include <sys/stat.h>
#include <sys/mnttab.h>
@ -54,6 +53,7 @@
#include "zfs_prop.h"
#include "zfeature_common.h"
#include <zfs_fletcher.h>
#include <libzutil.h>
int
libzfs_errno(libzfs_handle_t *hdl)
@ -677,135 +677,6 @@ zfs_strdup(libzfs_handle_t *hdl, const char *str)
return (ret);
}
/*
* Convert a number to an appropriately human-readable output.
*/
void
zfs_nicenum_format(uint64_t num, char *buf, size_t buflen,
enum zfs_nicenum_format format)
{
uint64_t n = num;
int index = 0;
const char *u;
const char *units[3][7] = {
[ZFS_NICENUM_1024] = {"", "K", "M", "G", "T", "P", "E"},
[ZFS_NICENUM_BYTES] = {"B", "K", "M", "G", "T", "P", "E"},
[ZFS_NICENUM_TIME] = {"ns", "us", "ms", "s", "?", "?", "?"}
};
const int units_len[] = {[ZFS_NICENUM_1024] = 6,
[ZFS_NICENUM_BYTES] = 6,
[ZFS_NICENUM_TIME] = 4};
const int k_unit[] = { [ZFS_NICENUM_1024] = 1024,
[ZFS_NICENUM_BYTES] = 1024,
[ZFS_NICENUM_TIME] = 1000};
double val;
if (format == ZFS_NICENUM_RAW) {
snprintf(buf, buflen, "%llu", (u_longlong_t)num);
return;
} else if (format == ZFS_NICENUM_RAWTIME && num > 0) {
snprintf(buf, buflen, "%llu", (u_longlong_t)num);
return;
} else if (format == ZFS_NICENUM_RAWTIME && num == 0) {
snprintf(buf, buflen, "%s", "-");
return;
}
while (n >= k_unit[format] && index < units_len[format]) {
n /= k_unit[format];
index++;
}
u = units[format][index];
/* Don't print zero latencies since they're invalid */
if ((format == ZFS_NICENUM_TIME) && (num == 0)) {
(void) snprintf(buf, buflen, "-");
} else if ((index == 0) || ((num %
(uint64_t)powl(k_unit[format], index)) == 0)) {
/*
* If this is an even multiple of the base, always display
* without any decimal precision.
*/
(void) snprintf(buf, buflen, "%llu%s", (u_longlong_t)n, u);
} else {
/*
* We want to choose a precision that reflects the best choice
* for fitting in 5 characters. This can get rather tricky when
* we have numbers that are very close to an order of magnitude.
* For example, when displaying 10239 (which is really 9.999K),
* we want only a single place of precision for 10.0K. We could
* develop some complex heuristics for this, but it's much
* easier just to try each combination in turn.
*/
int i;
for (i = 2; i >= 0; i--) {
val = (double)num /
(uint64_t)powl(k_unit[format], index);
/*
* Don't print floating point values for time. Note,
* we use floor() instead of round() here, since
* round can result in undesirable results. For
* example, if "num" is in the range of
* 999500-999999, it will print out "1000us". This
* doesn't happen if we use floor().
*/
if (format == ZFS_NICENUM_TIME) {
if (snprintf(buf, buflen, "%d%s",
(unsigned int) floor(val), u) <= 5)
break;
} else {
if (snprintf(buf, buflen, "%.*f%s", i,
val, u) <= 5)
break;
}
}
}
}
/*
* Convert a number to an appropriately human-readable output.
*/
void
zfs_nicenum(uint64_t num, char *buf, size_t buflen)
{
zfs_nicenum_format(num, buf, buflen, ZFS_NICENUM_1024);
}
/*
* Convert a time to an appropriately human-readable output.
* @num: Time in nanoseconds
*/
void
zfs_nicetime(uint64_t num, char *buf, size_t buflen)
{
zfs_nicenum_format(num, buf, buflen, ZFS_NICENUM_TIME);
}
/*
* Print out a raw number with correct column spacing
*/
void
zfs_niceraw(uint64_t num, char *buf, size_t buflen)
{
zfs_nicenum_format(num, buf, buflen, ZFS_NICENUM_RAW);
}
/*
* Convert a number of bytes to an appropriately human-readable output.
*/
void
zfs_nicebytes(uint64_t num, char *buf, size_t buflen)
{
zfs_nicenum_format(num, buf, buflen, ZFS_NICENUM_BYTES);
}
void
libzfs_print_on_error(libzfs_handle_t *hdl, boolean_t printerr)
{
@ -1232,210 +1103,6 @@ zfs_path_to_zhandle(libzfs_handle_t *hdl, char *path, zfs_type_t argtype)
return (zfs_open(hdl, entry.mnt_special, ZFS_TYPE_FILESYSTEM));
}
/*
* Append partition suffix to an otherwise fully qualified device path.
* This is used to generate the name the full path as its stored in
* ZPOOL_CONFIG_PATH for whole disk devices. On success the new length
* of 'path' will be returned on error a negative value is returned.
*/
int
zfs_append_partition(char *path, size_t max_len)
{
int len = strlen(path);
if ((strncmp(path, UDISK_ROOT, strlen(UDISK_ROOT)) == 0) ||
(strncmp(path, ZVOL_ROOT, strlen(ZVOL_ROOT)) == 0)) {
if (len + 6 >= max_len)
