* 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.
+ * or https://opensource.org/licenses/CDDL-1.0.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* under sponsorship from the FreeBSD Foundation.
* Copyright (c) 2021 Allan Jude
* Copyright (c) 2021 Toomas Soome <tsoome@me.com>
+ * Copyright (c) 2023, Klara Inc.
+ * Copyright (c) 2023, Rob Norris <robn@despairlabs.com>
*/
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <ctype.h>
+#include <getopt.h>
+#include <openssl/evp.h>
#include <sys/zfs_context.h>
#include <sys/spa.h>
#include <sys/spa_impl.h>
#include <sys/dsl_crypt.h>
#include <sys/dsl_scan.h>
#include <sys/btree.h>
+#include <sys/brt.h>
+#include <sys/brt_impl.h>
#include <zfs_comutil.h>
#include <sys/zstd/zstd.h>
#include <libnvpair.h>
#include <libzutil.h>
-#include "zdb.h"
+#include <libzdb.h>
-#define ZDB_COMPRESS_NAME(idx) ((idx) < ZIO_COMPRESS_FUNCTIONS ? \
- zio_compress_table[(idx)].ci_name : "UNKNOWN")
-#define ZDB_CHECKSUM_NAME(idx) ((idx) < ZIO_CHECKSUM_FUNCTIONS ? \
- zio_checksum_table[(idx)].ci_name : "UNKNOWN")
-#define ZDB_OT_TYPE(idx) ((idx) < DMU_OT_NUMTYPES ? (idx) : \
- (idx) == DMU_OTN_ZAP_DATA || (idx) == DMU_OTN_ZAP_METADATA ? \
- DMU_OT_ZAP_OTHER : \
- (idx) == DMU_OTN_UINT64_DATA || (idx) == DMU_OTN_UINT64_METADATA ? \
- DMU_OT_UINT64_OTHER : DMU_OT_NUMTYPES)
-
-/* Some platforms require part of inode IDs to be remapped */
-#ifdef __APPLE__
-#define ZDB_MAP_OBJECT_ID(obj) INO_XNUTOZFS(obj, 2)
-#else
-#define ZDB_MAP_OBJECT_ID(obj) (obj)
-#endif
+#include "zdb.h"
-static char *
-zdb_ot_name(dmu_object_type_t type)
-{
- if (type < DMU_OT_NUMTYPES)
- return (dmu_ot[type].ot_name);
- else if ((type & DMU_OT_NEWTYPE) &&
- ((type & DMU_OT_BYTESWAP_MASK) < DMU_BSWAP_NUMFUNCS))
- return (dmu_ot_byteswap[type & DMU_OT_BYTESWAP_MASK].ob_name);
- else
- return ("UNKNOWN");
-}
extern int reference_tracking_enable;
extern int zfs_recover;
-extern unsigned long zfs_arc_meta_min, zfs_arc_meta_limit;
-extern int zfs_vdev_async_read_max_active;
+extern uint_t zfs_vdev_async_read_max_active;
extern boolean_t spa_load_verify_dryrun;
-extern int zfs_reconstruct_indirect_combinations_max;
-extern int zfs_btree_verify_intensity;
+extern boolean_t spa_mode_readable_spacemaps;
+extern uint_t zfs_reconstruct_indirect_combinations_max;
+extern uint_t zfs_btree_verify_intensity;
static const char cmdname[] = "zdb";
uint8_t dump_opt[256];
typedef void object_viewer_t(objset_t *, uint64_t, void *data, size_t size);
-uint64_t *zopt_metaslab = NULL;
+static uint64_t *zopt_metaslab = NULL;
static unsigned zopt_metaslab_args = 0;
-typedef struct zopt_object_range {
- uint64_t zor_obj_start;
- uint64_t zor_obj_end;
- uint64_t zor_flags;
-} zopt_object_range_t;
-zopt_object_range_t *zopt_object_ranges = NULL;
+
+static zopt_object_range_t *zopt_object_ranges = NULL;
static unsigned zopt_object_args = 0;
static int flagbits[256];
-#define ZOR_FLAG_PLAIN_FILE 0x0001
-#define ZOR_FLAG_DIRECTORY 0x0002
-#define ZOR_FLAG_SPACE_MAP 0x0004
-#define ZOR_FLAG_ZAP 0x0008
-#define ZOR_FLAG_ALL_TYPES -1
-#define ZOR_SUPPORTED_FLAGS (ZOR_FLAG_PLAIN_FILE | \
- ZOR_FLAG_DIRECTORY | \
- ZOR_FLAG_SPACE_MAP | \
- ZOR_FLAG_ZAP)
-#define ZDB_FLAG_CHECKSUM 0x0001
-#define ZDB_FLAG_DECOMPRESS 0x0002
-#define ZDB_FLAG_BSWAP 0x0004
-#define ZDB_FLAG_GBH 0x0008
-#define ZDB_FLAG_INDIRECT 0x0010
-#define ZDB_FLAG_RAW 0x0020
-#define ZDB_FLAG_PRINT_BLKPTR 0x0040
-#define ZDB_FLAG_VERBOSE 0x0080
-
-uint64_t max_inflight_bytes = 256 * 1024 * 1024; /* 256MB */
+static uint64_t max_inflight_bytes = 256 * 1024 * 1024; /* 256MB */
static int leaked_objects = 0;
static range_tree_t *mos_refd_objs;
static int dump_bpobj_cb(void *arg, const blkptr_t *bp, boolean_t free,
dmu_tx_t *tx);
-typedef struct sublivelist_verify {
- /* FREE's that haven't yet matched to an ALLOC, in one sub-livelist */
- zfs_btree_t sv_pair;
-
- /* ALLOC's without a matching FREE, accumulates across sub-livelists */
- zfs_btree_t sv_leftover;
-} sublivelist_verify_t;
-
-static int
-livelist_compare(const void *larg, const void *rarg)
-{
- const blkptr_t *l = larg;
- const blkptr_t *r = rarg;
-
- /* Sort them according to dva[0] */
- uint64_t l_dva0_vdev, r_dva0_vdev;
- l_dva0_vdev = DVA_GET_VDEV(&l->blk_dva[0]);
- r_dva0_vdev = DVA_GET_VDEV(&r->blk_dva[0]);
- if (l_dva0_vdev < r_dva0_vdev)
- return (-1);
- else if (l_dva0_vdev > r_dva0_vdev)
- return (+1);
-
- /* if vdevs are equal, sort by offsets. */
- uint64_t l_dva0_offset;
- uint64_t r_dva0_offset;
- l_dva0_offset = DVA_GET_OFFSET(&l->blk_dva[0]);
- r_dva0_offset = DVA_GET_OFFSET(&r->blk_dva[0]);
- if (l_dva0_offset < r_dva0_offset) {
- return (-1);
- } else if (l_dva0_offset > r_dva0_offset) {
- return (+1);
- }
- /*
- * Since we're storing blkptrs without cancelling FREE/ALLOC pairs,
- * it's possible the offsets are equal. In that case, sort by txg
- */
- if (l->blk_birth < r->blk_birth) {
- return (-1);
- } else if (l->blk_birth > r->blk_birth) {
- return (+1);
- }
- return (0);
-}
-
-typedef struct sublivelist_verify_block {
- dva_t svb_dva;
-
- /*
- * We need this to check if the block marked as allocated
- * in the livelist was freed (and potentially reallocated)
- * in the metaslab spacemaps at a later TXG.
- */
- uint64_t svb_allocated_txg;
-} sublivelist_verify_block_t;
static void zdb_print_blkptr(const blkptr_t *bp, int flags);
int err;
struct sublivelist_verify *sv = args;
- zfs_btree_create(&sv->sv_pair, sublivelist_block_refcnt_compare,
+ zfs_btree_create(&sv->sv_pair, sublivelist_block_refcnt_compare, NULL,
sizeof (sublivelist_verify_block_refcnt_t));
err = bpobj_iterate_nofree(&dle->dle_bpobj, sublivelist_verify_blkptr,
* Check for errors in the livelist entry and discard the intermediary
* data structures
*/
-/* ARGSUSED */
static int
sublivelist_verify_lightweight(void *args, dsl_deadlist_entry_t *dle)
{
+ (void) args;
sublivelist_verify_t sv;
- zfs_btree_create(&sv.sv_leftover, livelist_block_compare,
+ zfs_btree_create(&sv.sv_leftover, livelist_block_compare, NULL,
sizeof (sublivelist_verify_block_t));
int err = sublivelist_verify_func(&sv, dle);
zfs_btree_clear(&sv.sv_leftover);
verify_livelist_allocs(metaslab_verify_t *mv, uint64_t txg,
uint64_t offset, uint64_t size)
{
- sublivelist_verify_block_t svb;
+ sublivelist_verify_block_t svb = {{{0}}};
DVA_SET_VDEV(&svb.svb_dva, mv->mv_vdid);
DVA_SET_OFFSET(&svb.svb_dva, offset);
DVA_SET_ASIZE(&svb.svb_dva, size);
(void) printf("Verifying deleted livelist entries\n");
sublivelist_verify_t sv;
- zfs_btree_create(&sv.sv_leftover, livelist_block_compare,
+ zfs_btree_create(&sv.sv_leftover, livelist_block_compare, NULL,
sizeof (sublivelist_verify_block_t));
iterate_deleted_livelists(spa, livelist_verify, &sv);
mv.mv_start = m->ms_start;
mv.mv_end = m->ms_start + m->ms_size;
zfs_btree_create(&mv.mv_livelist_allocs,
- livelist_block_compare,
+ livelist_block_compare, NULL,
sizeof (sublivelist_verify_block_t));
mv_populate_livelist_allocs(&mv, &sv);
"Usage:\t%s [-AbcdDFGhikLMPsvXy] [-e [-V] [-p <path> ...]] "
"[-I <inflight I/Os>]\n"
"\t\t[-o <var>=<value>]... [-t <txg>] [-U <cache>] [-x <dumpdir>]\n"
+ "\t\t[-K <key>]\n"
"\t\t[<poolname>[/<dataset | objset id>] [<object | range> ...]]\n"
- "\t%s [-AdiPv] [-e [-V] [-p <path> ...]] [-U <cache>]\n"
+ "\t%s [-AdiPv] [-e [-V] [-p <path> ...]] [-U <cache>] [-K <key>]\n"
"\t\t[<poolname>[/<dataset | objset id>] [<object | range> ...]\n"
+ "\t%s -B [-e [-V] [-p <path> ...]] [-I <inflight I/Os>]\n"
+ "\t\t[-o <var>=<value>]... [-t <txg>] [-U <cache>] [-x <dumpdir>]\n"
+ "\t\t[-K <key>] <poolname>/<objset id> [<backupflags>]\n"
"\t%s [-v] <bookmark>\n"
- "\t%s -C [-A] [-U <cache>]\n"
+ "\t%s -C [-A] [-U <cache>] [<poolname>]\n"
"\t%s -l [-Aqu] <device>\n"
"\t%s -m [-AFLPX] [-e [-V] [-p <path> ...]] [-t <txg>] "
"[-U <cache>]\n\t\t<poolname> [<vdev> [<metaslab> ...]]\n"
- "\t%s -O <dataset> <path>\n"
- "\t%s -r <dataset> <path> <destination>\n"
+ "\t%s -O
[-K <key>] <dataset> <path>\n"
+ "\t%s -r
[-K <key>] <dataset> <path> <destination>\n"
"\t%s -R [-A] [-e [-V] [-p <path> ...]] [-U <cache>]\n"
"\t\t<poolname> <vdev>:<offset>:<size>[:<flags>]\n"
"\t%s -E [-A] word0:word1:...:word15\n"
"\t%s -S [-AP] [-e [-V] [-p <path> ...]] [-U <cache>] "
"<poolname>\n\n",
cmdname, cmdname, cmdname, cmdname, cmdname, cmdname, cmdname,
- cmdname, cmdname, cmdname, cmdname);
+ cmdname, cmdname, cmdname, cmdname, cmdname);
(void) fprintf(stderr, " Dataset name must include at least one "
"separator character '/' or '@'\n");
" z ZAPs\n"
" - Negate effect of next flag\n\n");
(void) fprintf(stderr, " Options to control amount of output:\n");
- (void) fprintf(stderr, " -b block statistics\n");
- (void) fprintf(stderr, " -c checksum all metadata (twice for "
- "all data) blocks\n");
- (void) fprintf(stderr, " -C config (or cachefile if alone)\n");
- (void) fprintf(stderr, " -d dataset(s)\n");
- (void) fprintf(stderr, " -D dedup statistics\n");
- (void) fprintf(stderr, " -E decode and display block from an "
- "embedded block pointer\n");
- (void) fprintf(stderr, " -h pool history\n");
- (void) fprintf(stderr, " -i intent logs\n");
- (void) fprintf(stderr, " -l read label contents\n");
- (void) fprintf(stderr, " -k examine the checkpointed state "
- "of the pool\n");
- (void) fprintf(stderr, " -L disable leak tracking (do not "
- "load spacemaps)\n");
- (void) fprintf(stderr, " -m metaslabs\n");
- (void) fprintf(stderr, " -M metaslab groups\n");
- (void) fprintf(stderr, " -O perform object lookups by path\n");
- (void) fprintf(stderr, " -r copy an object by path to file\n");
- (void) fprintf(stderr, " -R read and display block from a "
- "device\n");
- (void) fprintf(stderr, " -s report stats on zdb's I/O\n");
- (void) fprintf(stderr, " -S simulate dedup to measure effect\n");
- (void) fprintf(stderr, " -v verbose (applies to all "
- "others)\n");
- (void) fprintf(stderr, " -y perform livelist and metaslab "
- "validation on any livelists being deleted\n\n");
+ (void) fprintf(stderr, " -b --block-stats "
+ "block statistics\n");
+ (void) fprintf(stderr, " -B --backup "
+ "backup stream\n");
+ (void) fprintf(stderr, " -c --checksum "
+ "checksum all metadata (twice for all data) blocks\n");
