/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012 by Delphix. All rights reserved.
+ * Copyright (c) 2013 by Delphix. All rights reserved.
*/
#include <stdio.h>
+#include <unistd.h>
#include <stdio_ext.h>
#include <stdlib.h>
#include <ctype.h>
#include <sys/arc.h>
#include <sys/ddt.h>
#include <sys/zfeature.h>
+#include <zfs_comutil.h>
#undef ZFS_MAXNAMELEN
#include <libzfs.h>
(void) fprintf(stderr, " -t <txg> -- highest txg to use when "
"searching for uberblocks\n");
(void) fprintf(stderr, " -M <number of inflight I/Os> -- "
- "specify the maximum number of checksumming I/Os [default is 200]");
+ "specify the maximum number of checksumming I/Os "
+ "[default is 200]\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");
nvlist_free(nv);
}
+/* ARGSUSED */
+static void
+dump_history_offsets(objset_t *os, uint64_t object, void *data, size_t size)
+{
+ spa_history_phys_t *shp = data;
+
+ if (shp == NULL)
+ return;
+
+ (void) printf("\t\tpool_create_len = %llu\n",
+ (u_longlong_t)shp->sh_pool_create_len);
+ (void) printf("\t\tphys_max_off = %llu\n",
+ (u_longlong_t)shp->sh_phys_max_off);
+ (void) printf("\t\tbof = %llu\n",
+ (u_longlong_t)shp->sh_bof);
+ (void) printf("\t\teof = %llu\n",
+ (u_longlong_t)shp->sh_eof);
+ (void) printf("\t\trecords_lost = %llu\n",
+ (u_longlong_t)shp->sh_records_lost);
+}
+
static void
zdb_nicenum(uint64_t num, char *buf)
{
nicenum(num, buf);
}
-const char dump_zap_stars[] = "****************************************";
-const int dump_zap_width = sizeof (dump_zap_stars) - 1;
+const char histo_stars[] = "****************************************";
+const int histo_width = sizeof (histo_stars) - 1;
static void
-dump_zap_histogram(uint64_t histo[ZAP_HISTOGRAM_SIZE])
+dump_histogram(const uint64_t *histo, int size, int offset)
{
int i;
- int minidx = ZAP_HISTOGRAM_SIZE - 1;
+ int minidx = size - 1;
int maxidx = 0;
uint64_t max = 0;
- for (i = 0; i < ZAP_HISTOGRAM_SIZE; i++) {
+ for (i = 0; i < size; i++) {
if (histo[i] > max)
max = histo[i];
if (histo[i] > 0 && i > maxidx)
minidx = i;
}
- if (max < dump_zap_width)
- max = dump_zap_width;
+ if (max < histo_width)
+ max = histo_width;
- for (i = minidx; i <= maxidx; i++)
- (void) printf("\t\t\t%u: %6llu %s\n", i, (u_longlong_t)histo[i],
- &dump_zap_stars[(max - histo[i]) * dump_zap_width / max]);
+ for (i = minidx; i <= maxidx; i++) {
+ (void) printf("\t\t\t%3u: %6llu %s\n",
+ i + offset, (u_longlong_t)histo[i],
+ &histo_stars[(max - histo[i]) * histo_width / max]);
+ }
}
static void
(u_longlong_t)zs.zs_salt);
(void) printf("\t\tLeafs with 2^n pointers:\n");
- dump_zap_histogram(zs.zs_leafs_with_2n_pointers);
+ dump_histogram(zs.zs_leafs_with_2n_pointers, ZAP_HISTOGRAM_SIZE, 0);
(void) printf("\t\tBlocks with n*5 entries:\n");
- dump_zap_histogram(zs.zs_blocks_with_n5_entries);
+ dump_histogram(zs.zs_blocks_with_n5_entries, ZAP_HISTOGRAM_SIZE, 0);