return (-1);
(void) strcat(path, "-part1");
len += 6;
} else {
if (len + 2 >= max_len)
return (-1);
if (isdigit(path[len-1])) {
(void) strcat(path, "p1");
len += 2;
} else {
(void) strcat(path, "1");
len += 1;
}
}
return (len);
}
/*
* Given a shorthand device name check if a file by that name exists in any
* of the 'zpool_default_import_path' or ZPOOL_IMPORT_PATH directories. If
* one is found, store its fully qualified path in the 'path' buffer passed
* by the caller and return 0, otherwise return an error.
*/
int
zfs_resolve_shortname(const char *name, char *path, size_t len)
{
int i, error = -1;
char *dir, *env, *envdup;
env = getenv("ZPOOL_IMPORT_PATH");
errno = ENOENT;
if (env) {
envdup = strdup(env);
dir = strtok(envdup, ":");
while (dir && error) {
(void) snprintf(path, len, "%s/%s", dir, name);
error = access(path, F_OK);
dir = strtok(NULL, ":");
}
free(envdup);
} else {
for (i = 0; i < DEFAULT_IMPORT_PATH_SIZE && error < 0; i++) {
(void) snprintf(path, len, "%s/%s",
zpool_default_import_path[i], name);
error = access(path, F_OK);
}
}
return (error ? ENOENT : 0);
}
/*
* Given a shorthand device name look for a match against 'cmp_name'. This
* is done by checking all prefix expansions using either the default
* 'zpool_default_import_paths' or the ZPOOL_IMPORT_PATH environment
* variable. Proper partition suffixes will be appended if this is a
* whole disk. When a match is found 0 is returned otherwise ENOENT.
*/
static int
zfs_strcmp_shortname(char *name, char *cmp_name, int wholedisk)
{
int path_len, cmp_len, i = 0, error = ENOENT;
char *dir, *env, *envdup = NULL;
char path_name[MAXPATHLEN];
cmp_len = strlen(cmp_name);
env = getenv("ZPOOL_IMPORT_PATH");
if (env) {
envdup = strdup(env);
dir = strtok(envdup, ":");
} else {
dir = zpool_default_import_path[i];
}
while (dir) {
/* Trim trailing directory slashes from ZPOOL_IMPORT_PATH */
while (dir[strlen(dir)-1] == '/')
dir[strlen(dir)-1] = '\0';
path_len = snprintf(path_name, MAXPATHLEN, "%s/%s", dir, name);
if (wholedisk)
path_len = zfs_append_partition(path_name, MAXPATHLEN);
if ((path_len == cmp_len) && strcmp(path_name, cmp_name) == 0) {
error = 0;
break;
}
if (env) {
dir = strtok(NULL, ":");
} else if (++i < DEFAULT_IMPORT_PATH_SIZE) {
dir = zpool_default_import_path[i];
} else {
dir = NULL;
}
}
if (env)
free(envdup);
return (error);
}
/*
* Given either a shorthand or fully qualified path name look for a match
* against 'cmp'. The passed name will be expanded as needed for comparison
* purposes and redundant slashes stripped to ensure an accurate match.
*/
int
zfs_strcmp_pathname(char *name, char *cmp, int wholedisk)
{
int path_len, cmp_len;
char path_name[MAXPATHLEN];
char cmp_name[MAXPATHLEN];
char *dir, *dup;
/* Strip redundant slashes if one exists due to ZPOOL_IMPORT_PATH */
memset(cmp_name, 0, MAXPATHLEN);
dup = strdup(cmp);
dir = strtok(dup, "/");
while (dir) {
strlcat(cmp_name, "/", sizeof (cmp_name));
strlcat(cmp_name, dir, sizeof (cmp_name));
dir = strtok(NULL, "/");
}
free(dup);
if (name[0] != '/')
return (zfs_strcmp_shortname(name, cmp_name, wholedisk));
(void) strlcpy(path_name, name, MAXPATHLEN);
path_len = strlen(path_name);
cmp_len = strlen(cmp_name);
if (wholedisk) {
path_len = zfs_append_partition(path_name, MAXPATHLEN);
if (path_len == -1)
return (ENOMEM);
}
if ((path_len != cmp_len) || strcmp(path_name, cmp_name))
return (ENOENT);
return (0);
}
/*
* Given a full path to a device determine if that device appears in the
* import search path. If it does return the first match and store the
* index in the passed 'order' variable, otherwise return an error.
*/
int
zfs_path_order(char *name, int *order)
{
int i = 0, error = ENOENT;
char *dir, *env, *envdup;
env = getenv("ZPOOL_IMPORT_PATH");
if (env) {
envdup = strdup(env);
dir = strtok(envdup, ":");
while (dir) {
if (strncmp(name, dir, strlen(dir)) == 0) {
*order = i;
error = 0;
break;
}
dir = strtok(NULL, ":");
i++;
}
free(envdup);
} else {
for (i = 0; i < DEFAULT_IMPORT_PATH_SIZE; i++) {
if (strncmp(name, zpool_default_import_path[i],
strlen(zpool_default_import_path[i])) == 0) {
*order = i;
error = 0;
break;
}
}
}
return (error);
}
/*
* Initialize the zc_nvlist_dst member to prepare for receiving an nvlist from
* an ioctl().

