diff options
author | Brian Behlendorf <[email protected]> | 2008-12-11 11:26:43 -0800 |
---|---|---|
committer | Brian Behlendorf <[email protected]> | 2008-12-11 11:26:43 -0800 |
commit | f07f82e93ab189c6fbb64f3d8ec6ae77f8da47af (patch) | |
tree | 97ff21d483a4966eb8db8748e09512465671988c /cmd | |
parent | 6921187bfe3503bb7b106a29544c26bad13705c8 (diff) | |
parent | b453b61272db34cd097f8636f57033785b00d932 (diff) |
Merge commit 'refs/top-bases/fix-pthreads' into fix-pthreads
Diffstat (limited to 'cmd')
-rw-r--r-- | cmd/zdb/zdb.c | 2418 | ||||
-rw-r--r-- | cmd/zdb/zdb_il.c | 363 | ||||
-rw-r--r-- | cmd/zdump/zdump.c | 664 | ||||
-rw-r--r-- | cmd/zfs/zfs_iter.c | 420 | ||||
-rw-r--r-- | cmd/zfs/zfs_iter.h | 54 | ||||
-rw-r--r-- | cmd/zfs/zfs_main.c | 4229 | ||||
-rw-r--r-- | cmd/zfs/zfs_util.h | 44 | ||||
-rw-r--r-- | cmd/zinject/translate.c | 474 | ||||
-rw-r--r-- | cmd/zinject/zinject.c | 768 | ||||
-rw-r--r-- | cmd/zinject/zinject.h | 71 | ||||
-rw-r--r-- | cmd/zpool/zpool_iter.c | 252 | ||||
-rw-r--r-- | cmd/zpool/zpool_main.c | 3950 | ||||
-rw-r--r-- | cmd/zpool/zpool_util.c | 104 | ||||
-rw-r--r-- | cmd/zpool/zpool_util.h | 72 | ||||
-rw-r--r-- | cmd/zpool/zpool_vdev.c | 1401 | ||||
-rw-r--r-- | cmd/ztest/ztest.c | 3476 |
16 files changed, 18760 insertions, 0 deletions
diff --git a/cmd/zdb/zdb.c b/cmd/zdb/zdb.c new file mode 100644 index 000000000..253a1346a --- /dev/null +++ b/cmd/zdb/zdb.c @@ -0,0 +1,2418 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ +/* + * Copyright 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#include <stdio.h> +#include <stdio_ext.h> +#include <stdlib.h> +#include <ctype.h> +#include <sys/zfs_context.h> +#include <sys/spa.h> +#include <sys/spa_impl.h> +#include <sys/dmu.h> +#include <sys/zap.h> +#include <sys/fs/zfs.h> +#include <sys/zfs_znode.h> +#include <sys/vdev.h> +#include <sys/vdev_impl.h> +#include <sys/metaslab_impl.h> +#include <sys/dmu_objset.h> +#include <sys/dsl_dir.h> +#include <sys/dsl_dataset.h> +#include <sys/dsl_pool.h> +#include <sys/dbuf.h> +#include <sys/zil.h> +#include <sys/zil_impl.h> +#include <sys/stat.h> +#include <sys/resource.h> +#include <sys/dmu_traverse.h> +#include <sys/zio_checksum.h> +#include <sys/zio_compress.h> +#include <sys/zfs_fuid.h> +#include <sys/arc.h> +#undef ZFS_MAXNAMELEN +#undef verify +#include <libzfs.h> + +const char cmdname[] = "zdb"; +uint8_t dump_opt[256]; + +typedef void object_viewer_t(objset_t *, uint64_t, void *data, size_t size); + +extern void dump_intent_log(zilog_t *); +uint64_t *zopt_object = NULL; +int zopt_objects = 0; +libzfs_handle_t *g_zfs; +boolean_t zdb_sig_user_data = B_TRUE; +int zdb_sig_cksumalg = ZIO_CHECKSUM_SHA256; + +/* + * These libumem hooks provide a reasonable set of defaults for the allocator's + * debugging facilities. + */ +const char * +_umem_debug_init() +{ + return ("default,verbose"); /* $UMEM_DEBUG setting */ +} + +const char * +_umem_logging_init(void) +{ + return ("fail,contents"); /* $UMEM_LOGGING setting */ +} + +static void +usage(void) +{ + (void) fprintf(stderr, + "Usage: %s [-udibcsv] [-U cachefile_path] " + "[-S user:cksumalg] " + "dataset [object...]\n" + " %s -C [pool]\n" + " %s -l dev\n" + " %s -R pool:vdev:offset:size:flags\n" + " %s [-p path_to_vdev_dir]\n" + " %s -e pool | GUID | devid ...\n", + cmdname, cmdname, cmdname, cmdname, cmdname, cmdname); + + (void) fprintf(stderr, " -u uberblock\n"); + (void) fprintf(stderr, " -d datasets\n"); + (void) fprintf(stderr, " -C cached pool configuration\n"); + (void) fprintf(stderr, " -i intent logs\n"); + (void) fprintf(stderr, " -b block statistics\n"); + (void) fprintf(stderr, " -c checksum all data blocks\n"); + (void) fprintf(stderr, " -s report stats on zdb's I/O\n"); + (void) fprintf(stderr, " -S <user|all>:<cksum_alg|all> -- " + "dump blkptr signatures\n"); + (void) fprintf(stderr, " -v verbose (applies to all others)\n"); + (void) fprintf(stderr, " -l dump label contents\n"); + (void) fprintf(stderr, " -U cachefile_path -- use alternate " + "cachefile\n"); + (void) fprintf(stderr, " -R read and display block from a " + "device\n"); + (void) fprintf(stderr, " -e Pool is exported/destroyed/" + "has altroot\n"); + (void) fprintf(stderr, " -p <Path to vdev dir> (use with -e)\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); +} + +static void +fatal(const char *fmt, ...) +{ + va_list ap; + + va_start(ap, fmt); + (void) fprintf(stderr, "%s: ", cmdname); + (void) vfprintf(stderr, fmt, ap); + va_end(ap); + (void) fprintf(stderr, "\n"); + + abort(); +} + +static void +dump_nvlist(nvlist_t *list, int indent) +{ + nvpair_t *elem = NULL; + + while ((elem = nvlist_next_nvpair(list, elem)) != NULL) { + switch (nvpair_type(elem)) { + case DATA_TYPE_STRING: + { + char *value; + + VERIFY(nvpair_value_string(elem, &value) == 0); + (void) printf("%*s%s='%s'\n", indent, "", + nvpair_name(elem), value); + } + break; + + case DATA_TYPE_UINT64: + { + uint64_t value; + + VERIFY(nvpair_value_uint64(elem, &value) == 0); + (void) printf("%*s%s=%llu\n", indent, "", + nvpair_name(elem), (u_longlong_t)value); + } + break; + + case DATA_TYPE_NVLIST: + { + nvlist_t *value; + + VERIFY(nvpair_value_nvlist(elem, &value) == 0); + (void) printf("%*s%s\n", indent, "", + nvpair_name(elem)); + dump_nvlist(value, indent + 4); + } + break; + + case DATA_TYPE_NVLIST_ARRAY: + { + nvlist_t **value; + uint_t c, count; + + VERIFY(nvpair_value_nvlist_array(elem, &value, + &count) == 0); + + for (c = 0; c < count; c++) { + (void) printf("%*s%s[%u]\n", indent, "", + nvpair_name(elem), c); + dump_nvlist(value[c], indent + 8); + } + } + break; + + default: + + (void) printf("bad config type %d for %s\n", + nvpair_type(elem), nvpair_name(elem)); + } + } +} + +/* ARGSUSED */ +static void +dump_packed_nvlist(objset_t *os, uint64_t object, void *data, size_t size) +{ + nvlist_t *nv; + size_t nvsize = *(uint64_t *)data; + char *packed = umem_alloc(nvsize, UMEM_NOFAIL); + + VERIFY(0 == dmu_read(os, object, 0, nvsize, packed)); + + VERIFY(nvlist_unpack(packed, nvsize, &nv, 0) == 0); + + umem_free(packed, nvsize); + + dump_nvlist(nv, 8); + + nvlist_free(nv); +} + +const char dump_zap_stars[] = "****************************************"; +const int dump_zap_width = sizeof (dump_zap_stars) - 1; + +static void +dump_zap_histogram(uint64_t histo[ZAP_HISTOGRAM_SIZE]) +{ + int i; + int minidx = ZAP_HISTOGRAM_SIZE - 1; + int maxidx = 0; + uint64_t max = 0; + + for (i = 0; i < ZAP_HISTOGRAM_SIZE; i++) { + if (histo[i] > max) + max = histo[i]; + if (histo[i] > 0 && i > maxidx) + maxidx = i; + if (histo[i] > 0 && i < minidx) + minidx = i; + } + + if (max < dump_zap_width) + max = dump_zap_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]); +} + +static void +dump_zap_stats(objset_t *os, uint64_t object) +{ + int error; + zap_stats_t zs; + + error = zap_get_stats(os, object, &zs); + if (error) + return; + + if (zs.zs_ptrtbl_len == 0) { + ASSERT(zs.zs_num_blocks == 1); + (void) printf("\tmicrozap: %llu bytes, %llu entries\n", + (u_longlong_t)zs.zs_blocksize, + (u_longlong_t)zs.zs_num_entries); + return; + } + + (void) printf("\tFat ZAP stats:\n"); + + (void) printf("\t\tPointer table:\n"); + (void) printf("\t\t\t%llu elements\n", + (u_longlong_t)zs.zs_ptrtbl_len); + (void) printf("\t\t\tzt_blk: %llu\n", + (u_longlong_t)zs.zs_ptrtbl_zt_blk); + (void) printf("\t\t\tzt_numblks: %llu\n", + (u_longlong_t)zs.zs_ptrtbl_zt_numblks); + (void) printf("\t\t\tzt_shift: %llu\n", + (u_longlong_t)zs.zs_ptrtbl_zt_shift); + (void) printf("\t\t\tzt_blks_copied: %llu\n", + (u_longlong_t)zs.zs_ptrtbl_blks_copied); + (void) printf("\t\t\tzt_nextblk: %llu\n", + (u_longlong_t)zs.zs_ptrtbl_nextblk); + + (void) printf("\t\tZAP entries: %llu\n", + (u_longlong_t)zs.zs_num_entries); + (void) printf("\t\tLeaf blocks: %llu\n", + (u_longlong_t)zs.zs_num_leafs); + (void) printf("\t\tTotal blocks: %llu\n", + (u_longlong_t)zs.zs_num_blocks); + (void) printf("\t\tzap_block_type: 0x%llx\n", + (u_longlong_t)zs.zs_block_type); + (void) printf("\t\tzap_magic: 0x%llx\n", + (u_longlong_t)zs.zs_magic); + (void) printf("\t\tzap_salt: 0x%llx\n", + (u_longlong_t)zs.zs_salt); + + (void) printf("\t\tLeafs with 2^n pointers:\n"); + dump_zap_histogram(zs.zs_leafs_with_2n_pointers); + + (void) printf("\t\tBlocks with n*5 entries:\n"); + dump_zap_histogram(zs.zs_blocks_with_n5_entries); + + (void) printf("\t\tBlocks n/10 full:\n"); + dump_zap_histogram(zs.zs_blocks_n_tenths_full); + + (void) printf("\t\tEntries with n chunks:\n"); + dump_zap_histogram(zs.zs_entries_using_n_chunks); + + (void) printf("\t\tBuckets with n entries:\n"); + dump_zap_histogram(zs.zs_buckets_with_n_entries); +} + +/*ARGSUSED*/ +static void +dump_none(objset_t *os, uint64_t object, void *data, size_t size) +{ +} + +/*ARGSUSED*/ +void +dump_uint8(objset_t *os, uint64_t object, void *data, size_t size) +{ +} + +/*ARGSUSED*/ +static void +dump_uint64(objset_t *os, uint64_t object, void *data, size_t size) +{ +} + +/*ARGSUSED*/ +static void +dump_zap(objset_t *os, uint64_t object, void *data, size_t size) +{ + zap_cursor_t zc; + zap_attribute_t attr; + void *prop; + int i; + + dump_zap_stats(os, object); + (void) printf("\n"); + + for (zap_cursor_init(&zc, os, object); + zap_cursor_retrieve(&zc, &attr) == 0; + zap_cursor_advance(&zc)) { + (void) printf("\t\t%s = ", attr.za_name); + if (attr.za_num_integers == 0) { + (void) printf("\n"); + continue; + } + prop = umem_zalloc(attr.za_num_integers * + attr.za_integer_length, UMEM_NOFAIL); + (void) zap_lookup(os, object, attr.za_name, + attr.za_integer_length, attr.za_num_integers, prop); + if (attr.za_integer_length == 1) { + (void) printf("%s", (char *)prop); + } else { + for (i = 0; i < attr.za_num_integers; i++) { + switch (attr.za_integer_length) { + case 2: + (void) printf("%u ", + ((uint16_t *)prop)[i]); + break; + case 4: + (void) printf("%u ", + ((uint32_t *)prop)[i]); + break; + case 8: + (void) printf("%lld ", + (u_longlong_t)((int64_t *)prop)[i]); + break; + } + } + } + (void) printf("\n"); + umem_free(prop, attr.za_num_integers * attr.za_integer_length); + } + zap_cursor_fini(&zc); +} + +/*ARGSUSED*/ +static void +dump_zpldir(objset_t *os, uint64_t object, void *data, size_t size) +{ + zap_cursor_t zc; + zap_attribute_t attr; + const char *typenames[] = { + /* 0 */ "not specified", + /* 1 */ "FIFO", + /* 2 */ "Character Device", + /* 3 */ "3 (invalid)", + /* 4 */ "Directory", + /* 5 */ "5 (invalid)", + /* 6 */ "Block Device", + /* 7 */ "7 (invalid)", + /* 8 */ "Regular File", + /* 9 */ "9 (invalid)", + /* 10 */ "Symbolic Link", + /* 11 */ "11 (invalid)", + /* 12 */ "Socket", + /* 13 */ "Door", + /* 14 */ "Event Port", + /* 15 */ "15 (invalid)", + }; + + dump_zap_stats(os, object); + (void) printf("\n"); + + for (zap_cursor_init(&zc, os, object); + zap_cursor_retrieve(&zc, &attr) == 0; + zap_cursor_advance(&zc)) { + (void) printf("\t\t%s = %lld (type: %s)\n", + attr.za_name, ZFS_DIRENT_OBJ(attr.za_first_integer), + typenames[ZFS_DIRENT_TYPE(attr.za_first_integer)]); + } + zap_cursor_fini(&zc); +} + +static void +dump_spacemap(objset_t *os, space_map_obj_t *smo, 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) + return; + + /* + * Print out the freelist entries in both encoded and decoded form. + */ + alloc = 0; + for (offset = 0; offset < smo->smo_objsize; offset += sizeof (entry)) { + VERIFY(0 == dmu_read(os, smo->smo_object, offset, + sizeof (entry), &entry)); + if (SM_DEBUG_DECODE(entry)) { + (void) printf("\t\t[%4llu] %s: txg %llu, pass %llu\n", + (u_longlong_t)(offset / sizeof (entry)), + ddata[SM_DEBUG_ACTION_DECODE(entry)], + (u_longlong_t)SM_DEBUG_TXG_DECODE(entry), + (u_longlong_t)SM_DEBUG_SYNCPASS_DECODE(entry)); + } else { + (void) printf("\t\t[%4llu] %c range:" + " %08llx-%08llx size: %06llx\n", + (u_longlong_t)(offset / sizeof (entry)), + SM_TYPE_DECODE(entry) == SM_ALLOC ? 'A' : 'F', + (u_longlong_t)((SM_OFFSET_DECODE(entry) << + mapshift) + mapstart), + (u_longlong_t)((SM_OFFSET_DECODE(entry) << + mapshift) + mapstart + (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; + else + alloc -= SM_RUN_DECODE(entry) << mapshift; + } + } + if (alloc != smo->smo_alloc) { + (void) printf("space_map_object alloc (%llu) INCONSISTENT " + "with space map summary (%llu)\n", + (u_longlong_t)smo->smo_alloc, (u_longlong_t)alloc); + } +} + +static void +dump_metaslab(metaslab_t *msp) +{ + char freebuf[5]; + space_map_obj_t *smo = &msp->ms_smo; + vdev_t *vd = msp->ms_group->mg_vd; + spa_t *spa = vd->vdev_spa; + + nicenum(msp->ms_map.sm_size - smo->smo_alloc, freebuf); + + if (dump_opt['d'] <= 5) { + (void) printf("\t%10llx %10llu %5s\n", + (u_longlong_t)msp->ms_map.sm_start, + (u_longlong_t)smo->smo_object, + freebuf); + return; + } + + (void) printf( + "\tvdev %llu offset %08llx spacemap %4llu free %5s\n", + (u_longlong_t)vd->vdev_id, (u_longlong_t)msp->ms_map.sm_start, + (u_longlong_t)smo->smo_object, freebuf); + + ASSERT(msp->ms_map.sm_size == (1ULL << vd->vdev_ms_shift)); + + dump_spacemap(spa->spa_meta_objset, smo, &msp->ms_map); +} + +static void +dump_metaslabs(spa_t *spa) +{ + vdev_t *rvd = spa->spa_root_vdev; + vdev_t *vd; + int c, m; + + (void) printf("\nMetaslabs:\n"); + + for (c = 0; c < rvd->vdev_children; c++) { + vd = rvd->vdev_child[c]; + + (void) printf("\n vdev %llu\n\n", (u_longlong_t)vd->vdev_id); + + if (dump_opt['d'] <= 5) { + (void) printf("\t%10s %10s %5s\n", + "offset", "spacemap", "free"); + (void) printf("\t%10s %10s %5s\n", + "------", "--------", "----"); + } + for (m = 0; m < vd->vdev_ms_count; m++) + dump_metaslab(vd->vdev_ms[m]); + (void) printf("\n"); + } +} + +static void +dump_dtl(vdev_t *vd, int indent) +{ + avl_tree_t *t = &vd->vdev_dtl_map.sm_root; + space_seg_t *ss; + vdev_t *pvd; + int c; + + if (indent == 0) + (void) printf("\nDirty time logs:\n\n"); + + (void) printf("\t%*s%s\n", indent, "", + vd->vdev_path ? vd->vdev_path : + vd->vdev_parent ? vd->vdev_ops->vdev_op_type : + spa_name(vd->vdev_spa)); + + for (ss = avl_first(t); ss; ss = AVL_NEXT(t, ss)) { + /* + * Everything in this DTL must appear in all parent DTL unions. + */ + for (pvd = vd; pvd; pvd = pvd->vdev_parent) + ASSERT(vdev_dtl_contains(&pvd->vdev_dtl_map, + ss->ss_start, ss->ss_end - ss->ss_start)); + (void) printf("\t%*soutage [%llu,%llu] length %llu\n", + indent, "", + (u_longlong_t)ss->ss_start, + (u_longlong_t)ss->ss_end - 1, + (u_longlong_t)(ss->ss_end - ss->ss_start)); + } + + (void) printf("\n"); + + if (dump_opt['d'] > 5 && vd->vdev_children == 0) { + dump_spacemap(vd->vdev_spa->spa_meta_objset, &vd->vdev_dtl, + &vd->vdev_dtl_map); + (void) printf("\n"); + } + + for (c = 0; c < vd->vdev_children; c++) + dump_dtl(vd->vdev_child[c], indent + 4); +} + +/*ARGSUSED*/ +static void +dump_dnode(objset_t *os, uint64_t object, void *data, size_t size) +{ +} + +static uint64_t +blkid2offset(const dnode_phys_t *dnp, int level, uint64_t blkid) +{ + if (level < 0) + return (blkid); + + return ((blkid << (level * (dnp->dn_indblkshift - SPA_BLKPTRSHIFT))) * + dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT); +} + +static