(void) printf("\t\tBlocks n/10 full:\n");
- dump_zap_histogram(zs.zs_blocks_n_tenths_full);
+ dump_histogram(zs.zs_blocks_n_tenths_full, ZAP_HISTOGRAM_SIZE, 0);
(void) printf("\t\tEntries with n chunks:\n");
- dump_zap_histogram(zs.zs_entries_using_n_chunks);
+ dump_histogram(zs.zs_entries_using_n_chunks, ZAP_HISTOGRAM_SIZE, 0);
(void) printf("\t\tBuckets with n entries:\n");
- dump_zap_histogram(zs.zs_buckets_with_n_entries);
+ dump_histogram(zs.zs_buckets_with_n_entries, ZAP_HISTOGRAM_SIZE, 0);
}
/*ARGSUSED*/
zap_cursor_fini(&zc);
}
+int
+get_dtl_refcount(vdev_t *vd)
+{
+ int refcount = 0;
+ int c;
+
+ if (vd->vdev_ops->vdev_op_leaf) {
+ space_map_t *sm = vd->vdev_dtl_sm;
+
+ if (sm != NULL &&
+ sm->sm_dbuf->db_size == sizeof (space_map_phys_t))
+ return (1);
+ return (0);
+ }
+
+ for (c = 0; c < vd->vdev_children; c++)
+ refcount += get_dtl_refcount(vd->vdev_child[c]);
+ return (refcount);
+}
+
+int
+get_metaslab_refcount(vdev_t *vd)
+{
+ int refcount = 0;
+ int c, m;
+
+ if (vd->vdev_top == vd) {
+ for (m = 0; m < vd->vdev_ms_count; m++) {
+ space_map_t *sm = vd->vdev_ms[m]->ms_sm;
+
+ if (sm != NULL &&
+ sm->sm_dbuf->db_size == sizeof (space_map_phys_t))
+ refcount++;
+ }
+ }
+ for (c = 0; c < vd->vdev_children; c++)
+ refcount += get_metaslab_refcount(vd->vdev_child[c]);
+
+ return (refcount);
+}
+
+static int
+verify_spacemap_refcounts(spa_t *spa)
+{
+ uint64_t expected_refcount = 0;
+ uint64_t actual_refcount;
+
+ (void) feature_get_refcount(spa,
+ &spa_feature_table[SPA_FEATURE_SPACEMAP_HISTOGRAM],
+ &expected_refcount);
+ actual_refcount = get_dtl_refcount(spa->spa_root_vdev);
+ actual_refcount += get_metaslab_refcount(spa->spa_root_vdev);
+
+ if (expected_refcount != actual_refcount) {
+ (void) printf("space map refcount mismatch: expected %lld != "
+ "actual %lld\n",
+ (longlong_t)expected_refcount,
+ (longlong_t)actual_refcount);
+ return (2);
+ }
+ return (0);
+}
+
static void
-dump_spacemap(objset_t *os, space_map_obj_t *smo, space_map_t *sm)
+dump_spacemap(objset_t *os, space_map_t *sm)
{
uint64_t alloc, offset, entry;
- uint8_t mapshift = sm->sm_shift;
- uint64_t mapstart = sm->sm_start;
char *ddata[] = { "ALLOC", "FREE", "CONDENSE", "INVALID",
"INVALID", "INVALID", "INVALID", "INVALID" };
- if (smo->smo_object == 0)
+ if (sm == NULL)
return;
/*
* Print out the freelist entries in both encoded and decoded form.
*/
alloc = 0;
- for (offset = 0; offset < smo->smo_objsize; offset += sizeof (entry)) {
- VERIFY3U(0, ==, dmu_read(os, smo->smo_object, offset,
+ for (offset = 0; offset < space_map_length(sm);
+ offset += sizeof (entry)) {
+ uint8_t mapshift = sm->sm_shift;
+
+ VERIFY0(dmu_read(os, space_map_object(sm), offset,
sizeof (entry), &entry, DMU_READ_PREFETCH));
if (SM_DEBUG_DECODE(entry)) {
+
(void) printf("\t [%6llu] %s: txg %llu, pass %llu\n",
(u_longlong_t)(offset / sizeof (entry)),
ddata[SM_DEBUG_ACTION_DECODE(entry)],