5
lib/libzpool/Makefile.am
View File

@ -188,11 +188,10 @@ nodist_libzpool_la_SOURCES = \
$(LUA_C)
libzpool_la_LIBADD = \
$(top_builddir)/lib/libavl/libavl.la \
$(top_builddir)/lib/libicp/libicp.la \
$(top_builddir)/lib/libnvpair/libnvpair.la \
$(top_builddir)/lib/libspl/libspl.la \
$(top_builddir)/lib/libunicode/libunicode.la
$(top_builddir)/lib/libunicode/libunicode.la \
$(top_builddir)/lib/libzutil/libzutil.la
libzpool_la_LIBADD += $(ZLIB) -ldl
libzpool_la_LDFLAGS = -pthread -version-info 2:0:0

155
lib/libzpool/util.c
View File

@ -34,112 +34,14 @@
#include <sys/spa.h>
#include <sys/fs/zfs.h>
#include <sys/refcount.h>
#include <sys/zfs_ioctl.h>
#include <dlfcn.h>
#include <libzutil.h>
/*
* Routines needed by more than one client of libzpool.
*/
/* The largest suffix that can fit, aka an exabyte (2^60 / 10^18) */
#define INDEX_MAX (6)
/* Verify INDEX_MAX fits */
CTASSERT_GLOBAL(INDEX_MAX * 10 < sizeof (uint64_t) * 8);
void
nicenum_scale(uint64_t n, size_t units, char *buf, size_t buflen,
uint32_t flags)
{
uint64_t divamt = 1024;
uint64_t divisor = 1;
int index = 0;
int rc = 0;
char u;
if (units == 0)
units = 1;
if (n > 0) {
n *= units;
if (n < units)
goto overflow;
}
if (flags & NN_DIVISOR_1000)
divamt = 1000;
/*
* This tries to find the suffix S(n) such that
* S(n) <= n < S(n+1), where S(n) = 2^(n*10) | 10^(3*n)
* (i.e. 1024/1000, 1,048,576/1,000,000, etc). Stop once S(n)
* is the largest prefix supported (i.e. don't bother computing
* and checking S(n+1). Since INDEX_MAX should be the largest
* suffix that fits (currently an exabyte), S(INDEX_MAX + 1) is
* never checked as it would overflow.
*/
while (index < INDEX_MAX) {
uint64_t newdiv = divisor * divamt;
/* CTASSERT() guarantee these never trip */
VERIFY3U(newdiv, >=, divamt);
VERIFY3U(newdiv, >=, divisor);
if (n < newdiv)
break;
divisor = newdiv;
index++;
}
u = " KMGTPE"[index];
if (index == 0) {
rc = snprintf(buf, buflen, "%llu", (u_longlong_t)n);
} else if (n % divisor == 0) {
/*
* If this is an even multiple of the base, always display
* without any decimal precision.
*/
rc = snprintf(buf, buflen, "%llu%c",
(u_longlong_t)(n / divisor), u);
} else {
/*
* We want to choose a precision that reflects the best choice
* for fitting in 5 characters. This can get rather tricky
* when we have numbers that are very close to an order of
* magnitude. For example, when displaying 10239 (which is
* really 9.999K), we want only a single place of precision
* for 10.0K. We could develop some complex heuristics for
* this, but it's much easier just to try each combination
* in turn.
*/
int i;
for (i = 2; i >= 0; i--) {
if ((rc = snprintf(buf, buflen, "%.*f%c", i,
(double)n / divisor, u)) <= 5)
break;
}
}
if (rc + 1 > buflen || rc < 0)
goto overflow;
return;
overflow:
/* prefer a more verbose message if possible */
if (buflen > 10)
(void) strlcpy(buf, "<overflow>", buflen);
else
(void) strlcpy(buf, "??", buflen);
}
void
nicenum(uint64_t num, char *buf, size_t buflen)
{
nicenum_scale(num, 1, buf, buflen, 0);
}
static void
show_vdev_stats(const char *desc, const char *ctype, nvlist_t *nv, int indent)
{
@ -300,3 +202,56 @@ set_global_var(char *arg)
return (0);
}
static nvlist_t *
refresh_config(void *unused, nvlist_t *tryconfig)
{
return (spa_tryimport(tryconfig));
}
static int
pool_active(void *unused, const char *name, uint64_t guid,
boolean_t *isactive)
{
zfs_cmd_t *zcp;
nvlist_t *innvl;
char *packed = NULL;
size_t size = 0;
int fd, ret;
/*
* Use ZFS_IOC_POOL_SYNC to confirm if a pool is active
*/
fd = open("/dev/zfs", O_RDWR);
if (fd < 0)
return (-1);
zcp = umem_zalloc(sizeof (zfs_cmd_t), UMEM_NOFAIL);
innvl = fnvlist_alloc();
fnvlist_add_boolean_value(innvl, "force", B_FALSE);
(void) strlcpy(zcp->zc_name, name, sizeof (zcp->zc_name));
packed = fnvlist_pack(innvl, &size);
zcp->zc_nvlist_src = (uint64_t)(uintptr_t)packed;
zcp->zc_nvlist_src_size = size;
ret = ioctl(fd, ZFS_IOC_POOL_SYNC, zcp);
fnvlist_pack_free(packed, size);
free((void *)(uintptr_t)zcp->zc_nvlist_dst);
nvlist_free(innvl);
umem_free(zcp, sizeof (zfs_cmd_t));
(void) close(fd);
*isactive = (ret == 0);
return (0);
}
const pool_config_ops_t libzpool_config_ops = {
.pco_refresh_config = refresh_config,
.pco_pool_active = pool_active,
};

27
lib/libzutil/Makefile.am Normal file
View File

@ -0,0 +1,27 @@
include $(top_srcdir)/config/Rules.am
# Suppress unused but set variable warnings often due to ASSERTs
AM_CFLAGS += $(NO_UNUSED_BUT_SET_VARIABLE)
DEFAULT_INCLUDES += \
-I$(top_srcdir)/include \
-I$(top_srcdir)/lib/libspl/include
noinst_LTLIBRARIES = libzutil.la
USER_C = \
zutil_device_path.c \
zutil_import.c \
zutil_nicenum.c \
zutil_pool.c
nodist_libzutil_la_SOURCES = $(USER_C)
libzutil_la_LIBADD = \
$(top_builddir)/lib/libavl/libavl.la \
$(top_builddir)/lib/libefi/libefi.la \
$(top_builddir)/lib/libtpool/libtpool.la
libzutil_la_LIBADD += -lm $(LIBBLKID) $(LIBUDEV)
EXTRA_DIST = $(USER_C)