void +sprintf_blkptr_compact(char *blkbuf, blkptr_t *bp, int alldvas) +{ + dva_t *dva = bp->blk_dva; + int ndvas = alldvas ? BP_GET_NDVAS(bp) : 1; + int i; + + blkbuf[0] = '\0'; + + for (i = 0; i < ndvas; i++) + (void) sprintf(blkbuf + strlen(blkbuf), "%llu:%llx:%llx ", + (u_longlong_t)DVA_GET_VDEV(&dva[i]), + (u_longlong_t)DVA_GET_OFFSET(&dva[i]), + (u_longlong_t)DVA_GET_ASIZE(&dva[i])); + + (void) sprintf(blkbuf + strlen(blkbuf), "%llxL/%llxP F=%llu B=%llu", + (u_longlong_t)BP_GET_LSIZE(bp), + (u_longlong_t)BP_GET_PSIZE(bp), + (u_longlong_t)bp->blk_fill, + (u_longlong_t)bp->blk_birth); +} + +static void +print_indirect(blkptr_t *bp, const zbookmark_t *zb, + const dnode_phys_t *dnp) +{ + char blkbuf[BP_SPRINTF_LEN]; + int l; + + ASSERT3U(BP_GET_TYPE(bp), ==, dnp->dn_type); + ASSERT3U(BP_GET_LEVEL(bp), ==, zb->zb_level); + + (void) printf("%16llx ", + (u_longlong_t)blkid2offset(dnp, zb->zb_level, zb->zb_blkid)); + + ASSERT(zb->zb_level >= 0); + + for (l = dnp->dn_nlevels - 1; l >= -1; l--) { + if (l == zb->zb_level) { + (void) printf("L%llx", (u_longlong_t)zb->zb_level); + } else { + (void) printf(" "); + } + } + + sprintf_blkptr_compact(blkbuf, bp, dump_opt['d'] > 5 ? 1 : 0); + (void) printf("%s\n", blkbuf); +} + +#define SET_BOOKMARK(zb, objset, object, level, blkid) \ +{ \ + (zb)->zb_objset = objset; \ + (zb)->zb_object = object; \ + (zb)->zb_level = level; \ + (zb)->zb_blkid = blkid; \ +} + +static int +visit_indirect(spa_t *spa, const dnode_phys_t *dnp, + blkptr_t *bp, const zbookmark_t *zb) +{ + int err; + + if (bp->blk_birth == 0) + return (0); + + print_indirect(bp, zb, dnp); + + if (BP_GET_LEVEL(bp) > 0) { + uint32_t flags = ARC_WAIT; + int i; + blkptr_t *cbp; + int epb = BP_GET_LSIZE(bp) >> SPA_BLKPTRSHIFT; + arc_buf_t *buf; + uint64_t fill = 0; + + err = arc_read_nolock(NULL, spa, bp, arc_getbuf_func, &buf, + ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb); + if (err) + return (err); + + /* recursively visit blocks below this */ + cbp = buf->b_data; + for (i = 0; i < epb; i++, cbp++) { + zbookmark_t czb; + + SET_BOOKMARK(&czb, zb->zb_objset, zb->zb_object, + zb->zb_level - 1, + zb->zb_blkid * epb + i); + err = visit_indirect(spa, dnp, cbp, &czb); + if (err) + break; + fill += cbp->blk_fill; + } + ASSERT3U(fill, ==, bp->blk_fill); + (void) arc_buf_remove_ref(buf, &buf); + } + + return (err); +} + +/*ARGSUSED*/ +static void +dump_indirect(dnode_t *dn) +{ + dnode_phys_t *dnp = dn->dn_phys; + int j; + zbookmark_t czb; + + (void) printf("Indirect blocks:\n"); + + SET_BOOKMARK(&czb, dmu_objset_id(&dn->dn_objset->os), + dn->dn_object, dnp->dn_nlevels - 1, 0); + for (j = 0; j < dnp->dn_nblkptr; j++) { + czb.zb_blkid = j; + (void) visit_indirect(dmu_objset_spa(&dn->dn_objset->os), 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) +{ + dsl_dir_phys_t *dd = data; + time_t crtime; + char nice[6]; + + if (dd == NULL) + return; + + ASSERT3U(size, >=, sizeof (dsl_dir_phys_t)); + + crtime = dd->dd_creation_time; + (void) printf("\t\tcreation_time = %s", ctime(&crtime)); + (void) printf("\t\thead_dataset_obj = %llu\n", + (u_longlong_t)dd->dd_head_dataset_obj); + (void) printf("\t\tparent_dir_obj = %llu\n", + (u_longlong_t)dd->dd_parent_obj); + (void) printf("\t\torigin_obj = %llu\n", + (u_longlong_t)dd->dd_origin_obj); + (void) printf("\t\tchild_dir_zapobj = %llu\n", + (u_longlong_t)dd->dd_child_dir_zapobj); + nicenum(dd->dd_used_bytes, nice); + (void) printf("\t\tused_bytes = %s\n", nice); + nicenum(dd->dd_compressed_bytes, nice); + (void) printf("\t\tcompressed_bytes = %s\n", nice); + nicenum(dd->dd_uncompressed_bytes, nice); + (void) printf("\t\tuncompressed_bytes = %s\n", nice); + nicenum(dd->dd_quota, nice); + (void) printf("\t\tquota = %s\n", nice); + nicenum(dd->dd_reserved, nice); + (void) printf("\t\treserved = %s\n", nice); + (void) printf("\t\tprops_zapobj = %llu\n", + (u_longlong_t)dd->dd_props_zapobj); + (void) printf("\t\tdeleg_zapobj = %llu\n", + (u_longlong_t)dd->dd_deleg_zapobj); + (void) printf("\t\tflags = %llx\n", + (u_longlong_t)dd->dd_flags); + +#define DO(which) \ + nicenum(dd->dd_used_breakdown[DD_USED_ ## which], nice); \ + (void) printf("\t\tused_breakdown[" #which "] = %s\n", nice) + DO(HEAD); + DO(SNAP); + DO(CHILD); + DO(CHILD_RSRV); + DO(REFRSRV); +#undef DO +} + +/*ARGSUSED*/ +static void +dump_dsl_dataset(objset_t *os, uint64_t object, void *data, size_t size) +{ + dsl_dataset_phys_t *ds = data; + time_t crtime; + char used[6], compressed[6], uncompressed[6], unique[6]; + char blkbuf[BP_SPRINTF_LEN]; + + if (ds == NULL) + return; + + ASSERT(size == sizeof (*ds)); + crtime = ds->ds_creation_time; + nicenum(ds->ds_used_bytes, used); + nicenum(ds->ds_compressed_bytes, compressed); + nicenum(ds->ds_uncompressed_bytes, uncompressed); + nicenum(ds->ds_unique_bytes, unique); + sprintf_blkptr(blkbuf, BP_SPRINTF_LEN, &ds->ds_bp); + + (void) printf("\t\tdir_obj = %llu\n", + (u_longlong_t)ds->ds_dir_obj); + (void) printf("\t\tprev_snap_obj = %llu\n", + (u_longlong_t)ds->ds_prev_snap_obj); + (void) printf("\t\tprev_snap_txg = %llu\n", + (u_longlong_t)ds->ds_prev_snap_txg); + (void) printf("\t\tnext_snap_obj = %llu\n", + (u_longlong_t)ds->ds_next_snap_obj); + (void) printf("\t\tsnapnames_zapobj = %llu\n", + (u_longlong_t)ds->ds_snapnames_zapobj); + (void) printf("\t\tnum_children = %llu\n", + (u_longlong_t)ds->ds_num_children); + (void) printf("\t\tcreation_time = %s", ctime(&crtime)); + (void) printf("\t\tcreation_txg = %llu\n", + (u_longlong_t)ds->ds_creation_txg); + (void) printf("\t\tdeadlist_obj = %llu\n", + (u_longlong_t)ds->ds_deadlist_obj); + (void) printf("\t\tused_bytes = %s\n", used); + (void) printf("\t\tcompressed_bytes = %s\n", compressed); + (void) printf("\t\tuncompressed_bytes = %s\n", uncompressed); + (void) printf("\t\tunique = %s\n", unique); + (void) printf("\t\tfsid_guid = %llu\n", + (u_longlong_t)ds->ds_fsid_guid); + (void) printf("\t\tguid = %llu\n", + (u_longlong_t)ds->ds_guid); + (void) printf("\t\tflags = %llx\n", + (u_longlong_t)ds->ds_flags); + (void) printf("\t\tnext_clones_obj = %llu\n", + (u_longlong_t)ds->ds_next_clones_obj); + (void) printf("\t\tprops_obj = %llu\n", + (u_longlong_t)ds->ds_props_obj); + (void) printf("\t\tbp = %s\n", blkbuf); +} + +static void +dump_bplist(objset_t *mos, uint64_t object, char *name) +{ + bplist_t bpl = { 0 }; + blkptr_t blk, *bp = &blk; + uint64_t itor = 0; + char bytes[6]; + char comp[6]; + char uncomp[6]; + + if (dump_opt['d'] < 3) + return; + + mutex_init(&bpl.bpl_lock, NULL, MUTEX_DEFAULT, NULL); + VERIFY(0 == bplist_open(&bpl, mos, object)); + if (bplist_empty(&bpl)) { + bplist_close(&bpl); + mutex_destroy(&bpl.bpl_lock); + return; + } + + nicenum(bpl.bpl_phys->bpl_bytes, bytes); + if (bpl.bpl_dbuf->db_size == sizeof (bplist_phys_t)) { + nicenum(bpl.bpl_phys->bpl_comp, comp); + nicenum(bpl.bpl_phys->bpl_uncomp, uncomp); + (void) printf("\n %s: %llu entries, %s (%s/%s comp)\n", + name, (u_longlong_t)bpl.bpl_phys->bpl_entries, + bytes, comp, uncomp); + } else { + (void) printf("\n %s: %llu entries, %s\n", + name, (u_longlong_t)bpl.bpl_phys->bpl_entries, bytes); + } + + if (dump_opt['d'] < 5) { + bplist_close(&bpl); + mutex_destroy(&bpl.bpl_lock); + return; + } + + (void) printf("\n"); + + while (bplist_iterate(&bpl, &itor, bp) == 0) { + char blkbuf[BP_SPRINTF_LEN]; + + ASSERT(bp->blk_birth != 0); + sprintf_blkptr_compact(blkbuf, bp, dump_opt['d'] > 5 ? 