(u_longlong_t)(offset / sizeof (entry)),
SM_TYPE_DECODE(entry) == SM_ALLOC ? 'A' : 'F',
(u_longlong_t)((SM_OFFSET_DECODE(entry) <<
- mapshift) + mapstart),
+ mapshift) + sm->sm_start),
(u_longlong_t)((SM_OFFSET_DECODE(entry) <<
- mapshift) + mapstart + (SM_RUN_DECODE(entry) <<
- mapshift)),
+ mapshift) + sm->sm_start +
+ (SM_RUN_DECODE(entry) << mapshift)),
(u_longlong_t)(SM_RUN_DECODE(entry) << mapshift));
if (SM_TYPE_DECODE(entry) == SM_ALLOC)
alloc += SM_RUN_DECODE(entry) << mapshift;
alloc -= SM_RUN_DECODE(entry) << mapshift;
}
}
- if (alloc != smo->smo_alloc) {
+ if (alloc != space_map_allocated(sm)) {
(void) printf("space_map_object alloc (%llu) INCONSISTENT "
"with space map summary (%llu)\n",
- (u_longlong_t)smo->smo_alloc, (u_longlong_t)alloc);
+ (u_longlong_t)space_map_allocated(sm), (u_longlong_t)alloc);
}
}
dump_metaslab_stats(metaslab_t *msp)
{
char maxbuf[32];
- space_map_t *sm = msp->ms_map;
- avl_tree_t *t = sm->sm_pp_root;
- int free_pct = sm->sm_space * 100 / sm->sm_size;
+ range_tree_t *rt = msp->ms_tree;
+ avl_tree_t *t = &msp->ms_size_tree;
+ int free_pct = range_tree_space(rt) * 100 / msp->ms_size;
- zdb_nicenum(space_map_maxsize(sm), maxbuf);
+ zdb_nicenum(metaslab_block_maxsize(msp), maxbuf);
(void) printf("\t %25s %10lu %7s %6s %4s %4d%%\n",
"segments", avl_numnodes(t), "maxsize", maxbuf,
"freepct", free_pct);
+ (void) printf("\tIn-memory histogram:\n");
+ dump_histogram(rt->rt_histogram, RANGE_TREE_HISTOGRAM_SIZE, 0);
}
static void
{
vdev_t *vd = msp->ms_group->mg_vd;
spa_t *spa = vd->vdev_spa;
- space_map_t *sm = msp->ms_map;
- space_map_obj_t *smo = &msp->ms_smo;
+ space_map_t *sm = msp->ms_sm;
char freebuf[32];
- zdb_nicenum(sm->sm_size - smo->smo_alloc, freebuf);
+ zdb_nicenum(msp->ms_size - space_map_allocated(sm), freebuf);
(void) printf(
"\tmetaslab %6llu offset %12llx spacemap %6llu free %5s\n",
- (u_longlong_t)(sm->sm_start / sm->sm_size),
- (u_longlong_t)sm->sm_start, (u_longlong_t)smo->smo_object, freebuf);
+ (u_longlong_t)msp->ms_id, (u_longlong_t)msp->ms_start,
+ (u_longlong_t)space_map_object(sm), freebuf);
- if (dump_opt['m'] > 1 && !dump_opt['L']) {
+ if (dump_opt['m'] > 2 && !dump_opt['L']) {
mutex_enter(&msp->ms_lock);
- space_map_load_wait(sm);
- if (!sm->sm_loaded)
- VERIFY(space_map_load(sm, zfs_metaslab_ops,
- SM_FREE, smo, spa->spa_meta_objset) == 0);
+ metaslab_load_wait(msp);
+ if (!msp->ms_loaded) {
+ VERIFY0(metaslab_load(msp));
+ range_tree_stat_verify(msp->ms_tree);
+ }
dump_metaslab_stats(msp);
- space_map_unload(sm);
+ metaslab_unload(msp);
mutex_exit(&msp->ms_lock);
}
- if (dump_opt['d'] > 5 || dump_opt['m'] > 2) {
- ASSERT(sm->sm_size == (1ULL << vd->vdev_ms_shift));
+ if (dump_opt['m'] > 1 && sm != NULL &&
+ spa_feature_is_active(spa, SPA_FEATURE_SPACEMAP_HISTOGRAM)) {
+ /*
+ * The space map histogram represents free space in chunks
+ * of sm_shift (i.e. bucket 0 refers to 2^sm_shift).