625
lib/libzutil/zutil_device_path.c Normal file
View File

@ -0,0 +1,625 @@
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
*/
#include <ctype.h>
#include <errno.h>
#include <dirent.h>
#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <sys/efi_partition.h>
#include <libzutil.h>
#ifdef HAVE_LIBUDEV
#include <libudev.h>
#endif
/*
* Append partition suffix to an otherwise fully qualified device path.
* This is used to generate the name the full path as its stored in
* ZPOOL_CONFIG_PATH for whole disk devices. On success the new length
* of 'path' will be returned on error a negative value is returned.
*/
int
zfs_append_partition(char *path, size_t max_len)
{
int len = strlen(path);
if ((strncmp(path, UDISK_ROOT, strlen(UDISK_ROOT)) == 0) ||
(strncmp(path, ZVOL_ROOT, strlen(ZVOL_ROOT)) == 0)) {
if (len + 6 >= max_len)
return (-1);
(void) strcat(path, "-part1");
len += 6;
} else {
if (len + 2 >= max_len)
return (-1);
if (isdigit(path[len-1])) {
(void) strcat(path, "p1");
len += 2;
} else {
(void) strcat(path, "1");
len += 1;
}
}
return (len);
}
/*
* Given a shorthand device name check if a file by that name exists in any
* of the 'zpool_default_import_path' or ZPOOL_IMPORT_PATH directories. If
* one is found, store its fully qualified path in the 'path' buffer passed
* by the caller and return 0, otherwise return an error.
*/
int
zfs_resolve_shortname(const char *name, char *path, size_t len)
{
int i, error = -1;
char *dir, *env, *envdup;
env = getenv("ZPOOL_IMPORT_PATH");
errno = ENOENT;
if (env) {
envdup = strdup(env);
dir = strtok(envdup, ":");
while (dir && error) {
(void) snprintf(path, len, "%s/%s", dir, name);
error = access(path, F_OK);
dir = strtok(NULL, ":");
}
free(envdup);
} else {
const char * const *zpool_default_import_path;
size_t count;
zpool_default_import_path = zpool_default_search_paths(&count);
for (i = 0; i < count && error < 0; i++) {
(void) snprintf(path, len, "%s/%s",
zpool_default_import_path[i], name);
error = access(path, F_OK);
}
}
return (error ? ENOENT : 0);
}
/*
* Given a shorthand device name look for a match against 'cmp_name'. This
* is done by checking all prefix expansions using either the default
* 'zpool_default_import_paths' or the ZPOOL_IMPORT_PATH environment
* variable. Proper partition suffixes will be appended if this is a
* whole disk. When a match is found 0 is returned otherwise ENOENT.
*/
static int
zfs_strcmp_shortname(const char *name, const char *cmp_name, int wholedisk)
{
int path_len, cmp_len, i = 0, error = ENOENT;
char *dir, *env, *envdup = NULL;
char path_name[MAXPATHLEN];
const char * const *zpool_default_import_path;
size_t count;
zpool_default_import_path = zpool_default_search_paths(&count);
cmp_len = strlen(cmp_name);
env = getenv("ZPOOL_IMPORT_PATH");
if (env) {
envdup = strdup(env);
dir = strtok(envdup, ":");
} else {
dir = (char *)zpool_default_import_path[i];
}
while (dir) {
/* Trim trailing directory slashes from ZPOOL_IMPORT_PATH */
if (env) {
while (dir[strlen(dir)-1] == '/')
dir[strlen(dir)-1] = '\0';
}
path_len = snprintf(path_name, MAXPATHLEN, "%s/%s", dir, name);
if (wholedisk)
path_len = zfs_append_partition(path_name, MAXPATHLEN);
if ((path_len == cmp_len) && strcmp(path_name, cmp_name) == 0) {
error = 0;
break;
}
if (env) {
dir = strtok(NULL, ":");
} else if (++i < count) {
dir = (char *)zpool_default_import_path[i];
} else {
dir = NULL;
}
}
if (env)
free(envdup);
return (error);
}
/*
* Given either a shorthand or fully qualified path name look for a match
* against 'cmp'. The passed name will be expanded as needed for comparison
* purposes and redundant slashes stripped to ensure an accurate match.
*/
int
zfs_strcmp_pathname(const char *name, const char *cmp, int wholedisk)
{
int path_len, cmp_len;
char path_name[MAXPATHLEN];
char cmp_name[MAXPATHLEN];
char *dir, *dup;
/* Strip redundant slashes if one exists due to ZPOOL_IMPORT_PATH */
memset(cmp_name, 0, MAXPATHLEN);
dup = strdup(cmp);
dir = strtok(dup, "/");
while (dir) {
strlcat(cmp_name, "/", sizeof (cmp_name));
strlcat(cmp_name, dir, sizeof (cmp_name));
dir = strtok(NULL, "/");
}
free(dup);
if (name[0] != '/')
return (zfs_strcmp_shortname(name, cmp_name, wholedisk));
(void) strlcpy(path_name, name, MAXPATHLEN);
path_len = strlen(path_name);
cmp_len = strlen(cmp_name);
if (wholedisk) {
path_len = zfs_append_partition(path_name, MAXPATHLEN);
if (path_len == -1)
return (ENOMEM);
}
if ((path_len != cmp_len) || strcmp(path_name, cmp_name))
return (ENOENT);
return (0);
}
/*