+ (void) fprintf(stderr, " -C --config "
+ "config (or cachefile if alone)\n");
+ (void) fprintf(stderr, " -d --datasets "
+ "dataset(s)\n");
+ (void) fprintf(stderr, " -D --dedup-stats "
+ "dedup statistics\n");
+ (void) fprintf(stderr, " -E --embedded-block-pointer=INTEGER\n"
+ " decode and display block "
+ "from an embedded block pointer\n");
+ (void) fprintf(stderr, " -h --history "
+ "pool history\n");
+ (void) fprintf(stderr, " -i --intent-logs "
+ "intent logs\n");
+ (void) fprintf(stderr, " -l --label "
+ "read label contents\n");
+ (void) fprintf(stderr, " -k --checkpointed-state "
+ "examine the checkpointed state of the pool\n");
+ (void) fprintf(stderr, " -L --disable-leak-tracking "
+ "disable leak tracking (do not load spacemaps)\n");
+ (void) fprintf(stderr, " -m --metaslabs "
+ "metaslabs\n");
+ (void) fprintf(stderr, " -M --metaslab-groups "
+ "metaslab groups\n");
+ (void) fprintf(stderr, " -O --object-lookups "
+ "perform object lookups by path\n");
+ (void) fprintf(stderr, " -r --copy-object "
+ "copy an object by path to file\n");
+ (void) fprintf(stderr, " -R --read-block "
+ "read and display block from a device\n");
+ (void) fprintf(stderr, " -s --io-stats "
+ "report stats on zdb's I/O\n");
+ (void) fprintf(stderr, " -S --simulate-dedup "
+ "simulate dedup to measure effect\n");
+ (void) fprintf(stderr, " -v --verbose "
+ "verbose (applies to all others)\n");
+ (void) fprintf(stderr, " -y --livelist "
+ "perform livelist and metaslab validation on any livelists being "
+ "deleted\n\n");
(void) fprintf(stderr, " Below options are intended for use "
"with other options:\n");
- (void) fprintf(stderr, " -A ignore assertions (-A), enable "
- "panic recovery (-AA) or both (-AAA)\n");
- (void) fprintf(stderr, " -e pool is exported/destroyed/"
- "has altroot/not in a cachefile\n");
- (void) fprintf(stderr, " -F attempt automatic rewind within "
- "safe range of transaction groups\n");
- (void) fprintf(stderr, " -G dump zfs_dbgmsg buffer before "
- "exiting\n");
- (void) fprintf(stderr, " -I <number of inflight I/Os> -- "
- "specify the maximum number of\n "
- "checksumming I/Os [default is 200]\n");
- (void) fprintf(stderr, " -o <variable>=<value> set global "
- "variable to an unsigned 32-bit integer\n");
- (void) fprintf(stderr, " -p <path> -- use one or more with "
- "-e to specify path to vdev dir\n");
- (void) fprintf(stderr, " -P print numbers in parseable form\n");
- (void) fprintf(stderr, " -q don't print label contents\n");
- (void) fprintf(stderr, " -t <txg> -- highest txg to use when "
- "searching for uberblocks\n");
- (void) fprintf(stderr, " -u uberblock\n");
- (void) fprintf(stderr, " -U <cachefile_path> -- use alternate "
- "cachefile\n");
- (void) fprintf(stderr, " -V do verbatim import\n");
- (void) fprintf(stderr, " -x <dumpdir> -- "
+ (void) fprintf(stderr, " -A --ignore-assertions "
+ "ignore assertions (-A), enable panic recovery (-AA) or both "
+ "(-AAA)\n");
+ (void) fprintf(stderr, " -e --exported "
+ "pool is exported/destroyed/has altroot/not in a cachefile\n");
+ (void) fprintf(stderr, " -F --automatic-rewind "
+ "attempt automatic rewind within safe range of transaction "
+ "groups\n");
+ (void) fprintf(stderr, " -G --dump-debug-msg "
+ "dump zfs_dbgmsg buffer before exiting\n");
+ (void) fprintf(stderr, " -I --inflight=INTEGER "
+ "specify the maximum number of checksumming I/Os "
+ "[default is 200]\n");
+ (void) fprintf(stderr, " -K --key=KEY "
+ "decryption key for encrypted dataset\n");
+ (void) fprintf(stderr, " -o --option=\"OPTION=INTEGER\" "
+ "set global variable to an unsigned 32-bit integer\n");
+ (void) fprintf(stderr, " -p --path==PATH "
+ "use one or more with -e to specify path to vdev dir\n");
+ (void) fprintf(stderr, " -P --parseable "
+ "print numbers in parseable form\n");
+ (void) fprintf(stderr, " -q --skip-label "
+ "don't print label contents\n");
+ (void) fprintf(stderr, " -t --txg=INTEGER "
+ "highest txg to use when searching for uberblocks\n");
+ (void) fprintf(stderr, " -T --brt-stats "
+ "BRT statistics\n");
+ (void) fprintf(stderr, " -u --uberblock "
+ "uberblock\n");
+ (void) fprintf(stderr, " -U --cachefile=PATH "
+ "use alternate cachefile\n");
+ (void) fprintf(stderr, " -V --verbatim "
+ "do verbatim import\n");
+ (void) fprintf(stderr, " -x --dump-blocks=PATH "
"dump all read blocks into specified directory\n");
- (void) fprintf(stderr, " -X attempt extreme rewind (does not "
- "work with dataset)\n");
- (void) fprintf(stderr, " -Y attempt all reconstruction "
- "combinations for split blocks\n");
- (void) fprintf(stderr, " -Z show ZSTD headers \n");
+ (void) fprintf(stderr, " -X --extreme-rewind "
+ "attempt extreme rewind (does not work with dataset)\n");
+ (void) fprintf(stderr, " -Y --all-reconstruction "
+ "attempt all reconstruction combinations for split blocks\n");
+ (void) fprintf(stderr, " -Z --zstd-headers "
+ "show ZSTD headers \n");
(void) fprintf(stderr, "Specify an option more than once (e.g. -bb) "
"to make only that option verbose\n");
(void) fprintf(stderr, "Default is to dump everything non-verbosely\n");
exit(1);
}
-/* ARGSUSED */
static void
dump_packed_nvlist(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) size;
nvlist_t *nv;
size_t nvsize = *(uint64_t *)data;
char *packed = umem_alloc(nvsize, UMEM_NOFAIL);
nvlist_free(nv);
}
-/* ARGSUSED */
static void
dump_history_offsets(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) os, (void) object, (void) size;
spa_history_phys_t *shp = data;
if (shp == NULL)
if (dump_opt['P'])
(void) snprintf(buf, buflen, "%llu", (longlong_t)num);
else
- nicenum(num, buf, sizeof (buf));
+ nicenum(num, buf, buflen);
+}
+
+static void
+zdb_nicebytes(uint64_t bytes, char *buf, size_t buflen)
+{
+ if (dump_opt['P'])
+ (void) snprintf(buf, buflen, "%llu", (longlong_t)bytes);
+ else
+ zfs_nicebytes(bytes, buf, buflen);
}
static const char histo_stars[] = "****************************************";
uint64_t max = 0;
for (i = 0; i < size; i++) {
+ if (histo[i] == 0)
+ continue;
if (histo[i] > max)
max = histo[i];
- if (histo[i] > 0 && i > maxidx)
+ if (i > maxidx)
maxidx = i;
- if (histo[i] > 0 && i < minidx)
+ if (i < minidx)
minidx = i;
}
dump_histogram(zs.zs_buckets_with_n_entries, ZAP_HISTOGRAM_SIZE, 0);
}
-/*ARGSUSED*/
static void
dump_none(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) os, (void) object, (void) data, (void) size;
}
-/*ARGSUSED*/
static void
dump_unknown(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) os, (void) object, (void) data, (void) size;
(void) printf("\tUNKNOWN OBJECT TYPE\n");
}
-/*ARGSUSED*/
static void
dump_uint8(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) os, (void) object, (void) data, (void) size;
}
-/*ARGSUSED*/
static void
dump_uint64(objset_t *os, uint64_t object, void *data, size_t size)
{
}
if (size == 0) {
+ if (data == NULL)
+ kmem_free(arr, oursize);
(void) printf("\t\t[]\n");
return;
}
kmem_free(arr, oursize);
}
-/*ARGSUSED*/
static void
dump_zap(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) data, (void) size;
zap_cursor_t zc;
zap_attribute_t attr;
void *prop;
char bytes[32], comp[32], uncomp[32];
/* make sure the output won't get truncated */
- CTASSERT(sizeof (bytes) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (comp) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (uncomp) >= NN_NUMBUF_SZ);
+ _Static_assert(sizeof (bytes) >= NN_NUMBUF_SZ, "bytes truncated");
+ _Static_assert(sizeof (comp) >= NN_NUMBUF_SZ, "comp truncated");
+ _Static_assert(sizeof (uncomp) >= NN_NUMBUF_SZ, "uncomp truncated");
if (bpop == NULL)
return;
}
}
-/* ARGSUSED */
static void
dump_bpobj_subobjs(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) data, (void) size;
dmu_object_info_t doi;
int64_t i;
kmem_free(subobjs, doi.doi_max_offset);
}
-/*ARGSUSED*/
static void
dump_ddt_zap(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) data, (void) size;
dump_zap_stats(os, object);
/* contents are printed elsewhere, properly decoded */
}
-/*ARGSUSED*/
static void
dump_sa_attrs(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) data, (void) size;
zap_cursor_t zc;
zap_attribute_t attr;
zap_cursor_fini(&zc);
}
-/*ARGSUSED*/
static void
dump_sa_layouts(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) data, (void) size;
zap_cursor_t zc;
zap_attribute_t attr;
uint16_t *layout_attrs;
zap_cursor_fini(&zc);
}
-/*ARGSUSED*/
static void
dump_zpldir(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) data, (void) size;
zap_cursor_t zc;
zap_attribute_t attr;
const char *typenames[] = {
int free_pct = range_tree_space(rt) * 100 / msp->ms_size;
/* max sure nicenum has enough space */
- CTASSERT(sizeof (maxbuf) >= NN_NUMBUF_SZ);
+ _Static_assert(sizeof (maxbuf) >= NN_NUMBUF_SZ, "maxbuf truncated");
zdb_nicenum(metaslab_largest_allocatable(msp), maxbuf, sizeof (maxbuf));
}
static void
-dump_metaslab_groups(spa_t *spa)
+dump_metaslab_groups(spa_t *spa, boolean_t show_special)
{
vdev_t *rvd = spa->spa_root_vdev;
metaslab_class_t *mc = spa_normal_class(spa);
+ metaslab_class_t *smc = spa_special_class(spa);
uint64_t fragmentation;
metaslab_class_histogram_verify(mc);
vdev_t *tvd = rvd->vdev_child[c];
metaslab_group_t *mg = tvd->vdev_mg;
- if (mg == NULL || mg->mg_class != mc)
+ if (mg == NULL || (mg->mg_class != mc &&
+ (!show_special || mg->mg_class != smc)))
continue;
metaslab_group_histogram_verify(mg);
static void
dump_all_ddts(spa_t *spa)
{
- ddt_histogram_t ddh_total;
- ddt_stat_t dds_total;
-
- bzero(&ddh_total, sizeof (ddh_total));
- bzero(&dds_total, sizeof (dds_total));
+ ddt_histogram_t ddh_total = {{{0}}};
+ ddt_stat_t dds_total = {0};
for (enum zio_checksum c = 0; c < ZIO_CHECKSUM_FUNCTIONS; c++) {
ddt_t *ddt = spa->spa_ddt[c];
dump_dedup_ratio(&dds_total);
}
+static void
+dump_brt(spa_t *spa)
+{
+ if (!spa_feature_is_enabled(spa, SPA_FEATURE_BLOCK_CLONING)) {
+ printf("BRT: unsupported on this pool\n");
+ return;
+ }
+
+ if (!spa_feature_is_active(spa, SPA_FEATURE_BLOCK_CLONING)) {
+ printf("BRT: empty\n");
+ return;
+ }
+
+ brt_t *brt = spa->spa_brt;
+ VERIFY(brt);
+
+ char count[32], used[32], saved[32];
+ zdb_nicebytes(brt_get_used(spa), used, sizeof (used));