+ */
+ (void) printf("\tOn-disk histogram:\n");
+ dump_histogram(sm->sm_phys->smp_histogram,
+ SPACE_MAP_HISTOGRAM_SIZE(sm), sm->sm_shift);
+ }
+
+ if (dump_opt['d'] > 5 || dump_opt['m'] > 3) {
+ ASSERT(msp->ms_size == (1ULL << vd->vdev_ms_shift));
mutex_enter(&msp->ms_lock);
- dump_spacemap(spa->spa_meta_objset, smo, sm);
+ dump_spacemap(spa->spa_meta_objset, msp->ms_sm);
mutex_exit(&msp->ms_lock);
}
}
}
static void
-dump_dtl_seg(space_map_t *sm, uint64_t start, uint64_t size)
+dump_dtl_seg(void *arg, uint64_t start, uint64_t size)
{
- char *prefix = (void *)sm;
+ char *prefix = arg;
(void) printf("%s [%llu,%llu) length %llu\n",
prefix,
required ? "DTL-required" : "DTL-expendable");
for (t = 0; t < DTL_TYPES; t++) {
- space_map_t *sm = &vd->vdev_dtl[t];
- if (sm->sm_space == 0)
+ range_tree_t *rt = vd->vdev_dtl[t];
+ if (range_tree_space(rt) == 0)
continue;
(void) snprintf(prefix, sizeof (prefix), "\t%*s%s",
indent + 2, "", name[t]);
- mutex_enter(sm->sm_lock);
- space_map_walk(sm, dump_dtl_seg, (void *)prefix);
- mutex_exit(sm->sm_lock);
+ mutex_enter(rt->rt_lock);
+ range_tree_walk(rt, dump_dtl_seg, prefix);
+ mutex_exit(rt->rt_lock);
if (dump_opt['d'] > 5 && vd->vdev_children == 0)
dump_spacemap(spa->spa_meta_objset,
- &vd->vdev_dtl_smo, sm);
+ vd->vdev_dtl_sm);
}
for (c = 0; c < vd->vdev_children; c++)
for (i = 0; i < num; i++) {
uint64_t time, txg, ievent;
char *cmd, *intstr;
+ boolean_t printed = B_FALSE;
if (nvlist_lookup_uint64(events[i], ZPOOL_HIST_TIME,
&time) != 0)
- continue;
+ goto next;
if (nvlist_lookup_string(events[i], ZPOOL_HIST_CMD,
&cmd) != 0) {
if (nvlist_lookup_uint64(events[i],
ZPOOL_HIST_INT_EVENT, &ievent) != 0)
- continue;
+ goto next;
verify(nvlist_lookup_uint64(events[i],
ZPOOL_HIST_TXG, &txg) == 0);
verify(nvlist_lookup_string(events[i],
ZPOOL_HIST_INT_STR, &intstr) == 0);
- if (ievent >= LOG_END)
- continue;
+ if (ievent >= ZFS_NUM_LEGACY_HISTORY_EVENTS)
+ goto next;
(void) snprintf(internalstr,
sizeof (internalstr),
(void) localtime_r(&tsec, &t);
(void) strftime(tbuf, sizeof (tbuf), "%F.%T", &t);
(void) printf("%s %s\n", tbuf, cmd);
+ printed = B_TRUE;
+
+next:
+ if (dump_opt['h'] > 1) {
+ if (!printed)
+ (void) printf("unrecognized record:\n");
+ dump_nvlist(events[i], 2);
+ }
}
}
int ndvas = dump_opt['d'] > 5 ? BP_GET_NDVAS(bp) : 1;
int i;
- if (dump_opt['b'] >= 5) {
+ if (dump_opt['b'] >= 6) {
sprintf_blkptr(blkbuf, bp);
return;
}
arc_buf_t *buf;
uint64_t fill = 0;
- err = arc_read_nolock(NULL, spa, bp, arc_getbuf_func, &buf,
+ err = arc_read(NULL, spa, bp, arc_getbuf_func, &buf,
ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb);
if (err)
return (err);
}
static void
-dump_bpobj(bpobj_t *bpo, char *name)
+dump_bpobj(bpobj_t *bpo, char *name, int indent)
{
char bytes[32];
char comp[32];
char uncomp[32];
+ uint64_t i;
if (dump_opt['d'] < 3)
return;
zdb_nicenum(bpo->bpo_phys->bpo_bytes, bytes);
- if (bpo->bpo_havesubobj) {
+ if (bpo->bpo_havesubobj && bpo->bpo_phys->bpo_subobjs != 0) {
zdb_nicenum(bpo->bpo_phys->bpo_comp, comp);
zdb_nicenum(bpo->bpo_phys->bpo_uncomp, uncomp);
- (void) printf("\n %s: %llu local blkptrs, %llu subobjs, "
- "%s (%s/%s comp)\n",
- name, (u_longlong_t)bpo->bpo_phys->bpo_num_blkptrs,
+ (void) printf(" %*s: object %llu, %llu local blkptrs, "
+ "%llu subobjs, %s (%s/%s comp)\n",
+ indent * 8, name,
+ (u_longlong_t)bpo->bpo_object,
+ (u_longlong_t)bpo->bpo_phys->bpo_num_blkptrs,
(u_longlong_t)bpo->bpo_phys->bpo_num_subobjs,
bytes, comp, uncomp);
+
+ for (i = 0; i < bpo->bpo_phys->bpo_num_subobjs; i++) {
+ uint64_t subobj;
+ bpobj_t subbpo;
+ int error;
+ VERIFY0(dmu_read(bpo->bpo_os,
+ bpo->bpo_phys->bpo_subobjs,
+ i * sizeof (subobj), sizeof (subobj), &subobj, 0));
+ error = bpobj_open(&subbpo, bpo->bpo_os, subobj);