* Allocate and return the underlying device name for a device mapper device.
* If a device mapper device maps to multiple devices, return the first device.
*
* For example, dm_name = "/dev/dm-0" could return "/dev/sda". Symlinks to a
* DM device (like /dev/disk/by-vdev/A0) are also allowed.
*
* Returns device name, or NULL on error or no match. If dm_name is not a DM
* device then return NULL.
*
* NOTE: The returned name string must be *freed*.
*/
static char *
dm_get_underlying_path(const char *dm_name)
{
DIR *dp = NULL;
struct dirent *ep;
char *realp;
char *tmp = NULL;
char *path = NULL;
char *dev_str;
int size;
if (dm_name == NULL)
return (NULL);
/* dm name may be a symlink (like /dev/disk/by-vdev/A0) */
realp = realpath(dm_name, NULL);
if (realp == NULL)
return (NULL);
/*
* If they preface 'dev' with a path (like "/dev") then strip it off.
* We just want the 'dm-N' part.
*/
tmp = strrchr(realp, '/');
if (tmp != NULL)
dev_str = tmp + 1; /* +1 since we want the chr after '/' */
else
dev_str = tmp;
size = asprintf(&tmp, "/sys/block/%s/slaves/", dev_str);
if (size == -1 || !tmp)
goto end;
dp = opendir(tmp);
if (dp == NULL)
goto end;
/* Return first sd* entry in /sys/block/dm-N/slaves/ */
while ((ep = readdir(dp))) {
if (ep->d_type != DT_DIR) { /* skip "." and ".." dirs */
size = asprintf(&path, "/dev/%s", ep->d_name);
break;
}
}
end:
if (dp != NULL)
closedir(dp);
free(tmp);
free(realp);
return (path);
}
/*
* Return 1 if device is a device mapper or multipath device.
* Return 0 if not.
*/
int
zfs_dev_is_dm(const char *dev_name)
{
char *tmp;
tmp = dm_get_underlying_path(dev_name);
if (tmp == NULL)
return (0);
free(tmp);
return (1);
}
/*
* By "whole disk" we mean an entire physical disk (something we can
* label, toggle the write cache on, etc.) as opposed to the full
* capacity of a pseudo-device such as lofi or did. We act as if we
* are labeling the disk, which should be a pretty good test of whether
* it's a viable device or not. Returns B_TRUE if it is and B_FALSE if
* it isn't.
*/
int
zfs_dev_is_whole_disk(const char *dev_name)
{
struct dk_gpt *label;
int fd;
if ((fd = open(dev_name, O_RDONLY | O_DIRECT)) < 0)
return (0);
if (efi_alloc_and_init(fd, EFI_NUMPAR, &label) != 0) {
(void) close(fd);
return (0);
}
efi_free(label);
(void) close(fd);
return (1);
}
/*
* Lookup the underlying device for a device name
*
* Often you'll have a symlink to a device, a partition device,
* or a multipath device, and want to look up the underlying device.
* This function returns the underlying device name. If the device
* name is already the underlying device, then just return the same
* name. If the device is a DM device with multiple underlying devices
* then return the first one.
*
* For example:
*
* 1. /dev/disk/by-id/ata-QEMU_HARDDISK_QM00001 -> ../../sda
* dev_name: /dev/disk/by-id/ata-QEMU_HARDDISK_QM00001
* returns: /dev/sda
*
* 2. /dev/mapper/mpatha (made up of /dev/sda and /dev/sdb)
* dev_name: /dev/mapper/mpatha
* returns: /dev/sda (first device)
*
* 3. /dev/sda (already the underlying device)
* dev_name: /dev/sda
* returns: /dev/sda
*
* 4. /dev/dm-3 (mapped to /dev/sda)
* dev_name: /dev/dm-3
* returns: /dev/sda
*
* 5. /dev/disk/by-id/scsi-0QEMU_drive-scsi0-0-0-0-part9 -> ../../sdb9
* dev_name: /dev/disk/by-id/scsi-0QEMU_drive-scsi0-0-0-0-part9
* returns: /dev/sdb
*
* 6. /dev/disk/by-uuid/5df030cf-3cd9-46e4-8e99-3ccb462a4e9a -> ../dev/sda2
* dev_name: /dev/disk/by-uuid/5df030cf-3cd9-46e4-8e99-3ccb462a4e9a
* returns: /dev/sda
*
* Returns underlying device name, or NULL on error or no match.
*
* NOTE: The returned name string must be *freed*.
*/
char *
zfs_get_underlying_path(const char *dev_name)
{
char *name = NULL;
char *tmp;
if (dev_name == NULL)
return (NULL);
tmp = dm_get_underlying_path(dev_name);
/* dev_name not a DM device, so just un-symlinkize it */
if (tmp == NULL)
tmp = realpath(dev_name, NULL);
if (tmp != NULL) {
name = zfs_strip_partition_path(tmp);
free(tmp);
}
return (name);
}
/*
* Given a dev name like "sda", return the full enclosure sysfs path to
* the disk. You can also pass in the name with "/dev" prepended
* to it (like /dev/sda).
*
* For example, disk "sda" in enclosure slot 1:
* dev: "sda"
* returns: "/sys/class/enclosure/1:0:3:0/Slot 1"
*
* 'dev' must be a non-devicemapper device.
*
* Returned string must be freed.
*/
char *
zfs_get_enclosure_sysfs_path(const char *dev_name)
{
DIR *dp = NULL;
struct dirent *ep;
char buf[MAXPATHLEN];
char *tmp1 = NULL;
char *tmp2 = NULL;
char *tmp3 = NULL;
char *path = NULL;
size_t size;
int tmpsize;