1 : 0); + (void) printf("\tItem %3llu: %s\n", + (u_longlong_t)itor - 1, blkbuf); + } + + bplist_close(&bpl); + mutex_destroy(&bpl.bpl_lock); +} + +static avl_tree_t idx_tree; +static avl_tree_t domain_tree; +static boolean_t fuid_table_loaded; + +static void +fuid_table_destroy() +{ + if (fuid_table_loaded) { + zfs_fuid_table_destroy(&idx_tree, &domain_tree); + fuid_table_loaded = B_FALSE; + } +} + +/* + * 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. + */ +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)); + (void) printf("\t%s %llx [%s-%d]\n", id_type, + (u_longlong_t)id, domain, (int)FUID_RID(id)); + } else { + (void) printf("\t%s %llu\n", id_type, (u_longlong_t)id); + } + +} + +static void +dump_uidgid(objset_t *os, znode_phys_t *zp) +{ + uint32_t uid_idx, gid_idx; + + uid_idx = FUID_INDEX(zp->zp_uid); + gid_idx = FUID_INDEX(zp->zp_gid); + + /* Load domain table, if not already loaded */ + if (!fuid_table_loaded && (uid_idx || gid_idx)) { + uint64_t fuid_obj; + + /* first find the fuid object. It lives in the master node */ + VERIFY(zap_lookup(os, MASTER_NODE_OBJ, ZFS_FUID_TABLES, + 8, 1, &fuid_obj) == 0); + (void) zfs_fuid_table_load(os, fuid_obj, + &idx_tree, &domain_tree); + fuid_table_loaded = B_TRUE; + } + + print_idstr(zp->zp_uid, "uid"); + print_idstr(zp->zp_gid, "gid"); +} + +/*ARGSUSED*/ +static void +dump_znode(objset_t *os, uint64_t object, void *data, size_t size) +{ + znode_phys_t *zp = data; + time_t z_crtime, z_atime, z_mtime, z_ctime; + char path[MAXPATHLEN * 2]; /* allow for xattr and failure prefix */ + int error; + + ASSERT(size >= sizeof (znode_phys_t)); + + error = zfs_obj_to_path(os, object, path, sizeof (path)); + if (error != 0) { + (void) snprintf(path, sizeof (path), "\?\?\?<object#%llu>", + (u_longlong_t)object); + } + + if (dump_opt['d'] < 3) { + (void) printf("\t%s\n", path); + return; + } + + z_crtime = (time_t)zp->zp_crtime[0]; + z_atime = (time_t)zp->zp_atime[0]; + z_mtime = (time_t)zp->zp_mtime[0]; + z_ctime = (time_t)zp->zp_ctime[0]; + + (void) printf("\tpath %s\n", path); + dump_uidgid(os, zp); + (void) printf("\tatime %s", ctime(&z_atime)); + (void) printf("\tmtime %s", ctime(&z_mtime)); + (void) printf("\tctime %s", ctime(&z_ctime)); + (void) printf("\tcrtime %s", ctime(&z_crtime)); + (void) printf("\tgen %llu\n", (u_longlong_t)zp->zp_gen); + (void) printf("\tmode %llo\n", (u_longlong_t)zp->zp_mode); + (void) printf("\tsize %llu\n", (u_longlong_t)zp->zp_size); + (void) printf("\tparent %llu\n", (u_longlong_t)zp->zp_parent); + (void) printf("\tlinks %llu\n", (u_longlong_t)zp->zp_links); + (void) printf("\txattr %llu\n", (u_longlong_t)zp->zp_xattr); + (void) printf("\trdev 0x%016llx\n", (u_longlong_t)zp->zp_rdev); +} + +/*ARGSUSED*/ +static void +dump_acl(objset_t *os, uint64_t object, void *data, size_t size) +{ +} + +/*ARGSUSED*/ +static void +dump_dmu_objset(objset_t *os, uint64_t object, void *data, size_t size) +{ +} + +static object_viewer_t *object_viewer[DMU_OT_NUMTYPES] = { + dump_none, /* unallocated */ + dump_zap, /* object directory */ + dump_uint64, /* object array */ + dump_none, /* packed nvlist */ + dump_packed_nvlist, /* packed nvlist size */ + dump_none, /* bplist */ + dump_none, /* bplist header */ + dump_none, /* SPA space map header */ + dump_none, /* SPA space map */ + dump_none, /* ZIL intent log */ + dump_dnode, /* DMU dnode */ + dump_dmu_objset, /* DMU objset */ + dump_dsl_dir, /* DSL directory */ + dump_zap, /* DSL directory child map */ + dump_zap, /* DSL dataset snap map */ + dump_zap, /* DSL props */ + dump_dsl_dataset, /* DSL dataset */ + dump_znode, /* ZFS znode */ + dump_acl, /* ZFS V0 ACL */ + dump_uint8, /* ZFS plain file */ + dump_zpldir, /* ZFS directory */ + dump_zap, /* ZFS master node */ + dump_zap, /* ZFS delete queue */ + dump_uint8, /* zvol object */ + dump_zap, /* zvol prop */ + dump_uint8, /* other uint8[] */ + dump_uint64, /* other uint64[] */ + dump_zap, /* other ZAP */ + dump_zap, /* persistent error log */ + dump_uint8, /* SPA history */ + dump_uint64, /* SPA history offsets */ + dump_zap, /* Pool properties */ + dump_zap, /* DSL permissions */ + dump_acl, /* ZFS ACL */ + dump_uint8, /* ZFS SYSACL */ + dump_none, /* FUID nvlist */ + dump_packed_nvlist, /* FUID nvlist size */ + dump_zap, /* DSL dataset next clones */ + dump_zap, /* DSL scrub queue */ +}; + +static void +dump_object(objset_t *os, uint64_t object, int verbosity, int *print_header) +{ + dmu_buf_t *db = NULL; + dmu_object_info_t doi; + dnode_t *dn; + void *bonus = NULL; + size_t bsize = 0; + char iblk[6], dblk[6], lsize[6], asize[6], bonus_size[6], segsize[6]; + char aux[50]; + int error; + + if (*print_header) { + (void) printf("\n Object lvl iblk dblk lsize" + " asize type\n"); + *print_header = 0; + } + + if (object == 0) { + dn = os->os->os_meta_dnode; + } else { + error = dmu_bonus_hold(os, object, FTAG, &db); + if (error) + fatal("dmu_bonus_hold(%llu) failed, errno %u", + object, error); + bonus = db->db_data; + bsize = db->db_size; + dn = ((dmu_buf_impl_t *)db)->db_dnode; + } + dmu_object_info_from_dnode(dn, &doi); + + nicenum(doi.doi_metadata_block_size, iblk); + nicenum(doi.doi_data_block_size, dblk); + nicenum(doi.doi_data_block_size * (doi.doi_max_block_offset + 1), + lsize); + nicenum(doi.doi_physical_blks << 9, asize); + nicenum(doi.doi_bonus_size, bonus_size); + + aux[0] = '\0'; + + if (doi.doi_checksum != ZIO_CHECKSUM_INHERIT || verbosity >= 6) { + (void) snprintf(aux + strlen(aux), sizeof (aux), " (K=%s)", + zio_checksum_table[doi.doi_checksum].ci_name); + } + + if (doi.doi_compress != ZIO_COMPRESS_INHERIT || verbosity >= 6) { + (void) snprintf(aux + strlen(aux), sizeof (aux), " (Z=%s)", + zio_compress_table[doi.doi_compress].ci_name); + } + + (void) printf("%10lld %3u %5s %5s %5s %5s %s%s\n", + (u_longlong_t)object, doi.doi_indirection, iblk, dblk, lsize, + asize, dmu_ot[doi.doi_type].ot_name, aux); + + if (doi.doi_bonus_type != DMU_OT_NONE && verbosity > 3) { + (void) printf("%10s %3s %5s %5s %5s %5s %s\n", + "", "", "", "", bonus_size, "bonus", + dmu_ot[doi.doi_bonus_type].ot_name); + } + + if (verbosity >= 4) { + object_viewer[doi.doi_bonus_type](os, object, bonus, bsize); + object_viewer[doi.doi_type](os, object, NULL, 0); + *print_header = 1; + } + + if (verbosity >= 5) + dump_indirect(dn); + + if (verbosity >= 5) { + /* + * Report the list of segments that comprise the object. + */ + uint64_t start = 0; + uint64_t end; + uint64_t blkfill = 1; + int minlvl = 1; + + if (dn->dn_type == DMU_OT_DNODE) { + minlvl = 0; + blkfill = DNODES_PER_BLOCK; + } + + for (;;) { + error = dnode_next_offset(dn, + 0, &start, minlvl, blkfill, 0); + if (error) + break; + end = start; + error = dnode_next_offset(dn, + DNODE_FIND_HOLE, &end, minlvl, blkfill, 0); + nicenum(end - start, segsize); + (void) printf("\t\tsegment [%016llx, %016llx)" + " size %5s\n", (u_longlong_t)start, + (u_longlong_t)end, segsize); + if (error) + break; + start = end; + } + } + + if (db != NULL) + dmu_buf_rele(db, FTAG); +} + +static char *objset_types[DMU_OST_NUMTYPES] = { + "NONE", "META", "ZPL", "ZVOL", "OTHER", "ANY" }; + +static void +dump_dir(objset_t *os) +{ + dmu_objset_stats_t dds; + uint64_t object, object_count; + uint64_t refdbytes, usedobjs, scratch; + char numbuf[8]; + char blkbuf[BP_SPRINTF_LEN]; + char osname[MAXNAMELEN]; + char *type = "UNKNOWN"; + int verbosity = dump_opt['d']; + int print_header = 1; + int i, error; + + dmu_objset_fast_stat(os, &dds); + + if (dds.dds_type < DMU_OST_NUMTYPES) + type = objset_types[dds.dds_type]; + + if (dds.dds_type == DMU_OST_META) { + dds.dds_creation_txg = TXG_INITIAL; + usedobjs = os->os->os_rootbp->blk_fill; + refdbytes = os->os->os_spa->spa_dsl_pool-> + dp_mos_dir->dd_phys->dd_used_bytes; + } else { + dmu_objset_space(os, &refdbytes, &scratch, &usedobjs, &scratch); + } + + ASSERT3U(usedobjs, ==, os->os->os_rootbp->blk_fill); + + nicenum(refdbytes, numbuf); + + if (verbosity >= 4) { + (void) strcpy(blkbuf, ", rootbp "); + sprintf_blkptr(blkbuf + strlen(blkbuf), + BP_SPRINTF_LEN - strlen(blkbuf), os->os->os_rootbp); + } else { + blkbuf[0] = '\0'; + } + + dmu_objset_name(os, osname); + + (void) printf("Dataset %s [%s], ID %llu, cr_txg %llu, " + "%s, %llu objects%s\n", + osname, type, (u_longlong_t)dmu_objset_id(os), + (u_longlong_t)dds.dds_creation_txg, + numbuf, (u_longlong_t)usedobjs, blkbuf); + + dump_intent_log(dmu_objset_zil(os)); + + if (dmu_objset_ds(os) != NULL) + dump_bplist(dmu_objset_pool(os)->dp_meta_objset, + dmu_objset_ds(os)->ds_phys->ds_deadlist_obj, "Deadlist"); + + if (verbosity < 2) + return; + + if (os->os->os_rootbp->blk_birth == 0) + return; + + if (zopt_objects != 0) { + for (i = 0; i < zopt_objects; i++) + dump_object(os, zopt_object[i], verbosity, + &print_header); + (void) printf("\n"); + return; + } + + dump_object(os, 0, verbosity, &print_header); + object_count = 1; + + object = 0; + while ((error = dmu_object_next(os, &object, B_FALSE, 0)) == 0) { + dump_object(os, object, verbosity, &print_header); + object_count++; + } + + ASSERT3U(object_count, ==, usedobjs); + + (void) printf("\n"); + + if (error != ESRCH) + fatal("dmu_object_next() = %d", error); +} + +static void +dump_uberblock(uberblock_t *ub) +{ + time_t timestamp = ub->ub_timestamp; + + (void) printf("Uberblock\n\n"); + (void) printf("\tmagic = %016llx\n", (u_longlong_t)ub->ub_magic); + (void) printf("\tversion = %llu\n", (u_longlong_t)ub->ub_version); + (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))); + if (dump_opt['u'] >= 3) { + char blkbuf[BP_SPRINTF_LEN]; + sprintf_blkptr(blkbuf, BP_SPRINTF_LEN, &ub->ub_rootbp); + (void) printf("\trootbp = %s\n", blkbuf); + } + (void) printf("\n"); +} + +static void +dump_config(const char *pool) +{ + spa_t *spa = NULL; + + mutex_enter(&spa_namespace_lock); + while ((spa = spa_next(spa)) != NULL) { + if (pool == NULL) + (void) printf("%s\n", spa_name(spa)); + if (pool == NULL || strcmp(pool, spa_name(spa)) == 0) + dump_nvlist(spa->spa_config, 4); + } + mutex_exit(&spa_namespace_lock); +} + +static void +dump_cachefile(const char *cachefile) +{ + int fd; + struct stat64 statbuf; + char *buf; + nvlist_t *config; + + if ((fd = open64(cachefile, O_RDONLY)) < 0) { + (void) printf("cannot open '%s': %s\n", cachefile, + strerror(errno)); + exit(1); + } + + if (fstat64(fd, &statbuf) != 0) { + (void) printf("failed to stat '%s': %s\n", cachefile, + strerror(errno)); + exit(1); + } + + if ((buf = malloc(statbuf.st_size)) == NULL) { + (void) fprintf(stderr, "failed to allocate %llu bytes\n", + (u_longlong_t)statbuf.st_size); + exit(1); + } + + if (read(fd, buf, statbuf.st_size) != statbuf.st_size) { + (void) fprintf(stderr, "failed to read %llu bytes\n", + (u_longlong_t)statbuf.st_size); + exit(1); + } + + (void) close(fd); + + if (nvlist_unpack(buf, statbuf.st_size, &config, 0) != 0) { + (void) fprintf(stderr, "failed to unpack nvlist\n"); + exit(1); + } + + free(buf); + + dump_nvlist(config, 0); + + nvlist_free(config); +} + +static void +dump_label(const char *dev) +{ + int fd; + vdev_label_t label; + char *buf = label.vl_vdev_phys.vp_nvlist; + size_t buflen = sizeof (label.vl_vdev_phys.vp_nvlist); + struct stat64 statbuf; + uint64_t psize; + int l; + + if ((fd = open64(dev, O_RDONLY)) < 0) { + (void) printf("cannot open '%s': %s\n", dev, strerror(errno)); + exit(1); + } + + if (fstat64(fd, &statbuf) != 0) { + (void) printf("failed to stat '%s': %s\n", dev, + strerror(errno)); + exit(1); + } + + psize = statbuf.st_size; + psize = P2ALIGN(psize, (uint64_t)sizeof (vdev_label_t)); + + for (l = 0; l < VDEV_LABELS; l++) { + + nvlist_t *config = NULL; + + (void) printf("--------------------------------------------\n"); + (void) printf("LABEL %d\n", l); + (void) printf("--------------------------------------------\n"); + + if (pread64(fd, &label, sizeof (label), + vdev_label_offset(psize, l, 0)) != sizeof (label)) { + (void) printf("failed to read label %d\n", l); + continue; + } + + if (nvlist_unpack(buf, buflen, &config, 0) != 0) { + (void) printf("failed to unpack label %d\n", l); + continue; + } + dump_nvlist(config, 4); + nvlist_free(config); + } +} + +/*ARGSUSED*/ +static int +dump_one_dir(char *dsname, void *arg) +{ + int error; + objset_t *os; + + error = dmu_objset_open(dsname, DMU_OST_ANY, + DS_MODE_USER | DS_MODE_READONLY, &os); + if (error) { + (void) printf("Could not open %s\n", dsname); + return (0); + } + dump_dir(os); + dmu_objset_close(os); + fuid_table_destroy(); + return (0); +} + +static void +zdb_leak(space_map_t *sm, uint64_t start, uint64_t size) +{ + vdev_t *vd = sm->sm_ppd; + + (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); +} + +/* ARGSUSED */ +static void +zdb_space_map_load(space_map_t *sm) +{ +} + +static void +zdb_space_map_unload(space_map_t *sm) +{ + space_map_vacate(sm, zdb_leak, sm); +} + +/* ARGSUSED */ +static void +zdb_space_map_claim(space_map_t *sm, uint64_t start, uint64_t size) +{ +} + +static space_map_ops_t zdb_space_map_ops = { + zdb_space_map_load, + zdb_space_map_unload, + NULL, /* alloc */ + zdb_space_map_claim, + NULL /* free */ +}; + +static void +zdb_leak_init(spa_t *spa) +{ + vdev_t *rvd = spa->spa_root_vdev; + + for (int c = 0; c < rvd->vdev_children; c++) { + vdev_t *vd = rvd->vdev_child[c]; + for (int m = 0; m < vd->vdev_ms_count; m++) { + metaslab_t *msp = vd->vdev_ms[m]; + mutex_enter(&msp->ms_lock); + 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; + mutex_exit(&msp->ms_lock); + } + } +} + +static void +zdb_leak_fini(spa_t *spa) +{ + vdev_t *rvd = spa->spa_root_vdev; + + for (int c = 0; c < rvd->vdev_children; c++) { + vdev_t *vd = rvd->vdev_child[c]; + for (int 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); + mutex_exit(&msp->ms_lock); + } + } +} + +/* + * Verify that the sum of the sizes of all blocks in the pool adds up + * to the SPA's sa_alloc total. + */ +typedef struct zdb_blkstats { + uint64_t zb_asize; + uint64_t zb_lsize; + uint64_t zb_psize; + uint64_t zb_count; +} zdb_blkstats_t; + +#define DMU_OT_DEFERRED DMU_OT_NONE +#define DMU_OT_TOTAL DMU_OT_NUMTYPES + +#define ZB_TOTAL DN_MAX_LEVELS + +typedef struct zdb_cb { + zdb_blkstats_t zcb_type[ZB_TOTAL + 1][DMU_OT_TOTAL + 1]; + uint64_t zcb_errors[256]; + int zcb_readfails; + int zcb_haderrors; +} zdb_cb_t; + +static void +zdb_count_block(spa_t *spa, zdb_cb_t *zcb, blkptr_t *bp, dmu_object_type_t type) +{ + for (int i = 0; i < 4; i++) { + int l = (i < 2) ? BP_GET_LEVEL(bp) : ZB_TOTAL; + int t = (i & 1) ? type : DMU_OT_TOTAL; + zdb_blkstats_t *zb = &zcb->zcb_type[l][t]; + + zb->zb_asize += BP_GET_ASIZE(bp); + zb->zb_lsize += BP_GET_LSIZE(bp); + zb->zb_psize += BP_GET_PSIZE(bp); + zb->zb_count++; + } + + if (dump_opt['S']) { + boolean_t print_sig; + + print_sig = !zdb_sig_user_data || (BP_GET_LEVEL(bp) == 0 && + BP_GET_TYPE(bp) == DMU_OT_PLAIN_FILE_CONTENTS); + + if (BP_GET_CHECKSUM(bp) < zdb_sig_cksumalg) + print_sig = B_FALSE; + + if (print_sig) { + (void) printf("%llu\t%lld\t%lld\t%s\t%s\t%s\t" + "%llx:%llx:%llx:%llx\n", + (u_longlong_t)BP_GET_LEVEL(bp), + (longlong_t)BP_GET_PSIZE(bp), + (longlong_t)BP_GET_NDVAS(bp), + dmu_ot[BP_GET_TYPE(bp)].ot_name, + zio_checksum_table[BP_GET_CHECKSUM(bp)].ci_name, + zio_compress_table[BP_GET_COMPRESS(bp)].ci_name, + (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], + (u_longlong_t)bp->blk_cksum.zc_word[3]); + } + } + + VERIFY(zio_wait(zio_claim(NULL, spa, spa_first_txg(spa), bp, + NULL, NULL, ZIO_FLAG_MUSTSUCCEED)) == 0); +} + +static int +zdb_blkptr_cb(spa_t *spa, blkptr_t *bp, const zbookmark_t *zb, + const dnode_phys_t *dnp, void *arg) +{ + zdb_cb_t *zcb = arg; + char blkbuf[BP_SPRINTF_LEN]; + + if (bp == NULL) + return (0); + + zdb_count_block(spa, zcb, bp, BP_GET_TYPE(bp)); + + if (dump_opt['c'] || dump_opt['S']) { + int ioerr, size; + void *data; + + size = BP_GET_LSIZE(bp); + data = malloc(size); + ioerr = zio_wait(zio_read(NULL, spa, bp, data, size, + NULL, NULL, ZIO_PRIORITY_ASYNC_READ, + ZIO_FLAG_CANFAIL | ZIO_FLAG_SCRUB, zb)); + free(data); + + /* We expect io errors on intent log */ + if (ioerr && BP_GET_TYPE(bp) != DMU_OT_INTENT_LOG) { + zcb->zcb_haderrors = 1; + zcb->zcb_errors[ioerr]++; + + if (dump_opt['b'] >= 2) + sprintf_blkptr(blkbuf, BP_SPRINTF_LEN, bp); + else + blkbuf[0] = '\0'; + + if (!dump_opt['S']) { + (void) printf("zdb_blkptr_cb: " + "Got error %d reading " + "<%llu, %llu, %lld, %llx> %s -- skipping\n", + ioerr, + (u_longlong_t)zb->zb_objset, + (u_longlong_t)zb->zb_object, + (u_longlong_t)zb->zb_level, + (u_longlong_t)zb->zb_blkid, + blkbuf); + } + } + } + + zcb->zcb_readfails = 0; + + if (dump_opt['b'] >= 4) { + sprintf_blkptr(blkbuf, BP_SPRINTF_LEN, bp); + (void) printf("objset %llu object %llu offset 0x%llx %s\n", + (u_longlong_t)zb->zb_objset, + (u_longlong_t)zb->zb_object, + (u_longlong_t)blkid2offset(dnp, zb->zb_level, zb->zb_blkid), + blkbuf); + } + + return (0); +} + +static int +dump_block_stats(spa_t *spa) +{ + zdb_cb_t zcb = { 0 }; + zdb_blkstats_t *zb, *tzb; + uint64_t alloc, space, logalloc; + vdev_t *rvd = spa->spa_root_vdev; + int leaks = 0; + int c, e; + + if (!dump_opt['S']) { + (void) printf("\nTraversing all blocks to %sverify" + " nothing leaked ...\n", + dump_opt['c'] ? "verify checksums and " : ""); + } + + /* + * Load all space maps as SM_ALLOC maps, then traverse the pool + * claiming each block we discover. If the pool is perfectly + * consistent, the space maps will be empty when we're done. + * Anything left over is a leak; any block we can't claim (because + * it's not part of any space map) is a double allocation, + * reference to a freed block, or an unclaimed log block. + */ + zdb_leak_init(spa); + + /* + * If there's a deferred-free bplist, process that first. + */ + if (spa->spa_sync_bplist_obj != 0) { + bplist_t *bpl = &spa->spa_sync_bplist; + blkptr_t blk; + uint64_t itor = 0; + + VERIFY(0 == bplist_open(bpl, spa->spa_meta_objset, + spa->spa_sync_bplist_obj)); + + while (bplist_iterate(bpl, &itor, &blk) == 0) { + if (dump_opt['b'] >= 4) { + char blkbuf[BP_SPRINTF_LEN]; + sprintf_blkptr(blkbuf, BP_SPRINTF_LEN, &blk); + (void) printf("[%s] %s\n", + "deferred free", blkbuf); + } + zdb_count_block(spa, &zcb, &blk, DMU_OT_DEFERRED); + } + + bplist_close(bpl); + } + + zcb.zcb_haderrors |= traverse_pool(spa, zdb_blkptr_cb, &zcb); + + if (zcb.zcb_haderrors && !dump_opt['S']) { + (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) { + (void) printf("\t%5d %llu\n", + e, (u_longlong_t)zcb.zcb_errors[e]); + } + } + } + + /* + * Report any leaked segments. + */ + zdb_leak_fini(spa); + + /* + * If we're interested in printing out the blkptr signatures, + * return now as we don't print out anything else (including + * errors and leaks). + */ + if (dump_opt['S']) + return (zcb.zcb_haderrors ? 3 : 0); + + alloc = spa_get_alloc(spa); + space = spa_get_space(spa); + + /* + * Log blocks allocated from a separate log device don't count + * as part of the normal pool space; factor them in here. + */ + logalloc = 0; + + for (c = 0; c < rvd->vdev_children; c++) + if (rvd->vdev_child[c]->vdev_islog) + logalloc += rvd->vdev_child[c]->vdev_stat.vs_alloc; + + tzb = &zcb.zcb_type[ZB_TOTAL][DMU_OT_TOTAL]; + + if (tzb->zb_asize == alloc + logalloc) { + (void) printf("\n\tNo leaks (block sum matches space" + " maps exactly)\n"); + } else { + (void) printf("block traversal size %llu != alloc %llu " + "(leaked %lld)\n", + (u_longlong_t)tzb->zb_asize, + (u_longlong_t)alloc + logalloc, + (u_longlong_t)(alloc + logalloc - tzb->zb_asize)); + leaks = 1; + } + + if (tzb->zb_count == 0) + return (2); + + (void) printf("\n"); + (void) printf("\tbp count: %10llu\n", + (u_longlong_t)tzb->zb_count); + (void) printf("\tbp logical: %10llu\t avg: %6llu\n", + (u_longlong_t)tzb->zb_lsize, + (u_longlong_t)(tzb->zb_lsize / tzb->zb_count)); + (void) printf("\tbp physical: %10llu\t avg:" + " %6llu\tcompression: %6.2f\n", + (u_longlong_t)tzb->zb_psize, + (u_longlong_t)(tzb->zb_psize / tzb->zb_count), + (double)tzb->zb_lsize / tzb->zb_psize); + (void) printf("\tbp allocated: %10llu\t avg:" + " %6llu\tcompression: %6.2f\n", + (u_longlong_t)tzb->zb_asize, + (u_longlong_t)(tzb->zb_asize / tzb->zb_count), + (double)tzb->zb_lsize / tzb->zb_asize); + (void) printf("\tSPA allocated: %10llu\tused: %5.2f%%\n", + (u_longlong_t)alloc, 100.0 * alloc / space); + + if (dump_opt['b'] >= 2) { + int l, t, level; + (void) printf("\nBlocks\tLSIZE\tPSIZE\tASIZE" + "\t avg\t comp\t%%Total\tType\n"); + + for (t = 0; t <= DMU_OT_NUMTYPES; t++) { + char csize[6], lsize[6], psize[6], asize[6], avg[6]; + char *typename; + + typename = t == DMU_OT_DEFERRED ? "deferred free" : + t == DMU_OT_TOTAL ? "Total" : dmu_ot[t].ot_name; + + 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", + "-", + "-", + "-", + "-", + "-", + "-", + "-", + typename); + continue; + } + + for (l = ZB_TOTAL - 1; l >= -1; l--) { + level = (l == -1 ? ZB_TOTAL : l); + zb = &zcb.zcb_type[level][t]; + + if (zb->zb_asize == 0) + continue; + + if (dump_opt['b'] < 3 && level != ZB_TOTAL) + continue; + + if (level == 0 && zb->zb_asize == + zcb.zcb_type[ZB_TOTAL][t].zb_asize) + continue; + + nicenum(zb->zb_count, csize); + nicenum(zb->zb_lsize, lsize); + nicenum(zb->zb_psize, psize); + nicenum(zb->zb_asize, asize); + nicenum(zb->zb_asize / zb->zb_count, avg); + + (void) printf("%6s\t%5s\t%5s\t%5s\t%5s" + "\t%5.2f\t%6.2f\t", + csize, lsize, psize, asize, avg, + (double)zb->zb_lsize / zb->zb_psize, + 100.0 * zb->zb_asize / tzb->zb_asize); + + if (level == ZB_TOTAL) + (void) printf("%s\n", typename); + else + (void) printf(" L%d %s\n", + level, typename); + } + } + } + + (void) printf("\n"); + + if (leaks) + return (2); + + if (zcb.zcb_haderrors) + return (3); + + return (0); +} + +static void +dump_zpool(spa_t *spa) +{ + dsl_pool_t *dp = spa_get_dsl(spa); + int rc = 0; + + if (dump_opt['u']) + dump_uberblock(&spa->spa_uberblock); + + if (dump_opt['d'] || dump_opt['i']) { + dump_dir(dp->dp_meta_objset); + if (dump_opt['d'] >= 3) { + dump_bplist(dp->dp_meta_objset, + spa->spa_sync_bplist_obj, "Deferred frees"); + dump_dtl(spa->spa_root_vdev, 0); + dump_metaslabs(spa); + } + (void) dmu_objset_find(spa_name(spa), dump_one_dir, NULL, + DS_FIND_SNAPSHOTS | DS_FIND_CHILDREN); + } + + if (dump_opt['b'] || dump_opt['c'] || dump_opt['S']) + rc = dump_block_stats(spa); + + if (dump_opt['s']) + show_pool_stats(spa); + + if (rc != 0) + exit(rc); +} + +#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_PHYS 0x0020 +#define ZDB_FLAG_RAW 0x0040 +#define ZDB_FLAG_PRINT_BLKPTR 0x0080 + +int flagbits[256]; + +static void +zdb_print_blkptr(blkptr_t *bp, int flags) +{ + dva_t *dva = bp->blk_dva; + int d; + + if (flags & ZDB_FLAG_BSWAP) + byteswap_uint64_array((void *)bp, sizeof (blkptr_t)); + /* + * Super-ick warning: This code is also duplicated in + * cmd/mdb/common/modules/zfs/zfs.c . Yeah, I hate code + * replication, too. + */ + for (d = 0; d < BP_GET_NDVAS(bp); d++) { + (void) printf("\tDVA[%d]: vdev_id %lld / %llx\n", d, + (longlong_t)DVA_GET_VDEV(&dva[d]), + (longlong_t)DVA_GET_OFFSET(&dva[d])); + (void) printf("\tDVA[%d]: GANG: %-5s GRID: %04llx\t" + "ASIZE: %llx\n", d, + DVA_GET_GANG(&dva[d]) ? "TRUE" : "FALSE", + (longlong_t)DVA_GET_GRID(&dva[d]), + (longlong_t)DVA_GET_ASIZE(&dva[d])); + (void) printf("\tDVA[%d]: :%llu:%llx:%llx:%s%s%s%s\n", d, + (u_longlong_t)DVA_GET_VDEV(&dva[d]), + (longlong_t)DVA_GET_OFFSET(&dva[d]), + (longlong_t)BP_GET_PSIZE(bp), + BP_SHOULD_BYTESWAP(bp) ? "e" : "", + !DVA_GET_GANG(&dva[d]) && BP_GET_LEVEL(bp) != 0 ? + "d" : "", + DVA_GET_GANG(&dva[d]) ? "g" : "", + BP_GET_COMPRESS(bp) != 0 ? "d" : ""); + } + (void) printf("\tLSIZE: %-16llx\t\tPSIZE: %llx\n", + (longlong_t)BP_GET_LSIZE(bp), (longlong_t)BP_GET_PSIZE(bp)); + (void) printf("\tENDIAN: %6s\t\t\t\t\tTYPE: %s\n", + BP_GET_BYTEORDER(bp) ? "LITTLE" : "BIG", + dmu_ot[BP_GET_TYPE(bp)].ot_name); + (void) printf("\tBIRTH: %-16llx LEVEL: %-2llu\tFILL: %llx\n", + (u_longlong_t)bp->blk_birth, (u_longlong_t)BP_GET_LEVEL(bp), + (u_longlong_t)bp->blk_fill); + (void) printf("\tCKFUNC: %-16s\t\tCOMP: %s\n", + zio_checksum_table[BP_GET_CHECKSUM(bp)].ci_name, + zio_compress_table[BP_GET_COMPRESS(bp)].ci_name); + (void) printf("\tCKSUM: %llx:%llx:%llx:%llx\n", + (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], + (u_longlong_t)bp->blk_cksum.zc_word[3]); +} + +static void +zdb_dump_indirect(blkptr_t *bp, int nbps, int flags) +{ + int i; + + for (i = 0; i < nbps; i++) + zdb_print_blkptr(&bp[i], flags); +} + +static void +zdb_dump_gbh(void *buf, int flags) +{ + zdb_dump_indirect((blkptr_t *)buf, SPA_GBH_NBLKPTRS, flags); +} + +static void +zdb_dump_block_raw(void *buf, uint64_t size, int flags) +{ + if (flags & ZDB_FLAG_BSWAP) + byteswap_uint64_array(buf, size); + (void) write(2, buf, size); +} + +static void +zdb_dump_block(char *label, void *buf, uint64_t size, int flags) +{ + uint64_t *d = (uint64_t *)buf; + int nwords = size / sizeof (uint64_t); + int do_bswap = !!(flags & ZDB_FLAG_BSWAP); + int i, j; + char *hdr, *c; + + + if (do_bswap) + hdr = " 7 6 5 4 3 2 1 0 f e d c b a 9 8"; + else + hdr = " 0 1 2 3 4 5 6 7 8 9 a b c d e f"; + + (void) printf("\n%s\n%6s %s 0123456789abcdef\n", label, "", hdr); + + for (i = 0; i < nwords; i += 2) { + (void) printf("%06llx: %016llx %016llx ", + (u_longlong_t)(i * sizeof (uint64_t)), + (u_longlong_t)(do_bswap ? BSWAP_64(d[i]) : d[i]), + (u_longlong_t)(do_bswap ? BSWAP_64(d[i + 1]) : d[i + 1])); + + c = (char *)&d[i]; + for (j = 0; j < 2 * sizeof (uint64_t); j++) + (void) printf("%c", isprint(c[j]) ? c[j] : '.'); + (void) printf("\n"); + } +} + +/* + * There are two acceptable formats: + * leaf_name - For example: c1t0d0 or /tmp/ztest.0a + * child[.child]* - For example: 0.1.1 + * + * The second form can be used to specify arbitrary vdevs anywhere + * in the heirarchy. For example, in a pool with a mirror of + * RAID-Zs, you can specify either RAID-Z vdev with 0.0 or 0.1 . + */ +static vdev_t * +zdb_vdev_lookup(vdev_t *vdev, char *path) +{ + char *s, *p, *q; + int i; + + if (vdev == NULL) + return (NULL); + + /* First, assume the x.x.x.x format */ + i = (int)strtoul(path, &s, 10); + if (s == path || (s && *s != '.' && *s != '\0')) + goto name; + if (i < 0 || i >= vdev->vdev_children) + return (NULL); + + vdev = vdev->vdev_child[i]; + if (*s == '\0') + return (vdev); + return (zdb_vdev_lookup(vdev, s+1)); + +name: + for (i = 0; i < vdev->vdev_children; i++) { + vdev_t *vc = vdev->vdev_child[i]; + + if (vc->vdev_path == NULL) { + vc = zdb_vdev_lookup(vc, path); + if (vc == NULL) + continue; + else + return (vc); + } + + p = strrchr(vc->vdev_path, '/'); + p = p ? p + 1 : vc->vdev_path; + q = &vc->vdev_path[strlen(vc->vdev_path) - 2]; + + if (strcmp(vc->vdev_path, path) == 0) + return (vc); + if (strcmp(p, path) == 0) + return (vc); + if (strcmp(q, "s0") == 0 && strncmp(p, path, q - p) == 0) + return (vc); + } + + return (NULL); +} + +/* + * Read a block from a pool and print it out. The syntax of the + * block descriptor is: + * + * pool:vdev_specifier:offset:size[:flags] + * + * pool - The name of the pool you wish to read from + * vdev_specifier - Which vdev (see comment for zdb_vdev_lookup) + * offset - offset, in hex, in bytes + * size - Amount of data to read, in hex, in bytes + * flags - A string of characters specifying options + * b: Decode a blkptr at given offset within block + * *c: Calculate and display checksums + * *d: Decompress data before dumping + * e: Byteswap data before dumping + * *g: Display data as a gang block header + * *i: Display as an indirect block + * p: Do I/O to physical offset + * r: Dump raw data to stdout + * + * * = not yet implemented + */ +static void +zdb_read_block(char *thing, spa_t **spap) +{ + spa_t *spa = *spap; + int flags = 0; + uint64_t offset = 0, size = 0, blkptr_offset = 0; + zio_t *zio; + vdev_t *vd; + void *buf; + char *s, *p, *dup, *pool, *vdev, *flagstr; + int i, error, zio_flags; + + dup = strdup(thing); + s = strtok(dup, ":"); + pool = s ? s : ""; + s = strtok(NULL, ":"); + vdev = s ? s : ""; + s = strtok(NULL, ":"); + offset = strtoull(s ? s : "", NULL, 16); + s = strtok(NULL, ":"); + size = strtoull(s ? s : "", NULL, 16); + s = strtok(NULL, ":"); + flagstr = s ? s : ""; + + s = NULL; + if (size == 0) + s = "size must not be zero"; + if (!IS_P2ALIGNED(size, DEV_BSIZE)) + s = "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); + free(dup); + return; + } + + for (s = strtok(flagstr, ":"); s; s = strtok(NULL, ":")) { + for (i = 0; flagstr[i]; i++) { + int bit = flagbits[(uchar_t)flagstr[i]]; + + if (bit == 0) { + (void) printf("***Invalid flag: %c\n", + flagstr[i]); + continue; + } + flags |= bit; + + /* If it's not something with an argument, keep going */ + if ((bit & (ZDB_FLAG_CHECKSUM | ZDB_FLAG_DECOMPRESS | + ZDB_FLAG_PRINT_BLKPTR)) == 0) + continue; + + p = &flagstr[i + 1]; + if (bit == ZDB_FLAG_PRINT_BLKPTR) + blkptr_offset = strtoull(p, &p, 16); + if (*p != ':' && *p != '\0') { + (void) printf("***Invalid flag arg: '%s'\n", s); + free(dup); + return; + } + } + } + + if (spa == NULL || strcmp(spa_name(spa), pool) != 0) { + if (spa) + spa_close(spa, (void *)zdb_read_block); + error = spa_open(pool, spap, (void *)zdb_read_block); + if (error) + fatal("Failed to open pool '%s': %s", + pool, strerror(error)); + spa = *spap; + } + + vd = zdb_vdev_lookup(spa->spa_root_vdev, vdev); + if (vd == NULL) { + (void) printf("***Invalid vdev: %s\n", vdev); + free(dup); + return; + } else { + if (vd->vdev_path) + (void) printf("Found vdev: %s\n", vd->vdev_path); + else + (void) printf("Found vdev type: %s\n", + vd->vdev_ops->vdev_op_type); + } + + buf = umem_alloc(size, UMEM_NOFAIL); + + zio_flags = ZIO_FLAG_DONT_CACHE | ZIO_FLAG_DONT_QUEUE | + ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_DONT_RETRY; + + spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); + zio = zio_root(spa, NULL, NULL, 0); + /* XXX todo - cons up a BP so RAID-Z will be happy */ + zio_nowait(zio_vdev_child_io(zio, NULL, vd, offset, buf, size, + ZIO_TYPE_READ, ZIO_PRIORITY_SYNC_READ, zio_flags, NULL, NULL)); + error = zio_wait(zio); + spa_config_exit(spa, SCL_STATE, FTAG); + + if (error) { + (void) printf("Read of %s failed, error: %d\n", thing, error); + goto