+ if (error != 0) {
+ (void) printf("ERROR %u while trying to open "
+ "subobj id %llu\n",
+ error, (u_longlong_t)subobj);
+ continue;
+ }
+ dump_bpobj(&subbpo, "subobj", indent + 1);
+ }
} else {
- (void) printf("\n %s: %llu blkptrs, %s\n",
- name, (u_longlong_t)bpo->bpo_phys->bpo_num_blkptrs, bytes);
+ (void) printf(" %*s: object %llu, %llu blkptrs, %s\n",
+ indent * 8, name,
+ (u_longlong_t)bpo->bpo_object,
+ (u_longlong_t)bpo->bpo_phys->bpo_num_blkptrs,
+ bytes);
}
if (dump_opt['d'] < 5)
return;
- (void) printf("\n");
- (void) bpobj_iterate_nofree(bpo, dump_bpobj_cb, NULL, NULL);
+ if (indent == 0) {
+ (void) bpobj_iterate_nofree(bpo, dump_bpobj_cb, NULL, NULL);
+ (void) printf("\n");
+ }
}
static void
dump_deadlist(dsl_deadlist_t *dl)
{
dsl_deadlist_entry_t *dle;
+ uint64_t unused;
char bytes[32];
char comp[32];
char uncomp[32];
(void) printf("\n");
+ /* force the tree to be loaded */
+ dsl_deadlist_space_range(dl, 0, UINT64_MAX, &unused, &unused, &unused);
+
for (dle = avl_first(&dl->dl_tree); dle;
dle = AVL_NEXT(&dl->dl_tree, dle)) {
- (void) printf(" mintxg %llu -> obj %llu\n",
- (longlong_t)dle->dle_mintxg,
- (longlong_t)dle->dle_bpobj.bpo_object);
+ if (dump_opt['d'] >= 5) {
+ char buf[128];
+ (void) snprintf(buf, sizeof (buf),
+ "mintxg %llu -> obj %llu",
+ (longlong_t)dle->dle_mintxg,
+ (longlong_t)dle->dle_bpobj.bpo_object);
+
+ dump_bpobj(&dle->dle_bpobj, buf, 0);
+ } else {
+ (void) printf("mintxg %llu -> obj %llu\n",
+ (longlong_t)dle->dle_mintxg,
+ (longlong_t)dle->dle_bpobj.bpo_object);
- if (dump_opt['d'] >= 5)
- dump_bpobj(&dle->dle_bpobj, "");
+ }
}
}
* print uid or gid information.
* For normal POSIX id just the id is printed in decimal format.
* For CIFS files with FUID the fuid is printed in hex followed by
- * the doman-rid string.
+ * the domain-rid string.
*/
static void
print_idstr(uint64_t id, const char *id_type)
print_idstr(gid, "gid");
}
+static void
+dump_znode_sa_xattr(sa_handle_t *hdl)
+{
+ nvlist_t *sa_xattr;
+ nvpair_t *elem = NULL;
+ int sa_xattr_size = 0;
+ int sa_xattr_entries = 0;
+ int error;
+ char *sa_xattr_packed;
+
+ error = sa_size(hdl, sa_attr_table[ZPL_DXATTR], &sa_xattr_size);
+ if (error || sa_xattr_size == 0)
+ return;
+
+ sa_xattr_packed = malloc(sa_xattr_size);
+ if (sa_xattr_packed == NULL)
+ return;
+
+ error = sa_lookup(hdl, sa_attr_table[ZPL_DXATTR],
+ sa_xattr_packed, sa_xattr_size);
+ if (error) {
+ free(sa_xattr_packed);
+ return;
+ }
+
+ error = nvlist_unpack(sa_xattr_packed, sa_xattr_size, &sa_xattr, 0);
+ if (error) {
+ free(sa_xattr_packed);
+ return;
+ }
+
+ while ((elem = nvlist_next_nvpair(sa_xattr, elem)) != NULL)
+ sa_xattr_entries++;
+
+ (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) {
+ 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]))
+ (void) putchar(value[idx]);
+ else
+ (void) printf("\\%3.3o", value[idx]);
+ }
+ (void) putchar('\n');
+ }
+
+ nvlist_free(sa_xattr);
+ free(sa_xattr_packed);
+}
+
/*ARGSUSED*/
static void
dump_znode(objset_t *os, uint64_t object, void *data, size_t size)
if (sa_lookup(hdl, sa_attr_table[ZPL_RDEV], &rdev,
sizeof (uint64_t)) == 0)
(void) printf("\trdev 0x%016llx\n", (u_longlong_t)rdev);
+ dump_znode_sa_xattr(hdl);
sa_handle_destroy(hdl);
}
dump_zap, /* other ZAP */
dump_zap, /* persistent error log */
dump_uint8, /* SPA history */
- dump_uint64, /* SPA history offsets */
+ dump_history_offsets, /* SPA history offsets */
dump_zap, /* Pool properties */
dump_zap, /* DSL permissions */
dump_acl, /* ZFS ACL */
int print_header = 1;
int i, error;
+ dsl_pool_config_enter(dmu_objset_pool(os), FTAG);
dmu_objset_fast_stat(os, &dds);
+ dsl_pool_config_exit(dmu_objset_pool(os), FTAG);
if (dds.dds_type < DMU_OST_NUMTYPES)
type = objset_types[dds.dds_type];
/*
* Block statistics.