+ zdb_nicebytes(brt_get_saved(spa), saved, sizeof (saved));
+ uint64_t ratio = brt_get_ratio(spa);
+ printf("BRT: used %s; saved %s; ratio %llu.%02llux\n", used, saved,
+ (u_longlong_t)(ratio / 100), (u_longlong_t)(ratio % 100));
+
+ if (dump_opt['T'] < 2)
+ return;
+
+ for (uint64_t vdevid = 0; vdevid < brt->brt_nvdevs; vdevid++) {
+ brt_vdev_t *brtvd = &brt->brt_vdevs[vdevid];
+ if (brtvd == NULL)
+ continue;
+
+ if (!brtvd->bv_initiated) {
+ printf("BRT: vdev %" PRIu64 ": empty\n", vdevid);
+ continue;
+ }
+
+ zdb_nicenum(brtvd->bv_totalcount, count, sizeof (count));
+ zdb_nicebytes(brtvd->bv_usedspace, used, sizeof (used));
+ zdb_nicebytes(brtvd->bv_savedspace, saved, sizeof (saved));
+ printf("BRT: vdev %" PRIu64 ": refcnt %s; used %s; saved %s\n",
+ vdevid, count, used, saved);
+ }
+
+ if (dump_opt['T'] < 3)
+ return;
+
+ char dva[64];
+ printf("\n%-16s %-10s\n", "DVA", "REFCNT");
+
+ for (uint64_t vdevid = 0; vdevid < brt->brt_nvdevs; vdevid++) {
+ brt_vdev_t *brtvd = &brt->brt_vdevs[vdevid];
+ if (brtvd == NULL || !brtvd->bv_initiated)
+ continue;
+
+ zap_cursor_t zc;
+ zap_attribute_t za;
+ for (zap_cursor_init(&zc, brt->brt_mos, brtvd->bv_mos_entries);
+ zap_cursor_retrieve(&zc, &za) == 0;
+ zap_cursor_advance(&zc)) {
+ uint64_t offset = *(uint64_t *)za.za_name;
+ uint64_t refcnt = za.za_first_integer;
+
+ snprintf(dva, sizeof (dva), "%" PRIu64 ":%llx", vdevid,
+ (u_longlong_t)offset);
+ printf("%-16s %-10llu\n", dva, (u_longlong_t)refcnt);
+ }
+ zap_cursor_fini(&zc);
+ }
+}
+
static void
dump_dtl_seg(void *arg, uint64_t start, uint64_t size)
{
free(buf);
}
-/*ARGSUSED*/
static void
dump_dnode(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) os, (void) object, (void) data, (void) size;
}
static uint64_t
snprintf_zstd_header(spa_t *spa, char *blkbuf, size_t buflen,
const blkptr_t *bp)
{
- abd_t *pabd;
+ static abd_t *pabd = NULL;
void *buf;
zio_t *zio;
zfs_zstdhdr_t zstd_hdr;
return;
}
- pabd = abd_alloc_for_io(SPA_MAXBLOCKSIZE, B_FALSE);
+ if (!pabd)
+ pabd = abd_alloc_for_io(SPA_MAXBLOCKSIZE, B_FALSE);
zio = zio_root(spa, NULL, NULL, 0);
/* Decrypt but don't decompress so we can read the compression header */
(void) snprintf(blkbuf + strlen(blkbuf),
buflen - strlen(blkbuf), " %s", "FREE");
(void) snprintf(blkbuf + strlen(blkbuf),
- buflen - strlen(blkbuf), " cksum=%llx:%llx:%llx:%llx",
+ buflen - strlen(blkbuf),
+ " cksum=%016llx:%016llx:%016llx:%016llx",
(u_longlong_t)bp->blk_cksum.zc_word[0],
(u_longlong_t)bp->blk_cksum.zc_word[1],
(u_longlong_t)bp->blk_cksum.zc_word[2],
return (err);
}
-/*ARGSUSED*/
static void
dump_indirect(dnode_t *dn)
{
dnode_phys_t *dnp = dn->dn_phys;
- int j;
zbookmark_phys_t czb;
(void) printf("Indirect blocks:\n");
SET_BOOKMARK(&czb, dmu_objset_id(dn->dn_objset),
dn->dn_object, dnp->dn_nlevels - 1, 0);
- for (j = 0; j < dnp->dn_nblkptr; j++) {
+ for (int j = 0; j < dnp->dn_nblkptr; j++) {
czb.zb_blkid = j;
(void) visit_indirect(dmu_objset_spa(dn->dn_objset), dnp,
&dnp->dn_blkptr[j], &czb);
(void) printf("\n");
}
-/*ARGSUSED*/
static void
dump_dsl_dir(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) os, (void) object;
dsl_dir_phys_t *dd = data;
time_t crtime;
char nice[32];
/* make sure nicenum has enough space */
- CTASSERT(sizeof (nice) >= NN_NUMBUF_SZ);
+ _Static_assert(sizeof (nice) >= NN_NUMBUF_SZ, "nice truncated");
if (dd == NULL)
return;
(u_longlong_t)dd->dd_clones);
}
-/*ARGSUSED*/
static void
dump_dsl_dataset(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) os, (void) object;
dsl_dataset_phys_t *ds = data;
time_t crtime;
char used[32], compressed[32], uncompressed[32], unique[32];
char blkbuf[BP_SPRINTF_LEN];
/* make sure nicenum has enough space */
- CTASSERT(sizeof (used) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (compressed) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (uncompressed) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (unique) >= NN_NUMBUF_SZ);
+ _Static_assert(sizeof (used) >= NN_NUMBUF_SZ, "used truncated");
+ _Static_assert(sizeof (compressed) >= NN_NUMBUF_SZ,
+ "compressed truncated");
+ _Static_assert(sizeof (uncompressed) >= NN_NUMBUF_SZ,
+ "uncompressed truncated");
+ _Static_assert(sizeof (unique) >= NN_NUMBUF_SZ, "unique truncated");
if (ds == NULL)
return;
(void) printf("\t\tbp = %s\n", blkbuf);
}
-/* ARGSUSED */
static int
dump_bptree_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx)
{
+ (void) arg, (void) tx;
char blkbuf[BP_SPRINTF_LEN];
if (bp->blk_birth != 0) {
dmu_buf_t *db;
/* make sure nicenum has enough space */
- CTASSERT(sizeof (bytes) >= NN_NUMBUF_SZ);
+ _Static_assert(sizeof (bytes) >= NN_NUMBUF_SZ, "bytes truncated");
if (dump_opt['d'] < 3)
return;
(void) bptree_iterate(os, obj, B_FALSE, dump_bptree_cb, NULL, NULL);
}
-/* ARGSUSED */
static int
dump_bpobj_cb(void *arg, const blkptr_t *bp, boolean_t bp_freed, dmu_tx_t *tx)
{
+ (void) arg, (void) tx;
char blkbuf[BP_SPRINTF_LEN];
ASSERT(bp->blk_birth != 0);
uint64_t i;
/* make sure nicenum has enough space */
- CTASSERT(sizeof (bytes) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (comp) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (uncomp) >= NN_NUMBUF_SZ);
+ _Static_assert(sizeof (bytes) >= NN_NUMBUF_SZ, "bytes truncated");
+ _Static_assert(sizeof (comp) >= NN_NUMBUF_SZ, "comp truncated");
+ _Static_assert(sizeof (uncomp) >= NN_NUMBUF_SZ, "uncomp truncated");
if (dump_opt['d'] < 3)
return;
zap_cursor_advance(&zc)) {
char osname[ZFS_MAX_DATASET_NAME_LEN];
char buf[ZFS_MAX_DATASET_NAME_LEN];
+ int len;
dmu_objset_name(os, osname);
- VERIFY3S(0, <=, snprintf(buf, sizeof (buf), "%s#%s", osname,
- attr.za_name));
+ len = snprintf(buf, sizeof (buf), "%s#%s", osname,
+ attr.za_name);
+ VERIFY3S(len, <, ZFS_MAX_DATASET_NAME_LEN);
(void) dump_bookmark(dp, buf, verbosity >= 5, verbosity >= 6);
}
zap_cursor_fini(&zc);
}
static void
-dump_blkptr_list(dsl_deadlist_t *dl, char *name)
+dump_blkptr_list(dsl_deadlist_t *dl, const char *name)
{
char bytes[32];
char comp[32];
}
/* make sure nicenum has enough space */
- CTASSERT(sizeof (bytes) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (comp) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (uncomp) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (entries) >= NN_NUMBUF_SZ);
+ _Static_assert(sizeof (bytes) >= NN_NUMBUF_SZ, "bytes truncated");
+ _Static_assert(sizeof (comp) >= NN_NUMBUF_SZ, "comp truncated");
+ _Static_assert(sizeof (uncomp) >= NN_NUMBUF_SZ, "uncomp truncated");
+ _Static_assert(sizeof (entries) >= NN_NUMBUF_SZ, "entries truncated");
if (dump_opt['d'] < 3)
return;
if (dump_opt['d'] < 4)
return;
- (void) printf("\n");
+ (void) putchar('\n');
dsl_deadlist_iterate(dl, dsl_deadlist_entry_dump, NULL);
}
return (0);
}
+static char *key_material = NULL;
+
+static boolean_t
+zdb_derive_key(dsl_dir_t *dd, uint8_t *key_out)
+{
+ uint64_t keyformat, salt, iters;
+ int i;
+ unsigned char c;
+
+ VERIFY0(zap_lookup(dd->dd_pool->dp_meta_objset, dd->dd_crypto_obj,
+ zfs_prop_to_name(ZFS_PROP_KEYFORMAT), sizeof (uint64_t),
+ 1, &keyformat));
+
+ switch (keyformat) {
+ case ZFS_KEYFORMAT_HEX:
+ for (i = 0; i < WRAPPING_KEY_LEN * 2; i += 2) {
+ if (!isxdigit(key_material[i]) ||
+ !isxdigit(key_material[i+1]))
+ return (B_FALSE);
+ if (sscanf(&key_material[i], "%02hhx", &c) != 1)
+ return (B_FALSE);
+ key_out[i / 2] = c;
+ }
+ break;
+
+ case ZFS_KEYFORMAT_PASSPHRASE:
+ VERIFY0(zap_lookup(dd->dd_pool->dp_meta_objset,
+ dd->dd_crypto_obj, zfs_prop_to_name(ZFS_PROP_PBKDF2_SALT),
+ sizeof (uint64_t), 1, &salt));
+ VERIFY0(zap_lookup(dd->dd_pool->dp_meta_objset,
+ dd->dd_crypto_obj, zfs_prop_to_name(ZFS_PROP_PBKDF2_ITERS),
+ sizeof (uint64_t), 1, &iters));
+
+ if (PKCS5_PBKDF2_HMAC_SHA1(key_material, strlen(key_material),
+ ((uint8_t *)&salt), sizeof (uint64_t), iters,
+ WRAPPING_KEY_LEN, key_out) != 1)
+ return (B_FALSE);
+
+ break;
+
+ default:
+ fatal("no support for key format %u\n",
+ (unsigned int) keyformat);
+ }
+
+ return (B_TRUE);
+}
+
+static char encroot[ZFS_MAX_DATASET_NAME_LEN];
+static boolean_t key_loaded = B_FALSE;
+
+static void
+zdb_load_key(objset_t *os)
+{
+ dsl_pool_t *dp;
+ dsl_dir_t *dd, *rdd;
+ uint8_t key[WRAPPING_KEY_LEN];
+ uint64_t rddobj;
+ int err;
+
+ dp = spa_get_dsl(os->os_spa);
+ dd = os->os_dsl_dataset->ds_dir;
+
+ dsl_pool_config_enter(dp, FTAG);
+ VERIFY0(zap_lookup(dd->dd_pool->dp_meta_objset, dd->dd_crypto_obj,
+ DSL_CRYPTO_KEY_ROOT_DDOBJ, sizeof (uint64_t), 1, &rddobj));
+ VERIFY0(dsl_dir_hold_obj(dd->dd_pool, rddobj, NULL, FTAG, &rdd));
+ dsl_dir_name(rdd, encroot);
+ dsl_dir_rele(rdd, FTAG);
+
+ if (!zdb_derive_key(dd, key))
+ fatal("couldn't derive encryption key");
+
+ dsl_pool_config_exit(dp, FTAG);
+
+ ASSERT3U(dsl_dataset_get_keystatus(dd), ==, ZFS_KEYSTATUS_UNAVAILABLE);
+
+ dsl_crypto_params_t *dcp;
+ nvlist_t *crypto_args;
+
+ crypto_args = fnvlist_alloc();
+ fnvlist_add_uint8_array(crypto_args, "wkeydata",
+ (uint8_t *)key, WRAPPING_KEY_LEN);
+ VERIFY0(dsl_crypto_params_create_nvlist(DCP_CMD_NONE,
+ NULL, crypto_args, &dcp));
+ err = spa_keystore_load_wkey(encroot, dcp, B_FALSE);
+
+ dsl_crypto_params_free(dcp, (err != 0));
+ fnvlist_free(crypto_args);
+
+ if (err != 0)
+ fatal(
+ "couldn't load encryption key for %s: %s",
+ encroot, err == ZFS_ERR_CRYPTO_NOTSUP ?
+ "crypto params not supported" : strerror(err));
+
+ ASSERT3U(dsl_dataset_get_keystatus(dd), ==, ZFS_KEYSTATUS_AVAILABLE);
+
+ printf("Unlocked encryption root: %s\n", encroot);
+ key_loaded = B_TRUE;
+}
+
+static void
+zdb_unload_key(void)
+{
+ if (!key_loaded)
+ return;
+
+ VERIFY0(spa_keystore_unload_wkey(encroot));
+ key_loaded = B_FALSE;
+}
+
static avl_tree_t idx_tree;
static avl_tree_t domain_tree;
static boolean_t fuid_table_loaded;
static sa_attr_type_t *sa_attr_table = NULL;
static int
-open_objset(const char *path, void *tag, objset_t **osp)
+open_objset(const char *path, const void *tag, objset_t **osp)
{
int err;
uint64_t sa_attrs = 0;
uint64_t version = 0;
VERIFY3P(sa_os, ==, NULL);
+
/*
* We can't own an objset if it's redacted. Therefore, we do this
* dance: hold the objset, then acquire a long hold on its dataset, then
* release the pool (which is held as part of holding the objset).