if (dev_name == NULL)
return (NULL);
/* If they preface 'dev' with a path (like "/dev") then strip it off */
tmp1 = strrchr(dev_name, '/');
if (tmp1 != NULL)
dev_name = tmp1 + 1; /* +1 since we want the chr after '/' */
tmpsize = asprintf(&tmp1, "/sys/block/%s/device", dev_name);
if (tmpsize == -1 || tmp1 == NULL) {
tmp1 = NULL;
goto end;
}
dp = opendir(tmp1);
if (dp == NULL) {
tmp1 = NULL; /* To make free() at the end a NOP */
goto end;
}
/*
* Look though all sysfs entries in /sys/block/<dev>/device for
* the enclosure symlink.
*/
while ((ep = readdir(dp))) {
/* Ignore everything that's not our enclosure_device link */
if (strstr(ep->d_name, "enclosure_device") == NULL)
continue;
if (asprintf(&tmp2, "%s/%s", tmp1, ep->d_name) == -1 ||
tmp2 == NULL)
break;
size = readlink(tmp2, buf, sizeof (buf));
/* Did readlink fail or crop the link name? */
if (size == -1 || size >= sizeof (buf)) {
free(tmp2);
tmp2 = NULL; /* To make free() at the end a NOP */
break;
}
/*
* We got a valid link. readlink() doesn't terminate strings
* so we have to do it.
*/
buf[size] = '\0';
/*
* Our link will look like:
*
* "../../../../port-11:1:2/..STUFF../enclosure/1:0:3:0/SLOT 1"
*
* We want to grab the "enclosure/1:0:3:0/SLOT 1" part
*/
tmp3 = strstr(buf, "enclosure");
if (tmp3 == NULL)
break;
if (asprintf(&path, "/sys/class/%s", tmp3) == -1) {
/* If asprintf() fails, 'path' is undefined */
path = NULL;
break;
}
if (path == NULL)
break;
}
end:
free(tmp2);
free(tmp1);
if (dp != NULL)
closedir(dp);
return (path);
}
/*
* Remove partition suffix from a vdev path. Partition suffixes may take three
* forms: "-partX", "pX", or "X", where X is a string of digits. The second
* case only occurs when the suffix is preceded by a digit, i.e. "md0p0" The
* third case only occurs when preceded by a string matching the regular
* expression "^([hsv]|xv)d[a-z]+", i.e. a scsi, ide, virtio or xen disk.
*
* caller must free the returned string
*/
char *
zfs_strip_partition(char *path)
{
char *tmp = strdup(path);
char *part = NULL, *d = NULL;
if (!tmp)
return (NULL);
if ((part = strstr(tmp, "-part")) && part != tmp) {
d = part + 5;
} else if ((part = strrchr(tmp, 'p')) &&
part > tmp + 1 && isdigit(*(part-1))) {
d = part + 1;
} else if ((tmp[0] == 'h' || tmp[0] == 's' || tmp[0] == 'v') &&
tmp[1] == 'd') {
for (d = &tmp[2]; isalpha(*d); part = ++d) { }
} else if (strncmp("xvd", tmp, 3) == 0) {
for (d = &tmp[3]; isalpha(*d); part = ++d) { }
}
if (part && d && *d != '\0') {
for (; isdigit(*d); d++) { }
if (*d == '\0')
*part = '\0';
}
return (tmp);
}
/*
* Same as zfs_strip_partition, but allows "/dev/" to be in the pathname
*
* path: /dev/sda1
* returns: /dev/sda
*
* Returned string must be freed.
*/
char *
zfs_strip_partition_path(char *path)
{
char *newpath = strdup(path);
char *sd_offset;
char *new_sd;
if (!newpath)
return (NULL);
/* Point to "sda1" part of "/dev/sda1" */
sd_offset = strrchr(newpath, '/') + 1;
/* Get our new name "sda" */
new_sd = zfs_strip_partition(sd_offset);
if (!new_sd) {
free(newpath);
return (NULL);
}
/* Paste the "sda" where "sda1" was */
strlcpy(sd_offset, new_sd, strlen(sd_offset) + 1);
/* Free temporary "sda" */
free(new_sd);
return (newpath);
}
#ifdef HAVE_LIBUDEV
/*
* A disk is considered a multipath whole disk when:
* DEVNAME key value has "dm-"
* DM_NAME key value has "mpath" prefix
* DM_UUID key exists
* ID_PART_TABLE_TYPE key does not exist or is not gpt
*/
static boolean_t
udev_mpath_whole_disk(struct udev_device *dev)
{
const char *devname, *type, *uuid;
devname = udev_device_get_property_value(dev, "DEVNAME");
type = udev_device_get_property_value(dev, "ID_PART_TABLE_TYPE");
uuid = udev_device_get_property_value(dev, "DM_UUID");
if ((devname != NULL && strncmp(devname, "/dev/dm-", 8) == 0) &&
((type == NULL) || (strcmp(type, "gpt") != 0)) &&
(uuid != NULL)) {
return (B_TRUE);
}
return (B_FALSE);
}
/*
* Check if a disk is effectively a multipath whole disk
*/
boolean_t
is_mpath_whole_disk(const char *path)
{
struct udev *udev;
struct udev_device *dev = NULL;
char nodepath[MAXPATHLEN];
char *sysname;
boolean_t wholedisk = B_FALSE;
if (realpath(path, nodepath) == NULL)
return (B_FALSE);
sysname = strrchr(nodepath, '/') + 1;
if (strncmp(sysname, "dm-", 3) != 0)
return (B_FALSE);
if ((udev = udev_new()) == NULL)
return (B_FALSE);
if ((dev = udev_device_new_from_subsystem_sysname(udev, "block",
sysname)) == NULL) {
udev_device_unref(dev);
return (B_FALSE);
}
wholedisk = udev_mpath_whole_disk(dev);
udev_device_unref(dev);
return (wholedisk);
}
#endif