*/
+#define PSIZE_HISTO_SIZE (SPA_MAXBLOCKSIZE / SPA_MINBLOCKSIZE + 1)
typedef struct zdb_blkstats {
- uint64_t zb_asize;
- uint64_t zb_lsize;
- uint64_t zb_psize;
- uint64_t zb_count;
+ uint64_t zb_asize;
+ uint64_t zb_lsize;
+ uint64_t zb_psize;
+ uint64_t zb_count;
+ uint64_t zb_psize_histogram[PSIZE_HISTO_SIZE];
} zdb_blkstats_t;
/*
zdb_blkstats_t zcb_type[ZB_TOTAL + 1][ZDB_OT_TOTAL + 1];
uint64_t zcb_dedup_asize;
uint64_t zcb_dedup_blocks;
+ uint64_t zcb_start;
+ uint64_t zcb_lastprint;
+ uint64_t zcb_totalasize;
uint64_t zcb_errors[256];
int zcb_readfails;
int zcb_haderrors;
zb->zb_lsize += BP_GET_LSIZE(bp);
zb->zb_psize += BP_GET_PSIZE(bp);
zb->zb_count++;
+ zb->zb_psize_histogram[BP_GET_PSIZE(bp) >> SPA_MINBLOCKSHIFT]++;
}
if (dump_opt['L'])
mutex_exit(&spa->spa_scrub_lock);
}
-/* ARGSUSED */
static int
-zdb_blkptr_cb(spa_t *spa, zilog_t *zilog, const blkptr_t *bp, arc_buf_t *pbuf,
+zdb_blkptr_cb(spa_t *spa, zilog_t *zilog, const blkptr_t *bp,
const zbookmark_t *zb, const dnode_phys_t *dnp, void *arg)
{
zdb_cb_t *zcb = arg;
zio_nowait(zio_read(NULL, spa, bp, data, size,
zdb_blkptr_done, zcb, ZIO_PRIORITY_ASYNC_READ, flags, zb));
-
}
zcb->zcb_readfails = 0;
- if (dump_opt['b'] >= 4) {
+ if (dump_opt['b'] >= 5) {
sprintf_blkptr(blkbuf, bp);
(void) printf("objset %llu object %llu "
"level %lld offset 0x%llx %s\n",
blkbuf);
}
- return (0);
-}
+ if (dump_opt['b'] < 5 && isatty(STDERR_FILENO) &&
+ gethrtime() > zcb->zcb_lastprint + NANOSEC) {
+ uint64_t now = gethrtime();
+ char buf[10];
+ uint64_t bytes = zcb->zcb_type[ZB_TOTAL][ZDB_OT_TOTAL].zb_asize;
+ int kb_per_sec =
+ 1 + bytes / (1 + ((now - zcb->zcb_start) / 1000 / 1000));
+ int sec_remaining =
+ (zcb->zcb_totalasize - bytes) / 1024 / kb_per_sec;
-static void
-zdb_leak(space_map_t *sm, uint64_t start, uint64_t size)
-{
- vdev_t *vd = sm->sm_ppd;
+ zfs_nicenum(bytes, buf, sizeof (buf));
+ (void) fprintf(stderr,
+ "\r%5s completed (%4dMB/s) "
+ "estimated time remaining: %uhr %02umin %02usec ",
+ buf, kb_per_sec / 1024,
+ sec_remaining / 60 / 60,
+ sec_remaining / 60 % 60,
+ sec_remaining % 60);
- (void) printf("leaked space: vdev %llu, offset 0x%llx, size %llu\n",
- (u_longlong_t)vd->vdev_id, (u_longlong_t)start, (u_longlong_t)size);
-}
+ zcb->zcb_lastprint = now;
+ }
-/* ARGSUSED */
-static void
-zdb_space_map_load(space_map_t *sm)
-{
+ return (0);
}
static void