*/
- err = dmu_objset_hold(path, tag, osp);
+
+ if (dump_opt['K']) {
+ /* decryption requested, try to load keys */
+ err = dmu_objset_hold(path, tag, osp);
+ if (err != 0) {
+ (void) fprintf(stderr, "failed to hold dataset "
+ "'%s': %s\n",
+ path, strerror(err));
+ return (err);
+ }
+ dsl_dataset_long_hold(dmu_objset_ds(*osp), tag);
+ dsl_pool_rele(dmu_objset_pool(*osp), tag);
+
+ /* succeeds or dies */
+ zdb_load_key(*osp);
+
+ /* release it all */
+ dsl_dataset_long_rele(dmu_objset_ds(*osp), tag);
+ dsl_dataset_rele(dmu_objset_ds(*osp), tag);
+ }
+
+ int ds_hold_flags = key_loaded ? DS_HOLD_FLAG_DECRYPT : 0;
+
+ err = dmu_objset_hold_flags(path, ds_hold_flags, tag, osp);
if (err != 0) {
(void) fprintf(stderr, "failed to hold dataset '%s': %s\n",
path, strerror(err));
dsl_dataset_long_hold(dmu_objset_ds(*osp), tag);
dsl_pool_rele(dmu_objset_pool(*osp), tag);
- if (dmu_objset_type(*osp) == DMU_OST_ZFS && !(*osp)->os_encrypted) {
+ if (dmu_objset_type(*osp) == DMU_OST_ZFS &&
+ (key_loaded || !(*osp)->os_encrypted)) {
(void) zap_lookup(*osp, MASTER_NODE_OBJ, ZPL_VERSION_STR,
8, 1, &version);
if (version >= ZPL_VERSION_SA) {
(void) fprintf(stderr, "sa_setup failed: %s\n",
strerror(err));
dsl_dataset_long_rele(dmu_objset_ds(*osp), tag);
- dsl_dataset_rele(dmu_objset_ds(*osp), tag);
+ dsl_dataset_rele_flags(dmu_objset_ds(*osp),
+ ds_hold_flags, tag);
*osp = NULL;
}
}
sa_os = *osp;
- return (0);
+ return (err);
}
static void
-close_objset(objset_t *os, void *tag)
+close_objset(objset_t *os, const void *tag)
{
VERIFY3P(os, ==, sa_os);
if (os->os_sa != NULL)
sa_tear_down(os);
dsl_dataset_long_rele(dmu_objset_ds(os), tag);
- dsl_dataset_rele(dmu_objset_ds(os), tag);
+ dsl_dataset_rele_flags(dmu_objset_ds(os),
+ key_loaded ? DS_HOLD_FLAG_DECRYPT : 0, tag);
sa_attr_table = NULL;
sa_os = NULL;
+
+ zdb_unload_key();
}
static void
print_idstr(uint64_t id, const char *id_type)
{
if (FUID_INDEX(id)) {
- char *domain;
-
- domain = zfs_fuid_idx_domain(&idx_tree, FUID_INDEX(id));
+ const char *domain =
+ zfs_fuid_idx_domain(&idx_tree, FUID_INDEX(id));
(void) printf("\t%s %llx [%s-%d]\n", id_type,
(u_longlong_t)id, domain, (int)FUID_RID(id));
} else {
(void) printf("\tSA xattrs: %d bytes, %d entries\n\n",
sa_xattr_size, sa_xattr_entries);
while ((elem = nvlist_next_nvpair(sa_xattr, elem)) != NULL) {
+ boolean_t can_print = !dump_opt['P'];
uchar_t *value;
uint_t cnt, idx;
(void) printf("\t\t%s = ", nvpair_name(elem));
nvpair_value_byte_array(elem, &value, &cnt);
+
+ for (idx = 0; idx < cnt; ++idx) {
+ if (!isprint(value[idx])) {
+ can_print = B_FALSE;
+ break;
+ }
+ }
+
for (idx = 0; idx < cnt; ++idx) {
- if (isprint(value[idx]))
+ if (can_print)
(void) putchar(value[idx]);
else
(void) printf("\\%3.3o", value[idx]);
{
int sa_symlink_size = 0;
char linktarget[MAXPATHLEN];
- linktarget[0] = '\0';
int error;
error = sa_size(hdl, sa_attr_table[ZPL_SYMLINK], &sa_symlink_size);
if (error || sa_symlink_size == 0) {
return;
}
+ if (sa_symlink_size >= sizeof (linktarget)) {
+ (void) printf("symlink size %d is too large\n",
+ sa_symlink_size);
+ return;
+ }
+ linktarget[sa_symlink_size] = '\0';
if (sa_lookup(hdl, sa_attr_table[ZPL_SYMLINK],
&linktarget, sa_symlink_size) == 0)
(void) printf("\ttarget %s\n", linktarget);
}
-/*ARGSUSED*/
static void
dump_znode(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) data, (void) size;
char path[MAXPATHLEN * 2]; /* allow for xattr and failure prefix */
sa_handle_t *hdl;
uint64_t xattr, rdev, gen;
sa_handle_destroy(hdl);
}
-/*ARGSUSED*/
static void
dump_acl(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) os, (void) object, (void) data, (void) size;
}
-/*ARGSUSED*/
static void
dump_dmu_objset(objset_t *os, uint64_t object, void *data, size_t size)
{
+ (void) os, (void) object, (void) data, (void) size;
}
static object_viewer_t *object_viewer[DMU_OT_NUMTYPES + 1] = {
int error;
/* make sure nicenum has enough space */
- CTASSERT(sizeof (iblk) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (dblk) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (lsize) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (asize) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (bonus_size) >= NN_NUMBUF_SZ);
+ _Static_assert(sizeof (iblk) >= NN_NUMBUF_SZ, "iblk truncated");
+ _Static_assert(sizeof (dblk) >= NN_NUMBUF_SZ, "dblk truncated");
+ _Static_assert(sizeof (lsize) >= NN_NUMBUF_SZ, "lsize truncated");
+ _Static_assert(sizeof (asize) >= NN_NUMBUF_SZ, "asize truncated");
+ _Static_assert(sizeof (bonus_size) >= NN_NUMBUF_SZ,
+ "bonus_size truncated");
if (*print_header) {
(void) printf("\n%10s %3s %5s %5s %5s %6s %5s %6s %s\n",
if (error)
fatal("dmu_object_info() failed, errno %u", error);
- if (os->os_encrypted &&
+ if (!key_loaded && os->os_encrypted &&
DMU_OT_IS_ENCRYPTED(doi.doi_bonus_type)) {
error = dnode_hold(os, object, FTAG, &dn);
if (error)
zdb_nicenum(doi.doi_physical_blocks_512 << 9, asize, sizeof (asize));
zdb_nicenum(doi.doi_bonus_size, bonus_size, sizeof (bonus_size));
zdb_nicenum(doi.doi_dnodesize, dnsize, sizeof (dnsize));
- (void) sprintf(fill, "%6.2f", 100.0 * doi.doi_fill_count *
- doi.doi_data_block_size / (object == 0 ? DNODES_PER_BLOCK : 1) /
- doi.doi_max_offset);
+ (void) snprintf(fill, sizeof (fill), "%6.2f", 100.0 *
+ doi.doi_fill_count * doi.doi_data_block_size / (object == 0 ?
+ DNODES_PER_BLOCK : 1) / doi.doi_max_offset);
aux[0] = '\0';
(void) printf("\t\t(bonus encrypted)\n");
}
- if (!os->os_encrypted || !DMU_OT_IS_ENCRYPTED(doi.doi_type)) {
+ if (key_loaded ||
+ (!os->os_encrypted || !DMU_OT_IS_ENCRYPTED(doi.doi_type))) {
object_viewer[ZDB_OT_TYPE(doi.doi_type)](os, object,
NULL, 0);
} else {
*print_header = B_TRUE;
}
- if (verbosity >= 5)
+ if (verbosity >= 5) {
+ if (dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR) {
+ char blkbuf[BP_SPRINTF_LEN];
+ snprintf_blkptr_compact(blkbuf, sizeof (blkbuf),
+ DN_SPILL_BLKPTR(dn->dn_phys), B_FALSE);
+ (void) printf("\nSpill block: %s\n", blkbuf);
+ }
dump_indirect(dn);
+ }
if (verbosity >= 5) {
/*
for (;;) {
char segsize[32];
/* make sure nicenum has enough space */
- CTASSERT(sizeof (segsize) >= NN_NUMBUF_SZ);
+ _Static_assert(sizeof (segsize) >= NN_NUMBUF_SZ,
+ "segsize truncated");
error = dnode_next_offset(dn,
0, &start, minlvl, blkfill, 0);
if (error)
}
}
-static const char *objset_types[DMU_OST_NUMTYPES] = {
+static const char *const objset_types[DMU_OST_NUMTYPES] = {
"NONE", "META", "ZPL", "ZVOL", "OTHER", "ANY" };
/*
* pointer to point to a descriptive error message.
*/
static int
-parse_object_range(char *range, zopt_object_range_t *zor, char **msg)
+parse_object_range(char *range, zopt_object_range_t *zor, const char **msg)
{
uint64_t flags = 0;
char *p, *s, *dup, *flagstr, *tmp = NULL;
uint64_t flags;
/* make sure nicenum has enough space */
- CTASSERT(sizeof (numbuf) >= NN_NUMBUF_SZ);
+ _Static_assert(sizeof (numbuf) >= NN_NUMBUF_SZ, "numbuf truncated");
dsl_pool_config_enter(dmu_objset_pool(os), FTAG);
dmu_objset_fast_stat(os, &dds);
(void) printf("\ttxg = %llu\n", (u_longlong_t)ub->ub_txg);
(void) printf("\tguid_sum = %llu\n", (u_longlong_t)ub->ub_guid_sum);
(void) printf("\ttimestamp = %llu UTC = %s",
- (u_longlong_t)ub->ub_timestamp, asctime(localtime(×tamp)));
+ (u_longlong_t)ub->ub_timestamp, ctime(×tamp));
(void) printf("\tmmp_magic = %016llx\n",
(u_longlong_t)ub->ub_mmp_magic);
}
(void) printf("\tcheckpoint_txg = %llu\n",
(u_longlong_t)ub->ub_checkpoint_txg);
+
+ (void) printf("\traidz_reflow state=%u off=%llu\n",
+ (int)RRSS_GET_STATE(ub),
+ (u_longlong_t)RRSS_GET_OFFSET(ub));
+
(void) printf("%s", footer ? footer : "");
}
{
nvlist_t *list, **array;
nvpair_t *nvp = NULL;
- char *name;
+ const char *name;
uint_t i, items;
stats->zns_list_count++;
}
static void
-print_label_numbers(char *prefix, cksum_record_t *rec)
+print_label_numbers(const char *prefix, const cksum_record_t *rec)
{
- printf("%s", prefix);
+ fputs(prefix, stdout);
for (int i = 0; i < VDEV_LABELS; i++)
if (rec->labels[i] == B_TRUE)
printf("%d ", i);
- printf("\n");
+ putchar('\n');
}
#define MAX_UBERBLOCK_COUNT (VDEV_UBERBLOCK_RING >> UBERBLOCK_SHIFT)
typedef struct zdb_label {
vdev_label_t label;
+ uint64_t label_offset;
nvlist_t *config_nv;
cksum_record_t *config;
cksum_record_t *uberblocks[MAX_UBERBLOCK_COUNT];
boolean_t header_printed;
boolean_t read_failed;
+ boolean_t cksum_valid;
} zdb_label_t;
static void
return;
(void) printf("------------------------------------\n");
- (void) printf("LABEL %d\n", l);
+ (void) printf("LABEL %d %s\n", l,
+ label->cksum_valid ? "" : "(Bad label cksum)");
(void) printf("------------------------------------\n");
label->header_printed = B_TRUE;
}
static void
-dump_l2arc_log_blkptr(l2arc_log_blkptr_t lbps)
+dump_l2arc_log_blkptr(const l2arc_log_blkptr_t *lbps)
{
- (void) printf("|\t\tdaddr: %llu\n", (u_longlong_t)lbps.lbp_daddr);
+ (void) printf("|\t\tdaddr: %llu\n", (u_longlong_t)lbps->lbp_daddr);
(void) printf("|\t\tpayload_asize: %llu\n",
- (u_longlong_t)lbps.lbp_payload_asize);
+ (u_longlong_t)lbps->lbp_payload_asize);
(void) printf("|\t\tpayload_start: %llu\n",
- (u_longlong_t)lbps.lbp_payload_start);
+ (u_longlong_t)lbps->lbp_payload_start);
(void) printf("|\t\tlsize: %llu\n",
- (u_longlong_t)L2BLK_GET_LSIZE((&lbps)->lbp_prop));
+ (u_longlong_t)L2BLK_GET_LSIZE(lbps->lbp_prop));
(void) printf("|\t\tasize: %llu\n",
- (u_longlong_t)L2BLK_GET_PSIZE((&lbps)->lbp_prop));
+ (u_longlong_t)L2BLK_GET_PSIZE(lbps->lbp_prop));
(void) printf("|\t\tcompralgo: %llu\n",
- (u_longlong_t)L2BLK_GET_COMPRESS((&lbps)->lbp_prop));
+ (u_longlong_t)L2BLK_GET_COMPRESS(lbps->lbp_prop));
(void) printf("|\t\tcksumalgo: %llu\n",
- (u_longlong_t)L2BLK_GET_CHECKSUM((&lbps)->lbp_prop));
+ (u_longlong_t)L2BLK_GET_CHECKSUM(lbps->lbp_prop));
(void) printf("|\n\n");
}
static void
-dump_l2arc_log_blocks(int fd, l2arc_dev_hdr_phys_t l2dhdr,
+dump_l2arc_log_blocks(int fd, const l2arc_dev_hdr_phys_t *l2dhdr,
l2arc_dev_hdr_phys_t *rebuild)
{
l2arc_log_blk_phys_t this_lb;
if (!dump_opt['q'])
print_l2arc_log_blocks();
- bcopy((&l2dhdr)->dh_start_lbps, lbps, sizeof (lbps));
+ memcpy(lbps, l2dhdr->dh_start_lbps, sizeof (lbps));
- dev.l2ad_evict = l2dhdr.dh_evict;
- dev.l2ad_start = l2dhdr.dh_start;
- dev.l2ad_end = l2dhdr.dh_end;
+ dev.l2ad_evict = l2dhdr->dh_evict;
+ dev.l2ad_start = l2dhdr->dh_start;
+ dev.l2ad_end = l2dhdr->dh_end;
- if (l2dhdr.dh_start_lbps[0].lbp_daddr == 0) {
+ if (l2dhdr->dh_start_lbps[0].lbp_daddr == 0) {
/* no log blocks to read */
if (!dump_opt['q']) {
(void) printf("No log blocks to read\n");
L2BLK_GET_PSIZE((&lbps[0])->lbp_prop);
}
- dev.l2ad_first = !!(l2dhdr.dh_flags & L2ARC_DEV_HDR_EVICT_FIRST);
+ dev.l2ad_first = !!(l2dhdr->dh_flags & L2ARC_DEV_HDR_EVICT_FIRST);
for (;;) {
if (!l2arc_log_blkptr_valid(&dev, &lbps[0]))
failed++;
if (!dump_opt['q']) {
(void) printf("Invalid cksum\n");
- dump_l2arc_log_blkptr(lbps[0]);
+ dump_l2arc_log_blkptr(&lbps[0]);
}
break;
}
default:
abd = abd_alloc_for_io(asize, B_TRUE);
abd_copy_from_buf_off(abd, &this_lb, 0, asize);
- zio_decompress_data(L2BLK_GET_COMPRESS(
+ if (zio_decompress_data(L2BLK_GET_COMPRESS(
(&lbps[0])->lbp_prop), abd, &this_lb,
- asize, sizeof (this_lb), NULL);
+ asize, sizeof (this_lb), NULL) != 0) {
+ (void) printf("L2ARC block decompression "
+ "failed\n");
+ abd_free(abd);
+ goto out;
+ }
abd_free(abd);
break;
}
(void) printf("lb[%4llu]\tmagic: %llu\n",
(u_longlong_t)rebuild->dh_lb_count,
(u_longlong_t)this_lb.lb_magic);
- dump_l2arc_log_blkptr(lbps[0]);
+ dump_l2arc_log_blkptr(&lbps[0]);
}
if (dump_opt['l'] > 2 && !dump_opt['q'])
- dump_l2arc_log_entries(l2dhdr.dh_log_entries,
+ dump_l2arc_log_entries(l2dhdr->dh_log_entries,
this_lb.lb_entries,
rebuild->dh_lb_count);
lbps[0] = lbps[1];
lbps[1] = this_lb.lb_prev_lbp;
}
-
+out:
if (!dump_opt['q']) {
(void) printf("log_blk_count:\t %llu with valid cksum\n",