2389
lib/libzutil/zutil_import.c Normal file
View File

File diff suppressed because it is too large Load Diff

157
lib/libzutil/zutil_nicenum.c Normal file
View File

@ -0,0 +1,157 @@
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
*/
#include <math.h>
#include <stdio.h>
#include <libzutil.h>
/*
* Convert a number to an appropriately human-readable output.
*/
void
zfs_nicenum_format(uint64_t num, char *buf, size_t buflen,
enum zfs_nicenum_format format)
{
uint64_t n = num;
int index = 0;
const char *u;
const char *units[3][7] = {
[ZFS_NICENUM_1024] = {"", "K", "M", "G", "T", "P", "E"},
[ZFS_NICENUM_BYTES] = {"B", "K", "M", "G", "T", "P", "E"},
[ZFS_NICENUM_TIME] = {"ns", "us", "ms", "s", "?", "?", "?"}
};
const int units_len[] = {[ZFS_NICENUM_1024] = 6,
[ZFS_NICENUM_BYTES] = 6,
[ZFS_NICENUM_TIME] = 4};
const int k_unit[] = { [ZFS_NICENUM_1024] = 1024,
[ZFS_NICENUM_BYTES] = 1024,
[ZFS_NICENUM_TIME] = 1000};
double val;
if (format == ZFS_NICENUM_RAW) {
snprintf(buf, buflen, "%llu", (u_longlong_t)num);
return;
} else if (format == ZFS_NICENUM_RAWTIME && num > 0) {
snprintf(buf, buflen, "%llu", (u_longlong_t)num);
return;
} else if (format == ZFS_NICENUM_RAWTIME && num == 0) {
snprintf(buf, buflen, "%s", "-");
return;
}
while (n >= k_unit[format] && index < units_len[format]) {
n /= k_unit[format];
index++;
}
u = units[format][index];
/* Don't print zero latencies since they're invalid */
if ((format == ZFS_NICENUM_TIME) && (num == 0)) {
(void) snprintf(buf, buflen, "-");
} else if ((index == 0) || ((num %
(uint64_t)powl(k_unit[format], index)) == 0)) {
/*
* If this is an even multiple of the base, always display
* without any decimal precision.
*/
(void) snprintf(buf, buflen, "%llu%s", (u_longlong_t)n, u);
} else {
/*
* We want to choose a precision that reflects the best choice
* for fitting in 5 characters. This can get rather tricky when
* we have numbers that are very close to an order of magnitude.
* For example, when displaying 10239 (which is really 9.999K),
* we want only a single place of precision for 10.0K. We could
* develop some complex heuristics for this, but it's much
* easier just to try each combination in turn.
*/
int i;
for (i = 2; i >= 0; i--) {
val = (double)num /
(uint64_t)powl(k_unit[format], index);
/*
* Don't print floating point values for time. Note,
* we use floor() instead of round() here, since
* round can result in undesirable results. For
* example, if "num" is in the range of
* 999500-999999, it will print out "1000us". This
* doesn't happen if we use floor().
*/
if (format == ZFS_NICENUM_TIME) {
if (snprintf(buf, buflen, "%d%s",
(unsigned int) floor(val), u) <= 5)
break;
} else {
if (snprintf(buf, buflen, "%.*f%s", i,
val, u) <= 5)
break;
}
}
}
}
/*
* Convert a number to an appropriately human-readable output.
*/
void
zfs_nicenum(uint64_t num, char *buf, size_t buflen)
{
zfs_nicenum_format(num, buf, buflen, ZFS_NICENUM_1024);
}
/*
* Convert a time to an appropriately human-readable output.
* @num: Time in nanoseconds
*/
void
zfs_nicetime(uint64_t num, char *buf, size_t buflen)
{
zfs_nicenum_format(num, buf, buflen, ZFS_NICENUM_TIME);
}
/*
* Print out a raw number with correct column spacing
*/
void
zfs_niceraw(uint64_t num, char *buf, size_t buflen)
{
zfs_nicenum_format(num, buf, buflen, ZFS_NICENUM_RAW);
}
/*
* Convert a number of bytes to an appropriately human-readable output.
*/
void
zfs_nicebytes(uint64_t num, char *buf, size_t buflen)
{
zfs_nicenum_format(num, buf, buflen, ZFS_NICENUM_BYTES);
}