-zdb_space_map_unload(space_map_t *sm)
+zdb_leak(void *arg, uint64_t start, uint64_t size)
{
- space_map_vacate(sm, zdb_leak, sm);
-}
+ vdev_t *vd = arg;
-/* ARGSUSED */
-static void
-zdb_space_map_claim(space_map_t *sm, uint64_t start, uint64_t size)
-{
+ (void) printf("leaked space: vdev %llu, offset 0x%llx, size %llu\n",
+ (u_longlong_t)vd->vdev_id, (u_longlong_t)start, (u_longlong_t)size);
}
-static space_map_ops_t zdb_space_map_ops = {
- zdb_space_map_load,
- zdb_space_map_unload,
+static metaslab_ops_t zdb_metaslab_ops = {
NULL, /* alloc */
- zdb_space_map_claim,
- NULL, /* free */
- NULL /* maxsize */
+ NULL /* fragmented */
};
static void
for (m = 0; m < vd->vdev_ms_count; m++) {
metaslab_t *msp = vd->vdev_ms[m];
mutex_enter(&msp->ms_lock);
- space_map_unload(msp->ms_map);
- VERIFY(space_map_load(msp->ms_map,
- &zdb_space_map_ops, SM_ALLOC, &msp->ms_smo,
- spa->spa_meta_objset) == 0);
- msp->ms_map->sm_ppd = vd;
+ metaslab_unload(msp);
+
+ /*
+ * For leak detection, we overload the metaslab
+ * ms_tree to contain allocated segments
+ * instead of free segments. As a result,
+ * we can't use the normal metaslab_load/unload
+ * interfaces.
+ */
+ if (msp->ms_sm != NULL) {
+ msp->ms_ops = &zdb_metaslab_ops;
+ VERIFY0(space_map_load(msp->ms_sm,
+ msp->ms_tree, SM_ALLOC));
+ msp->ms_loaded = B_TRUE;
+ }
mutex_exit(&msp->ms_lock);
}
}
for (m = 0; m < vd->vdev_ms_count; m++) {
metaslab_t *msp = vd->vdev_ms[m];
mutex_enter(&msp->ms_lock);
- space_map_unload(msp->ms_map);
+
+ /*
+ * The ms_tree has been overloaded to
+ * contain allocated segments. Now that we
+ * finished traversing all blocks, any
+ * block that remains in the ms_tree
+ * represents an allocated block that we
+ * did not claim during the traversal.
+ * Claimed blocks would have been removed
+ * from the ms_tree.
+ */
+ range_tree_vacate(msp->ms_tree, zdb_leak, vd);
+ msp->ms_loaded = B_FALSE;
+
mutex_exit(&msp->ms_lock);
}
}
{
zdb_cb_t *zcb = arg;
- if (dump_opt['b'] >= 4) {
+ if (dump_opt['b'] >= 5) {
char blkbuf[BP_SPRINTF_LEN];
sprintf_blkptr(blkbuf, bp);
(void) printf("[%s] %s\n",
int leaks = 0;
int e;
- (void) printf("\nTraversing all blocks %s%s%s%s%s...\n",
+ (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 " : "",
dump_opt['c'] ? "checksums " : "",
* it's not part of any space map) is a double allocation,
* reference to a freed block, or an unclaimed log block.