(u_longlong_t)rebuild->dh_lb_count);
static int
dump_l2arc_header(int fd)
{
- l2arc_dev_hdr_phys_t l2dhdr, rebuild;
+ l2arc_dev_hdr_phys_t l2dhdr = {0}, rebuild = {0};
int error = B_FALSE;
- bzero(&l2dhdr, sizeof (l2dhdr));
- bzero(&rebuild, sizeof (rebuild));
-
if (pread64(fd, &l2dhdr, sizeof (l2dhdr),
VDEV_LABEL_START_SIZE) != sizeof (l2dhdr)) {
error = B_TRUE;
(u_longlong_t)l2dhdr.dh_trim_state);
}
- dump_l2arc_log_blocks(fd, l2dhdr, &rebuild);
+ dump_l2arc_log_blocks(fd, &l2dhdr, &rebuild);
/*
* The total aligned size of log blocks and the number of log blocks
* reported in the header of the device may be less than what zdb
case DMU_OT_DIRECTORY_CONTENTS:
if (s != NULL && *(s + 1) != '\0')
return (dump_path_impl(os, child_obj, s + 1, retobj));
- fallthrough;
+ zfs_fallthrough;
case DMU_OT_PLAIN_FILE_CONTENTS:
if (retobj != NULL) {
*retobj = child_obj;
return (err);
}
+static int
+dump_backup_bytes(objset_t *os, void *buf, int len, void *arg)
+{
+ const char *p = (const char *)buf;
+ ssize_t nwritten;
+
+ (void) os;
+ (void) arg;
+
+ /* Write the data out, handling short writes and signals. */
+ while ((nwritten = write(STDOUT_FILENO, p, len)) < len) {
+ if (nwritten < 0) {
+ if (errno == EINTR)
+ continue;
+ return (errno);
+ }
+ p += nwritten;
+ len -= nwritten;
+ }
+
+ return (0);
+}
+
+static void
+dump_backup(const char *pool, uint64_t objset_id, const char *flagstr)
+{
+ boolean_t embed = B_FALSE;
+ boolean_t large_block = B_FALSE;
+ boolean_t compress = B_FALSE;
+ boolean_t raw = B_FALSE;
+
+ const char *c;
+ for (c = flagstr; c != NULL && *c != '\0'; c++) {
+ switch (*c) {
+ case 'e':
+ embed = B_TRUE;
+ break;
+ case 'L':
+ large_block = B_TRUE;
+ break;
+ case 'c':
+ compress = B_TRUE;
+ break;
+ case 'w':
+ raw = B_TRUE;
+ break;
+ default:
+ fprintf(stderr, "dump_backup: invalid flag "
+ "'%c'\n", *c);
+ return;
+ }
+ }
+
+ if (isatty(STDOUT_FILENO)) {
+ fprintf(stderr, "dump_backup: stream cannot be written "
+ "to a terminal\n");
+ return;
+ }
+
+ offset_t off = 0;
+ dmu_send_outparams_t out = {
+ .dso_outfunc = dump_backup_bytes,
+ .dso_dryrun = B_FALSE,
+ };
+
+ int err = dmu_send_obj(pool, objset_id, /* fromsnap */0, embed,
+ large_block, compress, raw, /* saved */ B_FALSE, STDOUT_FILENO,
+ &off, &out);
+ if (err != 0) {
+ fprintf(stderr, "dump_backup: dmu_send_obj: %s\n",
+ strerror(err));
+ return;
+ }
+}
+
static int
zdb_copy_object(objset_t *os, uint64_t srcobj, char *destfile)
{
}
int fd = open(destfile, O_WRONLY | O_CREAT | O_TRUNC, 0644);
+ if (fd == -1)
+ return (errno);
/*
* We cap the size at 1 mebibyte here to prevent
* allocation failures and nigh-infinite printing if the
offset = 0;
char *buf = kmem_alloc(oursize, KM_NOSLEEP);
if (buf == NULL) {
+ (void) close(fd);
return (ENOMEM);
}
if (err != 0) {
(void) printf("got error %u from dmu_read\n", err);
kmem_free(buf, oursize);
+ (void) close(fd);
return (err);
}
if (dump_opt['v'] > 3) {
return (err);
}
+static boolean_t
+label_cksum_valid(vdev_label_t *label, uint64_t offset)
+{
+ zio_checksum_info_t *ci = &zio_checksum_table[ZIO_CHECKSUM_LABEL];
+ zio_cksum_t expected_cksum;
+ zio_cksum_t actual_cksum;
+ zio_cksum_t verifier;
+ zio_eck_t *eck;
+ int byteswap;
+
+ void *data = (char *)label + offsetof(vdev_label_t, vl_vdev_phys);
+ eck = (zio_eck_t *)((char *)(data) + VDEV_PHYS_SIZE) - 1;
+
+ offset += offsetof(vdev_label_t, vl_vdev_phys);
+ ZIO_SET_CHECKSUM(&verifier, offset, 0, 0, 0);
+
+ byteswap = (eck->zec_magic == BSWAP_64(ZEC_MAGIC));
+ if (byteswap)
+ byteswap_uint64_array(&verifier, sizeof (zio_cksum_t));
+
+ expected_cksum = eck->zec_cksum;
+ eck->zec_cksum = verifier;
+
+ abd_t *abd = abd_get_from_buf(data, VDEV_PHYS_SIZE);
+ ci->ci_func[byteswap](abd, VDEV_PHYS_SIZE, NULL, &actual_cksum);
+ abd_free(abd);
+
+ if (byteswap)
+ byteswap_uint64_array(&expected_cksum, sizeof (zio_cksum_t));
+
+ if (ZIO_CHECKSUM_EQUAL(actual_cksum, expected_cksum))
+ return (B_TRUE);
+
+ return (B_FALSE);
+}
+
static int
dump_label(const char *dev)
{
char path[MAXPATHLEN];
- zdb_label_t labels[VDEV_LABELS];
+ zdb_label_t labels[VDEV_LABELS] = {{{{0}}}};
uint64_t psize, ashift, l2cache;
struct stat64 statbuf;
boolean_t config_found = B_FALSE;
void *node, *cookie;
int fd;
- bzero(labels, sizeof (labels));
-
/*
* Check if we were given absolute path and use it as is.
* Otherwise if the provided vdev name doesn't point to a file,
/*
* 1. Read the label from disk
- * 2. Unpack the configuration and insert in config tree.
- * 3. Traverse all uberblocks and insert in uberblock tree.
+ * 2. Verify label cksum
+ * 3. Unpack the configuration and insert in config tree.
+ * 4. Traverse all uberblocks and insert in uberblock tree.
*/
for (int l = 0; l < VDEV_LABELS; l++) {
zdb_label_t *label = &labels[l];
zio_cksum_t cksum;
vdev_t vd;
+ label->label_offset = vdev_label_offset(psize, l, 0);
+
if (pread64(fd, &label->label, sizeof (label->label),
- vdev_label_offset(psize, l, 0)) != sizeof (label->label)) {
+ label->label_offset) != sizeof (label->label)) {
if (!dump_opt['q'])
(void) printf("failed to read label %d\n", l);
label->read_failed = B_TRUE;
}
label->read_failed = B_FALSE;
+ label->cksum_valid = label_cksum_valid(&label->label,
+ label->label_offset);
if (nvlist_unpack(buf, buflen, &config, 0) == 0) {
nvlist_t *vdev_tree = NULL;
if (nvlist_size(config, &size, NV_ENCODE_XDR) != 0)
size = buflen;
- /* If the device is a cache device clear the header. */
+ /* If the device is a cache device read the header. */
if (!read_l2arc_header) {
if (nvlist_lookup_uint64(config,
ZPOOL_CONFIG_POOL_STATE, &l2cache) == 0 &&
static uint64_t global_feature_count[SPA_FEATURES];
static uint64_t remap_deadlist_count = 0;
-/*ARGSUSED*/
static int
dump_one_objset(const char *dsname, void *arg)
{
+ (void) arg;
int error;
objset_t *os;
spa_feature_t f;
avl_first(&dmu_objset_ds(os)->ds_bookmarks); dbn != NULL;
dbn = AVL_NEXT(&dmu_objset_ds(os)->ds_bookmarks, dbn)) {
mos_obj_refd(dbn->dbn_phys.zbm_redaction_obj);
- if (dbn->dbn_phys.zbm_redaction_obj != 0)
- global_feature_count[SPA_FEATURE_REDACTION_BOOKMARKS]++;
+ if (dbn->dbn_phys.zbm_redaction_obj != 0) {
+ global_feature_count[
+ SPA_FEATURE_REDACTION_BOOKMARKS]++;
+ objset_t *mos = os->os_spa->spa_meta_objset;
+ dnode_t *rl;
+ VERIFY0(dnode_hold(mos,
+ dbn->dbn_phys.zbm_redaction_obj, FTAG, &rl));
+ if (rl->dn_have_spill) {
+ global_feature_count[
+ SPA_FEATURE_REDACTION_LIST_SPILL]++;
+ }
+ }
if (dbn->dbn_phys.zbm_flags & ZBM_FLAG_HAS_FBN)
global_feature_count[SPA_FEATURE_BOOKMARK_WRITTEN]++;
}
#define ZB_TOTAL DN_MAX_LEVELS
#define SPA_MAX_FOR_16M (SPA_MAXBLOCKSHIFT+1)
+typedef struct zdb_brt_entry {
+ dva_t zbre_dva;
+ uint64_t zbre_refcount;
+ avl_node_t zbre_node;
+} zdb_brt_entry_t;
+
typedef struct zdb_cb {
zdb_blkstats_t zcb_type[ZB_TOTAL + 1][ZDB_OT_TOTAL + 1];
uint64_t zcb_removing_size;
uint64_t zcb_checkpoint_size;
uint64_t zcb_dedup_asize;
uint64_t zcb_dedup_blocks;
+ uint64_t zcb_clone_asize;
+ uint64_t zcb_clone_blocks;
uint64_t zcb_psize_count[SPA_MAX_FOR_16M];
uint64_t zcb_lsize_count[SPA_MAX_FOR_16M];
uint64_t zcb_asize_count[SPA_MAX_FOR_16M];
int zcb_haderrors;
spa_t *zcb_spa;
uint32_t **zcb_vd_obsolete_counts;
+ avl_tree_t zcb_brt;
+ boolean_t zcb_brt_is_active;
} zdb_cb_t;
/* test if two DVA offsets from same vdev are within the same metaslab */
* Used to simplify reporting of the histogram data.
*/
typedef struct one_histo {
- char *name;
+ const char *name;
uint64_t *count;
uint64_t *len;
uint64_t cumulative;
zcb->zcb_asize_len[bin] += BP_GET_ASIZE(bp);
zcb->zcb_asize_total += BP_GET_ASIZE(bp);
+ if (zcb->zcb_brt_is_active && brt_maybe_exists(zcb->zcb_spa, bp)) {
+ /*
+ * Cloned blocks are special. We need to count them, so we can
+ * later uncount them when reporting leaked space, and we must
+ * only claim them them once.
+ *
+ * To do this, we keep our own in-memory BRT. For each block
+ * we haven't seen before, we look it up in the real BRT and
+ * if its there, we note it and its refcount then proceed as
+ * normal. If we see the block again, we count it as a clone
+ * and then give it no further consideration.
+ */
+ zdb_brt_entry_t zbre_search, *zbre;
+ avl_index_t where;
+
+ zbre_search.zbre_dva = bp->blk_dva[0];
+ zbre = avl_find(&zcb->zcb_brt, &zbre_search, &where);
+ if (zbre != NULL) {
+ zcb->zcb_clone_asize += BP_GET_ASIZE(bp);
+ zcb->zcb_clone_blocks++;
+
+ zbre->zbre_refcount--;
+ if (zbre->zbre_refcount == 0) {
+ avl_remove(&zcb->zcb_brt, zbre);
+ umem_free(zbre, sizeof (zdb_brt_entry_t));
+ }
+ return;
+ }
+
+ uint64_t crefcnt = brt_entry_get_refcount(zcb->zcb_spa, bp);
+ if (crefcnt > 0) {
+ zbre = umem_zalloc(sizeof (zdb_brt_entry_t),
+ UMEM_NOFAIL);
+ zbre->zbre_dva = bp->blk_dva[0];
+ zbre->zbre_refcount = crefcnt;
+ avl_insert(&zcb->zcb_brt, zbre, where);
+ }
+ }
+
if (dump_opt['L'])
return;
(zcb->zcb_totalasize - bytes) / 1024 / kb_per_sec;
/* make sure nicenum has enough space */
- CTASSERT(sizeof (buf) >= NN_NUMBUF_SZ);
+ _Static_assert(sizeof (buf) >= NN_NUMBUF_SZ, "buf truncated");
zfs_nicebytes(bytes, buf, sizeof (buf));
(void) fprintf(stderr,
NULL /* alloc */
};
-/* ARGSUSED */
static int
load_unflushed_svr_segs_cb(spa_t *spa, space_map_entry_t *sme,
uint64_t txg, void *arg)
return (0);
}
-/* ARGSUSED */
static void
claim_segment_impl_cb(uint64_t inner_offset, vdev_t *vd, uint64_t offset,
uint64_t size, void *arg)
{
+ (void) inner_offset, (void) arg;
+
/*
* This callback was called through a remap from
* a device being removed. Therefore, the vdev that
spa_config_exit(spa, SCL_CONFIG, FTAG);
}
-/* ARGSUSED */
static int
increment_indirect_mapping_cb(void *arg, const blkptr_t *bp, boolean_t bp_freed,
dmu_tx_t *tx)
{
+ (void) tx;
zdb_cb_t *zcb = arg;
spa_t *spa = zcb->zcb_spa;
vdev_t *vd;
static void
zdb_ddt_leak_init(spa_t *spa, zdb_cb_t *zcb)
{
- ddt_bookmark_t ddb;
+ ddt_bookmark_t ddb = {0};
ddt_entry_t dde;
int error;
int p;
ASSERT(!dump_opt['L']);
- bzero(&ddb, sizeof (ddb));
while ((error = ddt_walk(spa, &ddb, &dde)) == 0) {
blkptr_t blk;
ddt_phys_t *ddp = dde.dde_phys;
*/
for (uint64_t inner_offset = 0;
inner_offset < DVA_GET_ASIZE(&vimep->vimep_dst);
- inner_offset += 1 << vd->vdev_ashift) {
+ inner_offset += 1ULL << vd->vdev_ashift) {
if (range_tree_contains(msp->ms_allocatable,
- offset + inner_offset, 1 << vd->vdev_ashift)) {
- obsolete_bytes += 1 << vd->vdev_ashift;
+ offset + inner_offset, 1ULL << vd->vdev_ashift)) {
+ obsolete_bytes += 1ULL << vd->vdev_ashift;
}
}
return (leaks);
}
-/* ARGSUSED */
static int
count_block_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx)
{
+ (void) tx;
zdb_cb_t *zcb = arg;
if (dump_opt['b'] >= 5) {
iterate_deleted_livelists(spa, dump_livelist_cb, NULL);
}
+static int
+zdb_brt_entry_compare(const void *zcn1, const void *zcn2)
+{
+ const dva_t *dva1 = &((const zdb_brt_entry_t *)zcn1)->zbre_dva;
+ const dva_t *dva2 = &((const zdb_brt_entry_t *)zcn2)->zbre_dva;
+ int cmp;
+
+ cmp = TREE_CMP(DVA_GET_VDEV(dva1), DVA_GET_VDEV(dva2));
+ if (cmp == 0)
+ cmp = TREE_CMP(DVA_GET_OFFSET(dva1), DVA_GET_OFFSET(dva2));
+
+ return (cmp);
+}
+
static int
dump_block_stats(spa_t *spa)
{
- zdb_cb_t zcb;
+ zdb_cb_t *zcb;
zdb_blkstats_t *zb, *tzb;
uint64_t norm_alloc, norm_space, total_alloc, total_found;
int flags = TRAVERSE_PRE | TRAVERSE_PREFETCH_METADATA |
int e, c, err;
bp_embedded_type_t i;
- bzero(&zcb, sizeof (zcb));
+ zcb = umem_zalloc(sizeof (zdb_cb_t), UMEM_NOFAIL);
+
+ if (spa_feature_is_active(spa, SPA_FEATURE_BLOCK_CLONING)) {
+ avl_create(&zcb->zcb_brt, zdb_brt_entry_compare,
+ sizeof (zdb_brt_entry_t),
+ offsetof(zdb_brt_entry_t, zbre_node));
+ zcb->zcb_brt_is_active = B_TRUE;
+ }
+
(void) printf("\nTraversing all blocks %s%s%s%s%s...\n\n",
(dump_opt['c'] || !dump_opt['L']) ? "to verify " : "",
(dump_opt['c'] == 1) ? "metadata " : "",
* pool claiming each block we discover, but we skip opening any space
* maps.