145
lib/libzutil/zutil_pool.c Normal file
View File

@ -0,0 +1,145 @@
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
*/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/nvpair.h>
#include <sys/fs/zfs.h>
#include <libzutil.h>
static void
dump_ddt_stat(const ddt_stat_t *dds, int h)
{
char refcnt[6];
char blocks[6], lsize[6], psize[6], dsize[6];
char ref_blocks[6], ref_lsize[6], ref_psize[6], ref_dsize[6];
if (dds == NULL || dds->dds_blocks == 0)
return;
if (h == -1)
(void) strcpy(refcnt, "Total");
else
zfs_nicenum(1ULL << h, refcnt, sizeof (refcnt));
zfs_nicenum(dds->dds_blocks, blocks, sizeof (blocks));
zfs_nicebytes(dds->dds_lsize, lsize, sizeof (lsize));
zfs_nicebytes(dds->dds_psize, psize, sizeof (psize));
zfs_nicebytes(dds->dds_dsize, dsize, sizeof (dsize));
zfs_nicenum(dds->dds_ref_blocks, ref_blocks, sizeof (ref_blocks));
zfs_nicebytes(dds->dds_ref_lsize, ref_lsize, sizeof (ref_lsize));
zfs_nicebytes(dds->dds_ref_psize, ref_psize, sizeof (ref_psize));
zfs_nicebytes(dds->dds_ref_dsize, ref_dsize, sizeof (ref_dsize));
(void) printf("%6s %6s %5s %5s %5s %6s %5s %5s %5s\n",
refcnt,
blocks, lsize, psize, dsize,
ref_blocks, ref_lsize, ref_psize, ref_dsize);
}
/*
* Print the DDT histogram and the column totals.
*/
void
zpool_dump_ddt(const ddt_stat_t *dds_total, const ddt_histogram_t *ddh)
{
int h;
(void) printf("\n");
(void) printf("bucket "
" allocated "
" referenced \n");
(void) printf("______ "
"______________________________ "
"______________________________\n");
(void) printf("%6s %6s %5s %5s %5s %6s %5s %5s %5s\n",
"refcnt",
"blocks", "LSIZE", "PSIZE", "DSIZE",
"blocks", "LSIZE", "PSIZE", "DSIZE");
(void) printf("%6s %6s %5s %5s %5s %6s %5s %5s %5s\n",
"------",
"------", "-----", "-----", "-----",
"------", "-----", "-----", "-----");
for (h = 0; h < 64; h++)
dump_ddt_stat(&ddh->ddh_stat[h], h);
dump_ddt_stat(dds_total, -1);
(void) printf("\n");
}
/*
* Process the buffer of nvlists, unpacking and storing each nvlist record
* into 'records'. 'leftover' is set to the number of bytes that weren't
* processed as there wasn't a complete record.
*/
int
zpool_history_unpack(char *buf, uint64_t bytes_read, uint64_t *leftover,
nvlist_t ***records, uint_t *numrecords)
{
uint64_t reclen;
nvlist_t *nv;
int i;
void *tmp;
while (bytes_read > sizeof (reclen)) {
/* get length of packed record (stored as little endian) */
for (i = 0, reclen = 0; i < sizeof (reclen); i++)
reclen += (uint64_t)(((uchar_t *)buf)[i]) << (8*i);
if (bytes_read < sizeof (reclen) + reclen)
break;
/* unpack record */
if (nvlist_unpack(buf + sizeof (reclen), reclen, &nv, 0) != 0)
return (ENOMEM);
bytes_read -= sizeof (reclen) + reclen;
buf += sizeof (reclen) + reclen;
/* add record to nvlist array */
(*numrecords)++;
if (ISP2(*numrecords + 1)) {
tmp = realloc(*records,
*numrecords * 2 * sizeof (nvlist_t *));
if (tmp == NULL) {
nvlist_free(nv);
(*numrecords)--;
return (ENOMEM);
}
*records = tmp;
}
(*records)[*numrecords - 1] = nv;
}
*leftover = bytes_read;
return (0);
}

67
module/zfs/zio_inject.c
View File

@ -46,6 +46,7 @@
#include <sys/zfs_ioctl.h>
#include <sys/vdev_impl.h>
#include <sys/dmu_objset.h>
#include <sys/dsl_dataset.h>
#include <sys/fs/zfs.h>
uint32_t zio_injection_enabled = 0;
@ -659,6 +660,63 @@ zio_handle_io_delay(zio_t *zio)
return (min_target);
}
static int
zio_calculate_range(const char *pool, zinject_record_t *record)
{
dsl_pool_t *dp;
dsl_dataset_t *ds;
objset_t *os = NULL;
dnode_t *dn = NULL;
int error;
/*
* Obtain the dnode for object using pool, objset, and object
*/
error = dsl_pool_hold(pool, FTAG, &dp);
if (error)
return (error);
error = dsl_dataset_hold_obj(dp, record->zi_objset, FTAG, &ds);
dsl_pool_rele(dp, FTAG);
if (error)
return (error);
error = dmu_objset_from_ds(ds, &os);
dsl_dataset_rele(ds, FTAG);
if (error)
return (error);
error = dnode_hold(os, record->zi_object, FTAG, &dn);
if (error)
return (error);
/*
* Translate the range into block IDs
*/
if (record->zi_start != 0 || record->zi_end != -1ULL) {
record->zi_start >>= dn->dn_datablkshift;
record->zi_end >>= dn->dn_datablkshift;
}
if (record->zi_level > 0) {
if (record->zi_level >= dn->dn_nlevels) {
dnode_rele(dn, FTAG);
return (SET_ERROR(EDOM));
}
if (record->zi_start != 0 || record->zi_end != 0) {
int shift = dn->dn_indblkshift - SPA_BLKPTRSHIFT;
for (int level = record->zi_level; level > 0; level--) {
record->zi_start >>= shift;
record->zi_end >>= shift;
}
}
}
dnode_rele(dn, FTAG);
return (0);
}
/*
* Create a new handler for the given record. We add it to the list, adding
* a reference to the spa_t in the process. We increment zio_injection_enabled,
@ -698,6 +756,15 @@ zio_inject_fault(char *name, int flags, int *id, zinject_record_t *record)
return (SET_ERROR(EINVAL));
}
/*
* If the supplied range was in bytes -- calculate the actual blkid
*/
if (flags & ZINJECT_CALC_RANGE) {
error = zio_calculate_range(name, record);
if (error != 0)
return (error);
}
if (!(flags & ZINJECT_NULL)) {
/*
* spa_inject_ref() will add an injection reference, which will