*/
- bzero(&zcb, sizeof(zdb_cb_t));
+ bzero(&zcb, sizeof (zdb_cb_t));
zdb_leak_init(spa, &zcb);
/*
*/
(void) bpobj_iterate_nofree(&spa->spa_deferred_bpobj,
count_block_cb, &zcb, NULL);
- (void) bpobj_iterate_nofree(&spa->spa_dsl_pool->dp_free_bpobj,
- count_block_cb, &zcb, NULL);
- if (spa_feature_is_active(spa,
- &spa_feature_table[SPA_FEATURE_ASYNC_DESTROY])) {
+ if (spa_version(spa) >= SPA_VERSION_DEADLISTS) {
+ (void) bpobj_iterate_nofree(&spa->spa_dsl_pool->dp_free_bpobj,
+ count_block_cb, &zcb, NULL);
+ }
+ 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));
if (dump_opt['c'] > 1)
flags |= TRAVERSE_PREFETCH_DATA;
+ zcb.zcb_totalasize = metaslab_class_get_alloc(spa_normal_class(spa));
+ zcb.zcb_start = zcb.zcb_lastprint = gethrtime();
zcb.zcb_haderrors |= traverse_pool(spa, 0, flags, zdb_blkptr_cb, &zcb);
/*
else
(void) printf(" L%d %s\n",
level, typename);
+
+ if (dump_opt['b'] >= 4) {
+ (void) printf("psize "
+ "(in 512-byte sectors): "
+ "number of blocks\n");
+ dump_histogram(zb->zb_psize_histogram,
+ PSIZE_HISTO_SIZE, 0);
+ }
}
}
}
/* ARGSUSED */
static int
zdb_ddt_add_cb(spa_t *spa, zilog_t *zilog, const blkptr_t *bp,
- arc_buf_t *pbuf, const zbookmark_t *zb, const dnode_phys_t *dnp, void *arg)
+ const zbookmark_t *zb, const dnode_phys_t *dnp, void *arg)
{
avl_tree_t *t = arg;
avl_index_t where;
if (dump_opt['d'] || dump_opt['i']) {
dump_dir(dp->dp_meta_objset);
if (dump_opt['d'] >= 3) {
- dump_bpobj(&spa->spa_deferred_bpobj, "Deferred frees");
+ dump_bpobj(&spa->spa_deferred_bpobj,
+ "Deferred frees", 0);
if (spa_version(spa) >= SPA_VERSION_DEADLISTS) {
dump_bpobj(&spa->spa_dsl_pool->dp_free_bpobj,
- "Pool snapshot frees");
+ "Pool snapshot frees", 0);
}
if (spa_feature_is_active(spa,
- &spa_feature_table[SPA_FEATURE_ASYNC_DESTROY])) {
+ SPA_FEATURE_ASYNC_DESTROY)) {
dump_bptree(spa->spa_meta_objset,
spa->spa_dsl_pool->dp_bptree_obj,
"Pool dataset frees");
if (dump_opt['b'] || dump_opt['c'])
rc = dump_block_stats(spa);
+ if (rc == 0)
+ rc = verify_spacemap_refcounts(spa);
+
if (dump_opt['s'])
show_pool_stats(spa);
psize = size;
lsize = size;
- pbuf = umem_alloc(SPA_MAXBLOCKSIZE, UMEM_NOFAIL);
+ pbuf = umem_alloc_aligned(SPA_MAXBLOCKSIZE, 512, UMEM_NOFAIL);
lbuf = umem_alloc(SPA_MAXBLOCKSIZE, UMEM_NOFAIL);
BP_ZERO(bp);
char *target;
nvlist_t *policy = NULL;
uint64_t max_txg = UINT64_MAX;
+ int flags = ZFS_IMPORT_MISSING_LOG;
int rewind = ZPOOL_NEVER_REWIND;
+ char *spa_config_path_env;
+ const char *opts = "bcdhilmM:suCDRSAFLVXevp:t:U:P";
(void) setrlimit(RLIMIT_NOFILE, &rl);
(void) enable_extended_FILE_stdio(-1, -1);
dprintf_setup(&argc, argv);
- while ((c = getopt(argc, argv, "bcdhilmM:suCDRSAFLXevp:t:U:P")) != -1) {
+ /*
+ * If there is an environment variable SPA_CONFIG_PATH it overrides
+ * default spa_config_path setting. If -U flag is specified it will
+ * override this environment variable settings once again.
+ */
+ spa_config_path_env = getenv("SPA_CONFIG_PATH");
+ if (spa_config_path_env != NULL)
+ spa_config_path = spa_config_path_env;
+
+ while ((c = getopt(argc, argv, opts)) != -1) {
switch (c) {
case 'b':
case 'c':
case 'P':
dump_opt[c]++;
break;
+ case 'V':
+ flags = ZFS_IMPORT_VERBATIM;
+ break;
case 'v':
verbose++;
break;
fatal("can't open '%s': %s",
target, strerror(ENOMEM));
}
- if ((error = spa_import(name, cfg, NULL,
- ZFS_IMPORT_MISSING_LOG)) != 0) {
- error = spa_import(name, cfg, NULL,
- ZFS_IMPORT_VERBATIM);
- }
+ error = spa_import(name, cfg, NULL, flags);
}
}
projects / mirror_zfs.git / blobdiff