*/
- bzero(&zcb, sizeof (zdb_cb_t));
- zdb_leak_init(spa, &zcb);
+ zdb_leak_init(spa, zcb);
/*
* If there's a deferred-free bplist, process that first.
*/
(void) bpobj_iterate_nofree(&spa->spa_deferred_bpobj,
- bpobj_count_block_cb, &zcb, NULL);
+ bpobj_count_block_cb, zcb, NULL);
if (spa_version(spa) >= SPA_VERSION_DEADLISTS) {
(void) bpobj_iterate_nofree(&spa->spa_dsl_pool->dp_free_bpobj,
- bpobj_count_block_cb, &zcb, NULL);
+ bpobj_count_block_cb, zcb, NULL);
}
- zdb_claim_removing(spa, &zcb);
+ zdb_claim_removing(spa, zcb);
if (spa_feature_is_active(spa, SPA_FEATURE_ASYNC_DESTROY)) {
VERIFY3U(0, ==, bptree_iterate(spa->spa_meta_objset,
spa->spa_dsl_pool->dp_bptree_obj, B_FALSE, count_block_cb,
- &zcb, NULL));
+ zcb, NULL));
}
- deleted_livelists_count_blocks(spa, &zcb);
+ deleted_livelists_count_blocks(spa, zcb);
if (dump_opt['c'] > 1)
flags |= TRAVERSE_PREFETCH_DATA;
- zcb.zcb_totalasize = metaslab_class_get_alloc(spa_normal_class(spa));
- zcb.zcb_totalasize += metaslab_class_get_alloc(spa_special_class(spa));
- zcb.zcb_totalasize += metaslab_class_get_alloc(spa_dedup_class(spa));
- zcb.zcb_totalasize +=
+ zcb->zcb_totalasize = metaslab_class_get_alloc(spa_normal_class(spa));
+ zcb->zcb_totalasize += metaslab_class_get_alloc(spa_special_class(spa));
+ zcb->zcb_totalasize += metaslab_class_get_alloc(spa_dedup_class(spa));
+ zcb->zcb_totalasize +=
metaslab_class_get_alloc(spa_embedded_log_class(spa));
- zcb.zcb_start = zcb.zcb_lastprint = gethrtime();
- err = traverse_pool(spa, 0, flags, zdb_blkptr_cb, &zcb);
+ zcb->zcb_start = zcb->zcb_lastprint = gethrtime();
+ err = traverse_pool(spa, 0, flags, zdb_blkptr_cb, zcb);
/*
* If we've traversed the data blocks then we need to wait for those
* Done after zio_wait() since zcb_haderrors is modified in
* zdb_blkptr_done()
*/
- zcb.zcb_haderrors |= err;
+ zcb->zcb_haderrors |= err;
- if (zcb.zcb_haderrors) {
+ if (zcb->zcb_haderrors) {
(void) printf("\nError counts:\n\n");
(void) printf("\t%5s %s\n", "errno", "count");
for (e = 0; e < 256; e++) {
- if (zcb.zcb_errors[e] != 0) {
+ if (zcb->zcb_errors[e] != 0) {
(void) printf("\t%5d %llu\n",
- e, (u_longlong_t)zcb.zcb_errors[e]);
+ e, (u_longlong_t)zcb->zcb_errors[e]);
}
}
}
/*
* Report any leaked segments.
*/
- leaks |= zdb_leak_fini(spa, &zcb);
+ leaks |= zdb_leak_fini(spa, zcb);
- tzb = &zcb.zcb_type[ZB_TOTAL][ZDB_OT_TOTAL];
+ tzb = &zcb->zcb_type[ZB_TOTAL][ZDB_OT_TOTAL];
norm_alloc = metaslab_class_get_alloc(spa_normal_class(spa));
norm_space = metaslab_class_get_space(spa_normal_class(spa));
metaslab_class_get_alloc(spa_special_class(spa)) +
metaslab_class_get_alloc(spa_dedup_class(spa)) +
get_unflushed_alloc_space(spa);
- total_found = tzb->zb_asize - zcb.zcb_dedup_asize +
- zcb.zcb_removing_size + zcb.zcb_checkpoint_size;
+ total_found =
+ tzb->zb_asize - zcb->zcb_dedup_asize - zcb->zcb_clone_asize +
+ zcb->zcb_removing_size + zcb->zcb_checkpoint_size;
if (total_found == total_alloc && !dump_opt['L']) {
(void) printf("\n\tNo leaks (block sum matches space"
leaks = B_TRUE;
}
- if (tzb->zb_count == 0)
+ if (tzb->zb_count == 0) {
+ umem_free(zcb, sizeof (zdb_cb_t));
return (2);
+ }
(void) printf("\n");
(void) printf("\t%-16s %14llu\n", "bp count:",
(u_longlong_t)(tzb->zb_asize / tzb->zb_count),
(double)tzb->zb_lsize / tzb->zb_asize);
(void) printf("\t%-16s %14llu ref>1: %6llu deduplication: %6.2f\n",
- "bp deduped:", (u_longlong_t)zcb.zcb_dedup_asize,
- (u_longlong_t)zcb.zcb_dedup_blocks,
- (double)zcb.zcb_dedup_asize / tzb->zb_asize + 1.0);
+ "bp deduped:", (u_longlong_t)zcb->zcb_dedup_asize,
+ (u_longlong_t)zcb->zcb_dedup_blocks,
+ (double)zcb->zcb_dedup_asize / tzb->zb_asize + 1.0);
+ (void) printf("\t%-16s %14llu count: %6llu\n",
+ "bp cloned:", (u_longlong_t)zcb->zcb_clone_asize,
+ (u_longlong_t)zcb->zcb_clone_blocks);
(void) printf("\t%-16s %14llu used: %5.2f%%\n", "Normal class:",
(u_longlong_t)norm_alloc, 100.0 * norm_alloc / norm_space);
}
for (i = 0; i < NUM_BP_EMBEDDED_TYPES; i++) {
- if (zcb.zcb_embedded_blocks[i] == 0)
+ if (zcb->zcb_embedded_blocks[i] == 0)
continue;
(void) printf("\n");
(void) printf("\tadditional, non-pointer bps of type %u: "
"%10llu\n",
- i, (u_longlong_t)zcb.zcb_embedded_blocks[i]);
+ i, (u_longlong_t)zcb->zcb_embedded_blocks[i]);
if (dump_opt['b'] >= 3) {
(void) printf("\t number of (compressed) bytes: "
"number of bps\n");
- dump_histogram(zcb.zcb_embedded_histogram[i],
- sizeof (zcb.zcb_embedded_histogram[i]) /
- sizeof (zcb.zcb_embedded_histogram[i][0]), 0);
+ dump_histogram(zcb->zcb_embedded_histogram[i],
+ sizeof (zcb->zcb_embedded_histogram[i]) /
+ sizeof (zcb->zcb_embedded_histogram[i][0]), 0);
}
}
if (dump_opt['b'] >= 2) {
int l, t, level;
+ char csize[32], lsize[32], psize[32], asize[32];
+ char avg[32], gang[32];
(void) printf("\nBlocks\tLSIZE\tPSIZE\tASIZE"
"\t avg\t comp\t%%Total\tType\n");
+ zfs_blkstat_t *mdstats = umem_zalloc(sizeof (zfs_blkstat_t),
+ UMEM_NOFAIL);
+
for (t = 0; t <= ZDB_OT_TOTAL; t++) {
- char csize[32], lsize[32], psize[32], asize[32];
- char avg[32], gang[32];
const char *typename;
/* make sure nicenum has enough space */
- CTASSERT(sizeof (csize) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (lsize) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (psize) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (asize) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (avg) >= NN_NUMBUF_SZ);
- CTASSERT(sizeof (gang) >= NN_NUMBUF_SZ);
+ _Static_assert(sizeof (csize) >= NN_NUMBUF_SZ,
+ "csize truncated");
+ _Static_assert(sizeof (lsize) >= NN_NUMBUF_SZ,
+ "lsize truncated");
+ _Static_assert(sizeof (psize) >= NN_NUMBUF_SZ,
+ "psize truncated");
+ _Static_assert(sizeof (asize) >= NN_NUMBUF_SZ,
+ "asize truncated");
+ _Static_assert(sizeof (avg) >= NN_NUMBUF_SZ,
+ "avg truncated");
+ _Static_assert(sizeof (gang) >= NN_NUMBUF_SZ,
+ "gang truncated");
if (t < DMU_OT_NUMTYPES)
typename = dmu_ot[t].ot_name;
else
typename = zdb_ot_extname[t - DMU_OT_NUMTYPES];
- if (zcb.zcb_type[ZB_TOTAL][t].zb_asize == 0) {
+ if (zcb->zcb_type[ZB_TOTAL][t].zb_asize == 0) {
(void) printf("%6s\t%5s\t%5s\t%5s"
"\t%5s\t%5s\t%6s\t%s\n",
"-",
for (l = ZB_TOTAL - 1; l >= -1; l--) {
level = (l == -1 ? ZB_TOTAL : l);
- zb = &zcb.zcb_type[level][t];
+ zb = &zcb->zcb_type[level][t];
if (zb->zb_asize == 0)
continue;
+ if (level != ZB_TOTAL && t < DMU_OT_NUMTYPES &&
+ (level > 0 || DMU_OT_IS_METADATA(t))) {
+ mdstats->zb_count += zb->zb_count;
+ mdstats->zb_lsize += zb->zb_lsize;
+ mdstats->zb_psize += zb->zb_psize;
+ mdstats->zb_asize += zb->zb_asize;
+ mdstats->zb_gangs += zb->zb_gangs;
+ }
+
if (dump_opt['b'] < 3 && level != ZB_TOTAL)
continue;
if (level == 0 && zb->zb_asize ==
- zcb.zcb_type[ZB_TOTAL][t].zb_asize)
+ zcb->zcb_type[ZB_TOTAL][t].zb_asize)
continue;
zdb_nicenum(zb->zb_count, csize,
}
}
}
+ zdb_nicenum(mdstats->zb_count, csize,
+ sizeof (csize));
+ zdb_nicenum(mdstats->zb_lsize, lsize,
+ sizeof (lsize));
+ zdb_nicenum(mdstats->zb_psize, psize,
+ sizeof (psize));
+ zdb_nicenum(mdstats->zb_asize, asize,
+ sizeof (asize));
+ zdb_nicenum(mdstats->zb_asize / mdstats->zb_count, avg,
+ sizeof (avg));
+ zdb_nicenum(mdstats->zb_gangs, gang, sizeof (gang));
+
+ (void) printf("%6s\t%5s\t%5s\t%5s\t%5s"
+ "\t%5.2f\t%6.2f\t",
+ csize, lsize, psize, asize, avg,
+ (double)mdstats->zb_lsize / mdstats->zb_psize,
+ 100.0 * mdstats->zb_asize / tzb->zb_asize);
+ (void) printf("%s\n", "Metadata Total");
/* Output a table summarizing block sizes in the pool */
if (dump_opt['b'] >= 2) {
- dump_size_histograms(&zcb);
+ dump_size_histograms(zcb);
}
+
+ umem_free(mdstats, sizeof (zfs_blkstat_t));
}
(void) printf("\n");
- if (leaks)
+ if (leaks) {
+ umem_free(zcb, sizeof (zdb_cb_t));
return (2);
+ }
- if (zcb.zcb_haderrors)
+ if (zcb->zcb_haderrors) {
+ umem_free(zcb, sizeof (zdb_cb_t));
return (3);
+ }
+ umem_free(zcb, sizeof (zdb_cb_t));
return (0);
}
typedef struct zdb_ddt_entry {
+ /* key must be first for ddt_key_compare */
ddt_key_t zdde_key;
uint64_t zdde_ref_blocks;
uint64_t zdde_ref_lsize;
avl_node_t zdde_node;
} zdb_ddt_entry_t;
-/* ARGSUSED */
static int
zdb_ddt_add_cb(spa_t *spa, zilog_t *zilog, const blkptr_t *bp,
const zbookmark_phys_t *zb, const dnode_phys_t *dnp, void *arg)
{
+ (void) zilog, (void) dnp;
avl_tree_t *t = arg;
avl_index_t where;
zdb_ddt_entry_t *zdde, zdde_search;
avl_tree_t t;
void *cookie = NULL;
zdb_ddt_entry_t *zdde;
- ddt_histogram_t ddh_total;
- ddt_stat_t dds_total;
+ ddt_histogram_t ddh_total = {{{0}}};
+ ddt_stat_t dds_total = {0};
- bzero(&ddh_total, sizeof (ddh_total));
- bzero(&dds_total, sizeof (dds_total));
- avl_create(&t, ddt_entry_compare,
+ avl_create(&t, ddt_key_compare,
sizeof (zdb_ddt_entry_t), offsetof(zdb_ddt_entry_t, zdde_node));
spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
freecfg = B_TRUE;
}
- if (asprintf(&bogus_name, "%s%s", poolname, BOGUS_SUFFIX) == -1)
+ if (asprintf(&bogus_name, "%s%s", poolname, BOGUS_SUFFIX) == -1) {
+ if (target != poolname)
+ free(poolname);
return (NULL);
+ }
fnvlist_add_string(cfg, ZPOOL_CONFIG_POOL_NAME, bogus_name);
error = spa_import(bogus_name, cfg, NULL,
if (new_path != NULL && path_start != NULL) {
if (asprintf(new_path, "%s%s", bogus_name, path_start) == -1) {
+ free(bogus_name);
if (path_start != NULL)
free(poolname);
return (NULL);
return (error);
}
-/* ARGSUSED */
static void
mos_leaks_cb(void *arg, uint64_t start, uint64_t size)
{
+ (void) arg;
for (uint64_t i = start; i < size; i++) {
(void) printf("MOS object %llu referenced but not allocated\n",
(u_longlong_t)i);
mos_obj_refd(space_map_object(ms->ms_sm));
}
+ if (vd->vdev_root_zap != 0)
+ mos_obj_refd(vd->vdev_root_zap);
+
if (vd->vdev_top_zap != 0) {
mos_obj_refd(vd->vdev_top_zap);
mos_leak_vdev_top_zap(vd);
mos_obj_refd(sls->sls_sm_obj);
}
+static void
+errorlog_count_refd(objset_t *mos, uint64_t errlog)
+{
+ zap_cursor_t zc;
+ zap_attribute_t za;
+ for (zap_cursor_init(&zc, mos, errlog);
+ zap_cursor_retrieve(&zc, &za) == 0;
+ zap_cursor_advance(&zc)) {
+ mos_obj_refd(za.za_first_integer);
+ }
+ zap_cursor_fini(&zc);
+}
+
static int
dump_mos_leaks(spa_t *spa)
{
mos_obj_refd(spa->spa_history);
mos_obj_refd(spa->spa_errlog_last);
mos_obj_refd(spa->spa_errlog_scrub);
+
+ if (spa_feature_is_enabled(spa, SPA_FEATURE_HEAD_ERRLOG)) {
+ errorlog_count_refd(mos, spa->spa_errlog_last);
+ errorlog_count_refd(mos, spa->spa_errlog_scrub);
+ }
+
mos_obj_refd(spa->spa_all_vdev_zaps);
mos_obj_refd(spa->spa_dsl_pool->dp_bptree_obj);
mos_obj_refd(spa->spa_dsl_pool->dp_tmp_userrefs_obj);
}
}
+ if (spa->spa_brt != NULL) {
+ brt_t *brt = spa->spa_brt;
+ for (uint64_t vdevid = 0; vdevid < brt->brt_nvdevs; vdevid++) {
+ brt_vdev_t *brtvd = &brt->brt_vdevs[vdevid];
+ if (brtvd != NULL && brtvd->bv_initiated) {
+ mos_obj_refd(brtvd->bv_mos_brtvdev);
+ mos_obj_refd(brtvd->bv_mos_entries);
+ }
+ }
+ }
+
/*
* Visit all allocated objects and make sure they are referenced.
*/
} else {
dmu_object_info_t doi;
const char *name;
- dmu_object_info(mos, object, &doi);
+ VERIFY0(dmu_object_info(mos, object, &doi));
if (doi.doi_type & DMU_OT_NEWTYPE) {
dmu_object_byteswap_t bswap =
DMU_OT_BYTESWAP(doi.doi_type);
if (!spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP))
return;
- log_sm_obsolete_stats_arg_t lsos;
- bzero(&lsos, sizeof (lsos));
+ log_sm_obsolete_stats_arg_t lsos = {0};
(void) printf("Log Space Map Obsolete Entry Statistics:\n");
if (dump_opt['D'])
dump_all_ddts(spa);
+ if (dump_opt['T'])
+ dump_brt(spa);
+
if (dump_opt['d'] > 2 || dump_opt['m'])
dump_metaslabs(spa);
if (dump_opt['M'])
- dump_metaslab_groups(spa);
+ dump_metaslab_groups(spa, dump_opt['M'] > 1);
if (dump_opt['d'] > 2 || dump_opt['m']) {
dump_log_spacemaps(spa);
dump_log_spacemap_obsolete_stats(spa);
for (spa_feature_t f = 0; f < SPA_FEATURES; f++)
global_feature_count[f] = UINT64_MAX;
global_feature_count[SPA_FEATURE_REDACTION_BOOKMARKS] = 0;
+ global_feature_count[SPA_FEATURE_REDACTION_LIST_SPILL] = 0;
global_feature_count[SPA_FEATURE_BOOKMARK_WRITTEN] = 0;
global_feature_count[SPA_FEATURE_LIVELIST] = 0;
#define ZIO_COMPRESS_MASK(alg) (1ULL << (ZIO_COMPRESS_##alg))
static boolean_t
+try_decompress_block(abd_t *pabd, uint64_t lsize, uint64_t psize,
+ int flags, int cfunc, void *lbuf, void *lbuf2)
+{
+ if (flags & ZDB_FLAG_VERBOSE) {
+ (void) fprintf(stderr,
+ "Trying %05llx -> %05llx (%s)\n",
+ (u_longlong_t)psize,
+ (u_longlong_t)lsize,
+ zio_compress_table[cfunc].ci_name);
+ }
+
+ /*
+ * We set lbuf to all zeros and lbuf2 to all
+ * ones, then decompress to both buffers and
+ * compare their contents. This way we can
+ * know if decompression filled exactly to
+ * lsize or if it left some bytes unwritten.
+ */
+
+ memset(lbuf, 0x00, lsize);
+ memset(lbuf2, 0xff, lsize);
+
+ if (zio_decompress_data(cfunc, pabd,
+ lbuf, psize, lsize, NULL) == 0 &&
+ zio_decompress_data(cfunc, pabd,
+ lbuf2, psize, lsize, NULL) == 0 &&
+ memcmp(lbuf, lbuf2, lsize) == 0)
+ return (B_TRUE);
+ return (B_FALSE);
+}
+
+static uint64_t
zdb_decompress_block(abd_t *pabd, void *buf, void *lbuf, uint64_t lsize,
uint64_t psize, int flags)
{
- boolean_t exceeded = B_FALSE;
+ (void) buf;
+ uint64_t orig_lsize = lsize;
+ boolean_t tryzle = ((getenv("ZDB_NO_ZLE") == NULL));
+ boolean_t found = B_FALSE;
/*
* We don't know how the data was compressed, so just try
* every decompress function at every inflated blocksize.
uint64_t maxlsize = SPA_MAXBLOCKSIZE;
uint64_t mask = ZIO_COMPRESS_MASK(ON) | ZIO_COMPRESS_MASK(OFF) |
ZIO_COMPRESS_MASK(INHERIT) | ZIO_COMPRESS_MASK(EMPTY) |
- (getenv("ZDB_NO_ZLE") ? ZIO_COMPRESS_MASK(ZLE) : 0);
+ ZIO_COMPRESS_MASK(ZLE);
*cfuncp++ = ZIO_COMPRESS_LZ4;
*cfuncp++ = ZIO_COMPRESS_LZJB;
mask |= ZIO_COMPRESS_MASK(LZ4) | ZIO_COMPRESS_MASK(LZJB);
+ /*
+ * Every gzip level has the same decompressor, no need to
+ * run it 9 times per bruteforce attempt.
+ */
+ mask |= ZIO_COMPRESS_MASK(GZIP_2) | ZIO_COMPRESS_MASK(GZIP_3);
+ mask |= ZIO_COMPRESS_MASK(GZIP_4) | ZIO_COMPRESS_MASK(GZIP_5);
+ mask |= ZIO_COMPRESS_MASK(GZIP_6) | ZIO_COMPRESS_MASK(GZIP_7);
+ mask |= ZIO_COMPRESS_MASK(GZIP_8) | ZIO_COMPRESS_MASK(GZIP_9);
for (int c = 0; c < ZIO_COMPRESS_FUNCTIONS; c++)
if (((1ULL << c) & mask) == 0)
*cfuncp++ = c;
lsize += SPA_MINBLOCKSIZE;
else
maxlsize = lsize;
+
for (; lsize <= maxlsize; lsize += SPA_MINBLOCKSIZE) {
for (cfuncp = cfuncs; *cfuncp; cfuncp++) {
- if (flags & ZDB_FLAG_VERBOSE) {
- (void) fprintf(stderr,
- "Trying %05llx -> %05llx (%s)\n",
- (u_longlong_t)psize,
- (u_longlong_t)lsize,
- zio_compress_table[*cfuncp].\
- ci_name);
- }
-
- /*
- * We randomize lbuf2, and decompress to both
- * lbuf and lbuf2. This way, we will know if
- * decompression fill exactly to lsize.
- */
- VERIFY0(random_get_pseudo_bytes(lbuf2, lsize));
-
- if (zio_decompress_data(*cfuncp, pabd,
- lbuf, psize, lsize, NULL) == 0 &&
- zio_decompress_data(*cfuncp, pabd,
- lbuf2, psize, lsize, NULL) == 0 &&
- bcmp(lbuf, lbuf2, lsize) == 0)
+ if (try_decompress_block(pabd, lsize, psize, flags,
+ *cfuncp, lbuf, lbuf2)) {
+ found = B_TRUE;
break;
+ }
}
if (*cfuncp != 0)
break;
}
+ if (!found && tryzle) {
+ for (lsize = orig_lsize; lsize <= maxlsize;
+ lsize += SPA_MINBLOCKSIZE) {
+ if (try_decompress_block(pabd, lsize, psize, flags,
+ ZIO_COMPRESS_ZLE, lbuf, lbuf2)) {
+ *cfuncp = ZIO_COMPRESS_ZLE;
+ found = B_TRUE;
+ break;
+ }
+ }
+ }
umem_free(lbuf2, SPA_MAXBLOCKSIZE);
- if (lsize > maxlsize) {
- exceeded = B_TRUE;
- }
if (*cfuncp == ZIO_COMPRESS_ZLE) {
printf("\nZLE decompression was selected. If you "
"suspect the results are wrong,\ntry avoiding ZLE "
"by setting and exporting ZDB_NO_ZLE=\"true\"\n");
}
- return (exceeded);
+ return (lsize > maxlsize ? -1 : lsize);
}
/*
vdev_t *vd;
abd_t *pabd;
void *lbuf, *buf;
- char *s, *p, *dup, *vdev, *flagstr, *sizes, *tmp = NULL;
+ char *s, *p, *dup, *flagstr, *sizes, *tmp = NULL;
+ const char *vdev, *errmsg = NULL;
int i, error;
boolean_t borrowed = B_FALSE, found = B_FALSE;
dup = strdup(thing);
s = strtok_r(dup, ":", &tmp);
- vdev = s ? s : "";
+ vdev = s ?: "";
s = strtok_r(NULL, ":", &tmp);
offset = strtoull(s ? s : "", NULL, 16);
sizes = strtok_r(NULL, ":", &tmp);
s = strtok_r(NULL, ":", &tmp);
- flagstr = strdup(s ? s : "");
+ flagstr = strdup(s ?: "");
- s = NULL;
- tmp = NULL;
if (!zdb_parse_block_sizes(sizes, &lsize, &psize))
- s = "invalid size(s)";
+ errmsg = "invalid size(s)";
if (!IS_P2ALIGNED(psize, DEV_BSIZE) || !IS_P2ALIGNED(lsize, DEV_BSIZE))
- s = "size must be a multiple of sector size";
+ errmsg = "size must be a multiple of sector size";
if (!IS_P2ALIGNED(offset, DEV_BSIZE))
- s = "offset must be a multiple of sector size";
- if (s) {
- (void) printf("Invalid block specifier: %s - %s\n", thing, s);
+ errmsg = "offset must be a multiple of sector size";
+ if (errmsg) {
+ (void) printf("Invalid block specifier: %s - %s\n",
+ thing, errmsg);
goto done;
}
+ tmp = NULL;
for (s = strtok_r(flagstr, ":", &tmp);
s != NULL;
s = strtok_r(NULL, ":", &tmp)) {
vd = zdb_vdev_lookup(spa->spa_root_vdev, vdev);
if (vd == NULL) {
(void) printf("***Invalid vdev: %s\n", vdev);
- free(dup);
- return;
+ goto done;
} else {
if (vd->vdev_path)
(void) fprintf(stderr, "Found vdev: %s\n",
*/
zio_nowait(zio_vdev_child_io(zio, bp, vd, offset, pabd,
psize, ZIO_TYPE_READ, ZIO_PRIORITY_SYNC_READ,
- ZIO_FLAG_DONT_CACHE | ZIO_FLAG_DONT_PROPAGATE |
- ZIO_FLAG_DONT_RETRY | ZIO_FLAG_CANFAIL | ZIO_FLAG_RAW |
- ZIO_FLAG_OPTIONAL, NULL, NULL));
+ ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_DONT_RETRY |
+ ZIO_FLAG_CANFAIL | ZIO_FLAG_RAW | ZIO_FLAG_OPTIONAL,
+ NULL, NULL));
}
error = zio_wait(zio);
uint64_t orig_lsize = lsize;
buf = lbuf;
if (flags & ZDB_FLAG_DECOMPRESS) {
- boolean_t failed = zdb_decompress_block(pabd, buf, lbuf,
+ lsize = zdb_decompress_block(pabd, buf, lbuf,
lsize, psize, flags);
- if (failed) {
+ if (lsize == -1) {
(void) printf("Decompress of %s failed\n", thing);
goto out;
}
!(flags & ZDB_FLAG_DECOMPRESS)) {
const blkptr_t *b = (const blkptr_t *)(void *)
((uintptr_t)buf + (uintptr_t)blkptr_offset);
- if (zfs_blkptr_verify(spa, b, B_FALSE, BLK_VERIFY_ONLY) ==
- B_FALSE) {
+ if (zfs_blkptr_verify(spa, b,
+ BLK_CONFIG_NEEDED, BLK_VERIFY_ONLY) == B_FALSE) {
abd_return_buf_copy(pabd, buf, lsize);
borrowed = B_FALSE;
buf = lbuf;
- boolean_t failed = zdb_decompress_block(pabd, buf,
+ lsize = zdb_decompress_block(pabd, buf,
lbuf, lsize, psize, flags);
b = (const blkptr_t *)(void *)
((uintptr_t)buf + (uintptr_t)blkptr_offset);
- if (failed || zfs_blkptr_verify(spa, b, B_FALSE,
- BLK_VERIFY_LOG) == B_FALSE) {
+ if (lsize == -1 || zfs_blkptr_verify(spa, b,
+ BLK_CONFIG_NEEDED, BLK_VERIFY_LOG) == B_FALSE) {
printf("invalid block pointer at this DVA\n");
goto out;
}
BP_SET_CHECKSUM(bp, ck);
spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
czio = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL);
- czio->io_bp = bp;
-
if (vd == vd->vdev_top) {
zio_nowait(zio_read(czio, spa, bp, pabd, psize,
NULL, NULL,
zio_nowait(zio_vdev_child_io(czio, bp, vd,
offset, pabd, psize, ZIO_TYPE_READ,
ZIO_PRIORITY_SYNC_READ,
- ZIO_FLAG_DONT_CACHE |
ZIO_FLAG_DONT_PROPAGATE |
ZIO_FLAG_DONT_RETRY |
ZIO_FLAG_CANFAIL | ZIO_FLAG_RAW |
}
error = zio_wait(czio);
if (error == 0 || error == ECKSUM) {
- zio_t *ck_zio = zio_root(spa, NULL, NULL, 0);
+ zio_t *ck_zio = zio_null(NULL, spa, NULL,
+ NULL, NULL, 0);
ck_zio->io_offset =
DVA_GET_OFFSET(&bp->blk_dva[0]);
ck_zio->io_bp = bp;
zio_checksum_compute(ck_zio, ck, pabd, lsize);
- printf("%12s\tcksum=%llx:%llx:%llx:%llx\n",
+ printf(
+ "%12s\t"
+ "cksum=%016llx:%016llx:%016llx:%016llx\n",
zio_checksum_table[ck].ci_name,
(u_longlong_t)bp->blk_cksum.zc_word[0],
(u_longlong_t)bp->blk_cksum.zc_word[1],
static void
zdb_embedded_block(char *thing)
{
- blkptr_t bp;
+ blkptr_t bp = {{{{0}}}};
unsigned long long *words = (void *)&bp;
char *buf;
int err;
- bzero(&bp, sizeof (bp));
err = sscanf(thing, "%llx:%llx:%llx:%llx:%llx:%llx:%llx:%llx:"
"%llx:%llx:%llx:%llx:%llx:%llx:%llx:%llx",
words + 0, words + 1, words + 2, words + 3,
free(buf);
}
+/* check for valid hex or decimal numeric string */
+static boolean_t
+zdb_numeric(char *str)
+{
+ int i = 0;
+
+ if (strlen(str) == 0)
+ return (B_FALSE);
+ if (strncmp(str, "0x", 2) == 0 || strncmp(str, "0X", 2) == 0)
+ i = 2;
+ for (; i < strlen(str); i++) {
+ if (!isxdigit(str[i]))
+ return (B_FALSE);
+ }
+ return (B_TRUE);
+}
+
int
main(int argc, char **argv)
{
int c;
- struct rlimit rl = { 1024, 1024 };
spa_t *spa = NULL;
objset_t *os = NULL;
int dump_all = 1;
boolean_t target_is_spa = B_TRUE, dataset_lookup = B_FALSE;
nvlist_t *cfg = NULL;
- (void) setrlimit(RLIMIT_NOFILE, &rl);
- (void) enable_extended_FILE_stdio(-1, -1);
-
dprintf_setup(&argc, argv);
/*
*/
zfs_btree_verify_intensity = 3;
- while ((c = getopt(argc, argv,
- "AbcCdDeEFGhiI:klLmMo:Op:PqrRsSt:uU:vVx:XYyZ")) != -1) {
+ struct option long_options[] = {
+ {"ignore-assertions", no_argument, NULL, 'A'},
+ {"block-stats", no_argument, NULL, 'b'},
+ {"backup", no_argument, NULL, 'B'},
+ {"checksum", no_argument, NULL, 'c'},
+ {"config", no_argument, NULL, 'C'},
+ {"datasets", no_argument, NULL, 'd'},
+ {"dedup-stats", no_argument, NULL, 'D'},
+ {"exported", no_argument, NULL, 'e'},
+ {"embedded-block-pointer", no_argument, NULL, 'E'},
+ {"automatic-rewind", no_argument, NULL, 'F'},
+ {"dump-debug-msg", no_argument, NULL, 'G'},
+ {"history", no_argument, NULL, 'h'},
+ {"intent-logs", no_argument, NULL, 'i'},
+ {"inflight", required_argument, NULL, 'I'},
+ {"checkpointed-state", no_argument, NULL, 'k'},
+ {"key", required_argument, NULL, 'K'},
+ {"label", no_argument, NULL, 'l'},
+ {"disable-leak-tracking", no_argument, NULL, 'L'},
+ {"metaslabs", no_argument, NULL, 'm'},
+ {"metaslab-groups", no_argument, NULL, 'M'},
+ {"numeric", no_argument, NULL, 'N'},
+ {"option", required_argument, NULL, 'o'},
+ {"object-lookups", no_argument, NULL, 'O'},
+ {"path", required_argument, NULL, 'p'},
+ {"parseable", no_argument, NULL, 'P'},
+ {"skip-label", no_argument, NULL, 'q'},
+ {"copy-object", no_argument, NULL, 'r'},
+ {"read-block", no_argument, NULL, 'R'},
+ {"io-stats", no_argument, NULL, 's'},
+ {"simulate-dedup", no_argument, NULL, 'S'},
+ {"txg", required_argument, NULL, 't'},
+ {"brt-stats", no_argument, NULL, 'T'},
+ {"uberblock", no_argument, NULL, 'u'},
+ {"cachefile", required_argument, NULL, 'U'},
+ {"verbose", no_argument, NULL, 'v'},
+ {"verbatim", no_argument, NULL, 'V'},
+ {"dump-blocks", required_argument, NULL, 'x'},
+ {"extreme-rewind", no_argument, NULL, 'X'},
+ {"all-reconstruction", no_argument, NULL, 'Y'},
+ {"livelist", no_argument, NULL, 'y'},
+ {"zstd-headers", no_argument, NULL, 'Z'},
+ {0, 0, 0, 0}
+ };
+
+ while ((c = getopt_long(argc, argv,
+ "AbBcCdDeEFGhiI:kK:lLmMNo:Op:PqrRsSt:TuU:vVx:XYyZ",
+ long_options, NULL)) != -1) {
switch (c) {
case 'b':
+ case 'B':
case 'c':
case 'C':
case 'd':
case 'l':
case 'm':
case 'M':
+ case 'N':
case 'O':
case 'r':
case 'R':
case 's':
case 'S':
+ case 'T':
case 'u':
case 'y':
case 'Z':
usage();
}
break;
+ case 'K':
+ dump_opt[c]++;
+ key_material = strdup(optarg);
+ /* redact key material in process table */
+ while (*optarg != '\0') { *optarg++ = '*'; }
+ break;
case 'o':
error = set_global_var(optarg);
if (error != 0)
} else {
char **tmp = umem_alloc((nsearch + 1) *
sizeof (char *), UMEM_NOFAIL);
- bcopy(searchdirs, tmp, nsearch *
+ memcpy(tmp, searchdirs, nsearch *
sizeof (char *));
umem_free(searchdirs,
nsearch * sizeof (char *));
(void) fprintf(stderr, "-p option requires use of -e\n");
usage();
}
- if (dump_opt['d'] || dump_opt['r']) {
- /* <pool>[/<dataset | objset id> is accepted */
- if (argv[2] && (objset_str = strchr(argv[2], '/')) != NULL &&
- objset_str++ != NULL) {
- char *endptr;
- errno = 0;
- objset_id = strtoull(objset_str, &endptr, 0);
- /* dataset 0 is the same as opening the pool */
- if (errno == 0 && endptr != objset_str &&
- objset_id != 0) {
- target_is_spa = B_FALSE;
- dataset_lookup = B_TRUE;
- } else if (objset_id != 0) {
- printf("failed to open objset %s "
- "%llu %s", objset_str,
- (u_longlong_t)objset_id,
- strerror(errno));
- exit(1);
- }
- /* normal dataset name not an objset ID */
- if (endptr == objset_str) {
- objset_id = -1;
- }
- }
- }
-
#if defined(_LP64)
/*
* ZDB does not typically re-read blocks; therefore limit the ARC
* to 256 MB, which can be used entirely for metadata.
*/
- zfs_arc_min = zfs_arc_meta_min = 2ULL << SPA_MAXBLOCKSHIFT;
- zfs_arc_max = zfs_arc_meta_limit = 256 * 1024 * 1024;
+ zfs_arc_min = 2ULL << SPA_MAXBLOCKSHIFT;
+ zfs_arc_max = 256 * 1024 * 1024;
#endif
/*
*/
spa_load_verify_dryrun = B_TRUE;
+ /*
+ * ZDB should have ability to read spacemaps.
+ */
+ spa_mode_readable_spacemaps = B_TRUE;
+
kernel_init(SPA_MODE_READ);
if (dump_all)
verbose = MAX(verbose, 1);
for (c = 0; c < 256; c++) {
- if (dump_all && strchr("AeEFklLOPrRSXy", c) == NULL)
+ if (dump_all && strchr("ABeEFkKlLNOPrRSXy", c) == NULL)
dump_opt[c] = 1;
if (dump_opt[c])
dump_opt[c] += verbose;
}
- libspl_assert_ok = (dump_opt['A'] == 1) || (dump_opt['A'] > 2);
+ libspl_set_assert_ok((dump_opt['A'] == 1) || (dump_opt['A'] > 2));
zfs_recover = (dump_opt['A'] > 1);
argc -= optind;
if (dump_opt['l'])
return (dump_label(argv[0]));
- if (dump_opt['O']) {
- if (argc != 2)
- usage();
- dump_opt['v'] = verbose + 3;
- return (dump_path(argv[0], argv[1], NULL));
- }
- if (dump_opt['r']) {
- if (argc != 3)
- usage();
- dump_opt['v'] = verbose;
- error = dump_path(argv[0], argv[1], &object);
- }
-
if (dump_opt['X'] || dump_opt['F'])
rewind = ZPOOL_DO_REWIND |
(dump_opt['X'] ? ZPOOL_EXTREME_REWIND : 0);
+ /* -N implies -d */
+ if (dump_opt['N'] && dump_opt['d'] == 0)
+ dump_opt['d'] = dump_opt['N'];
+
if (nvlist_alloc(&policy, NV_UNIQUE_NAME_TYPE, 0) != 0 ||
nvlist_add_uint64(policy, ZPOOL_LOAD_REQUEST_TXG, max_txg) != 0 ||
nvlist_add_uint32(policy, ZPOOL_LOAD_REWIND_POLICY, rewind) != 0)
targetlen = strlen(target);
if (targetlen && target[targetlen - 1] == '/')
target[targetlen - 1] = '\0';
+
+ /*
+ * See if an objset ID was supplied (-d <pool>/<objset ID>).
+ * To disambiguate tank/100, consider the 100 as objsetID
+ * if -N was given, otherwise 100 is an objsetID iff
+ * tank/100 as a named dataset fails on lookup.
+ */
+ objset_str = strchr(target, '/');
+ if (objset_str && strlen(objset_str) > 1 &&
+ zdb_numeric(objset_str + 1)) {
+ char *endptr;
+ errno = 0;
+ objset_str++;
+ objset_id = strtoull(objset_str, &endptr, 0);
+ /* dataset 0 is the same as opening the pool */
+ if (errno == 0 && endptr != objset_str &&
+ objset_id != 0) {
+ if (dump_opt['N'])
+ dataset_lookup = B_TRUE;
+ }
+ /* normal dataset name not an objset ID */
+ if (endptr == objset_str) {
+ objset_id = -1;
+ }
+ } else if (objset_str && !zdb_numeric(objset_str + 1) &&
+ dump_opt['N']) {
+ printf("Supply a numeric objset ID with -N\n");
+ exit(1);
+ }
} else {
target_pool = target;
}
args.path = searchdirs;
args.can_be_active = B_TRUE;
- error = zpool_find_config(NULL, target_pool, &cfg, &args,
- &libzpool_config_ops);
+ libpc_handle_t lpch = {
+ .lpc_lib_handle = NULL,
+ .lpc_ops = &libzpool_config_ops,
+ .lpc_printerr = B_TRUE
+ };
+ error = zpool_find_config(&lpch, target_pool, &cfg, &args);
if (error == 0) {
searchdirs = NULL;
}
+ /*
+ * We need to make sure to process -O option or call
+ * dump_path after the -e option has been processed,
+ * which imports the pool to the namespace if it's
+ * not in the cachefile.
+ */
+ if (dump_opt['O']) {
+ if (argc != 2)
+ usage();
+ dump_opt['v'] = verbose + 3;
+ return (dump_path(argv[0], argv[1], NULL));
+ }
+
+ if (dump_opt['r']) {
+ target_is_spa = B_FALSE;
+ if (argc != 3)
+ usage();
+ dump_opt['v'] = verbose;
+ error = dump_path(argv[0], argv[1], &object);
+ if (error != 0)
+ fatal("internal error: %s", strerror(error));
+ }
+
/*
* import_checkpointed_state makes the assumption that the
* target pool that we pass it is already part of the spa
checkpoint_pool, error);
}
- } else if (target_is_spa || dump_opt['R'] || objset_id == 0) {
+ } else if (target_is_spa || dump_opt['R'] || dump_opt['B'] ||
+ objset_id == 0) {
zdb_set_skip_mmp(target);
error = spa_open_rewind(target, &spa, FTAG, policy,
NULL);
}
return (error);
} else {
+ target_pool = strdup(target);
+ if (strpbrk(target, "/@") != NULL)
+ *strpbrk(target_pool, "/@") = '\0';
+
zdb_set_skip_mmp(target);
+ /*
+ * If -N was supplied, the user has indicated that
+ * zdb -d <pool>/<objsetID> is in effect. Otherwise
+ * we first assume that the dataset string is the
+ * dataset name. If dmu_objset_hold fails with the
+ * dataset string, and we have an objset_id, retry the
+ * lookup with the objsetID.
+ */
+ boolean_t retry = B_TRUE;
+retry_lookup:
if (dataset_lookup == B_TRUE) {
/*
* Use the supplied id to get the name
* for open_objset.
*/
- error = spa_open(target, &spa, FTAG);
+ error = spa_open(target_pool, &spa, FTAG);
if (error == 0) {
error = name_from_objset_id(spa,
objset_id, dsname);
target = dsname;
}
}
- if (error == 0)
+ if (error == 0) {
+ if (objset_id > 0 && retry) {
+ int err = dmu_objset_hold(target, FTAG,
+ &os);
+ if (err) {
+ dataset_lookup = B_TRUE;
+ retry = B_FALSE;
+ goto retry_lookup;
+ } else {
+ dmu_objset_rele(os, FTAG);
+ }
+ }
error = open_objset(target, FTAG, &os);
+ }
if (error == 0)
spa = dmu_objset_spa(os);
+ free(target_pool);
}
}
nvlist_free(policy);
sizeof (zopt_object_range_t));
for (unsigned i = 0; i < zopt_object_args; i++) {
int err;
- char *msg = NULL;
+ const char *msg = NULL;
err = parse_object_range(argv[i],
&zopt_object_ranges[i], &msg);
if (err != 0)
fatal("Bad object or range: '%s': %s\n",
- argv[i], msg ? msg : "");
+ argv[i], msg ?: "");
}
} else if (argc > 0 && dump_opt['m']) {
zopt_metaslab_args = argc;
strerror(errno));
}
}
- if (os != NULL) {
+ if (dump_opt['B']) {
+ dump_backup(target, objset_id,
+ argc > 0 ? argv[0] : NULL);
+ } else if (os != NULL) {
dump_objset(os);
} else if (zopt_object_args > 0 && !dump_opt['m']) {
dump_objset(spa->spa_meta_objset);
projects / mirror_zfs.git / blobdiff