out; + } + + if (flags & ZDB_FLAG_PRINT_BLKPTR) + zdb_print_blkptr((blkptr_t *)(void *) + ((uintptr_t)buf + (uintptr_t)blkptr_offset), flags); + else if (flags & ZDB_FLAG_RAW) + zdb_dump_block_raw(buf, size, flags); + else if (flags & ZDB_FLAG_INDIRECT) + zdb_dump_indirect((blkptr_t *)buf, size / sizeof (blkptr_t), + flags); + else if (flags & ZDB_FLAG_GBH) + zdb_dump_gbh(buf, flags); + else + zdb_dump_block(thing, buf, size, flags); + +out: + umem_free(buf, size); + free(dup); +} + +static boolean_t +nvlist_string_match(nvlist_t *config, char *name, char *tgt) +{ + char *s; + + if (nvlist_lookup_string(config, name, &s) != 0) + return (B_FALSE); + + return (strcmp(s, tgt) == 0); +} + +static boolean_t +nvlist_uint64_match(nvlist_t *config, char *name, uint64_t tgt) +{ + uint64_t val; + + if (nvlist_lookup_uint64(config, name, &val) != 0) + return (B_FALSE); + + return (val == tgt); +} + +static boolean_t +vdev_child_guid_match(nvlist_t *vdev, uint64_t guid) +{ + nvlist_t **child; + uint_t c, children; + + verify(nvlist_lookup_nvlist_array(vdev, ZPOOL_CONFIG_CHILDREN, + &child, &children) == 0); + for (c = 0; c < children; ++c) + if (nvlist_uint64_match(child[c], ZPOOL_CONFIG_GUID, guid)) + return (B_TRUE); + return (B_FALSE); +} + +static boolean_t +vdev_child_string_match(nvlist_t *vdev, char *tgt) +{ + nvlist_t **child; + uint_t c, children; + + verify(nvlist_lookup_nvlist_array(vdev, ZPOOL_CONFIG_CHILDREN, + &child, &children) == 0); + for (c = 0; c < children; ++c) { + if (nvlist_string_match(child[c], ZPOOL_CONFIG_PATH, tgt) || + nvlist_string_match(child[c], ZPOOL_CONFIG_DEVID, tgt)) + return (B_TRUE); + } + return (B_FALSE); +} + +static boolean_t +vdev_guid_match(nvlist_t *config, uint64_t guid) +{ + nvlist_t *nvroot; + + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &nvroot) == 0); + + return (nvlist_uint64_match(nvroot, ZPOOL_CONFIG_GUID, guid) || + vdev_child_guid_match(nvroot, guid)); +} + +static boolean_t +vdev_string_match(nvlist_t *config, char *tgt) +{ + nvlist_t *nvroot; + + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &nvroot) == 0); + + return (vdev_child_string_match(nvroot, tgt)); +} + +static boolean_t +pool_match(nvlist_t *config, char *tgt) +{ + uint64_t guid = strtoull(tgt, NULL, 0); + + if (guid != 0) { + return ( + nvlist_uint64_match(config, ZPOOL_CONFIG_POOL_GUID, guid) || + vdev_guid_match(config, guid)); + } else { + return ( + nvlist_string_match(config, ZPOOL_CONFIG_POOL_NAME, tgt) || + vdev_string_match(config, tgt)); + } +} + +static int +find_exported_zpool(char *pool_id, nvlist_t **configp, char *vdev_dir) +{ + nvlist_t *pools; + int error = ENOENT; + nvlist_t *match = NULL; + + if (vdev_dir != NULL) + pools = zpool_find_import_activeok(g_zfs, 1, &vdev_dir); + else + pools = zpool_find_import_activeok(g_zfs, 0, NULL); + + if (pools != NULL) { + nvpair_t *elem = NULL; + + while ((elem = nvlist_next_nvpair(pools, elem)) != NULL) { + verify(nvpair_value_nvlist(elem, configp) == 0); + if (pool_match(*configp, pool_id)) { + if (match != NULL) { + (void) fatal( + "More than one matching pool - " + "specify guid/devid/device path."); + } else { + match = *configp; + error = 0; + } + } + } + } + + *configp = error ? NULL : match; + + return (error); +} + +int +main(int argc, char **argv) +{ + int i, c; + struct rlimit rl = { 1024, 1024 }; + spa_t *spa; + objset_t *os = NULL; + char *endstr; + int dump_all = 1; + int verbose = 0; + int error; + int exported = 0; + char *vdev_dir = NULL; + + (void) setrlimit(RLIMIT_NOFILE, &rl); + (void) enable_extended_FILE_stdio(-1, -1); + + dprintf_setup(&argc, argv); + + while ((c = getopt(argc, argv, "udibcsvCS:U:lRep:")) != -1) { + switch (c) { + case 'u': + case 'd': + case 'i': + case 'b': + case 'c': + case 's': + case 'C': + case 'l': + case 'R': + dump_opt[c]++; + dump_all = 0; + break; + case 'v': + verbose++; + break; + case 'U': + spa_config_path = optarg; + break; + case 'e': + exported = 1; + break; + case 'p': + vdev_dir = optarg; + break; + case 'S': + dump_opt[c]++; + dump_all = 0; + zdb_sig_user_data = (strncmp(optarg, "user:", 5) == 0); + if (!zdb_sig_user_data && strncmp(optarg, "all:", 4)) + usage(); + endstr = strchr(optarg, ':') + 1; + if (strcmp(endstr, "fletcher2") == 0) + zdb_sig_cksumalg = ZIO_CHECKSUM_FLETCHER_2; + else if (strcmp(endstr, "fletcher4") == 0) + zdb_sig_cksumalg = ZIO_CHECKSUM_FLETCHER_4; + else if (strcmp(endstr, "sha256") == 0) + zdb_sig_cksumalg = ZIO_CHECKSUM_SHA256; + else if (strcmp(endstr, "all") == 0) + zdb_sig_cksumalg = ZIO_CHECKSUM_FLETCHER_2; + else + usage(); + break; + default: + usage(); + break; + } + } + + if (vdev_dir != NULL && exported == 0) { + (void) fprintf(stderr, "-p option requires use of -e\n"); + usage(); + } + + kernel_init(FREAD); + g_zfs = libzfs_init(); + ASSERT(g_zfs != NULL); + + for (c = 0; c < 256; c++) { + if (dump_all && c != 'l' && c != 'R') + dump_opt[c] = 1; + if (dump_opt[c]) + dump_opt[c] += verbose; + } + + argc -= optind; + argv += optind; + + if (argc < 1) { + if (dump_opt['C']) { + dump_cachefile(spa_config_path); + return (0); + } + usage(); + } + + if (dump_opt['l']) { + dump_label(argv[0]); + return (0); + } + + if (dump_opt['R']) { + flagbits['b'] = ZDB_FLAG_PRINT_BLKPTR; + flagbits['c'] = ZDB_FLAG_CHECKSUM; + flagbits['d'] = ZDB_FLAG_DECOMPRESS; + flagbits['e'] = ZDB_FLAG_BSWAP; + flagbits['g'] = ZDB_FLAG_GBH; + flagbits['i'] = ZDB_FLAG_INDIRECT; + flagbits['p'] = ZDB_FLAG_PHYS; + flagbits['r'] = ZDB_FLAG_RAW; + + spa = NULL; + while (argv[0]) { + zdb_read_block(argv[0], &spa); + argv++; + argc--; + } + if (spa) + spa_close(spa, (void *)zdb_read_block); + return (0); + } + + if (dump_opt['C']) + dump_config(argv[0]); + + error = 0; + if (exported) { + /* + * Check to see if the name refers to an exported zpool + */ + char *slash; + nvlist_t *exported_conf = NULL; + + if ((slash = strchr(argv[0], '/')) != NULL) + *slash = '\0'; + + error = find_exported_zpool(argv[0], &exported_conf, vdev_dir); + if (error == 0) { + nvlist_t *nvl = NULL; + + if (vdev_dir != NULL) { + if (nvlist_alloc(&nvl, NV_UNIQUE_NAME, 0) != 0) + error = ENOMEM; + else if (nvlist_add_string(nvl, + zpool_prop_to_name(ZPOOL_PROP_ALTROOT), + vdev_dir) != 0) + error = ENOMEM; + } + + if (error == 0) + error = spa_import_faulted(argv[0], + exported_conf, nvl); + + nvlist_free(nvl); + } + + if (slash != NULL) + *slash = '/'; + } + + if (error == 0) { + if (strchr(argv[0], '/') != NULL) { + error = dmu_objset_open(argv[0], DMU_OST_ANY, + DS_MODE_USER | DS_MODE_READONLY, &os); + } else { + error = spa_open(argv[0], &spa, FTAG); + } + } + + if (error) + fatal("can't open %s: %s", argv[0], strerror(error)); + + argv++; + if (--argc > 0) { + zopt_objects = argc; + zopt_object = calloc(zopt_objects, sizeof (uint64_t)); + for (i = 0; i < zopt_objects; i++) { + errno = 0; + zopt_object[i] = strtoull(argv[i], NULL, 0); + if (zopt_object[i] == 0 && errno != 0) + fatal("bad object number %s: %s", + argv[i], strerror(errno)); + } + } + + if (os != NULL) { + dump_dir(os); + dmu_objset_close(os); + } else { + dump_zpool(spa); + spa_close(spa, FTAG); + } + + fuid_table_destroy(); + + libzfs_fini(g_zfs); + kernel_fini(); + + return (0); +} diff --git a/cmd/zdb/zdb_il.c b/cmd/zdb/zdb_il.c new file mode 100644 index 000000000..02d35a050 --- /dev/null +++ b/cmd/zdb/zdb_il.c @@ -0,0 +1,363 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ +/* + * Copyright 2007 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#pragma ident "%Z%%M% %I% %E% SMI" + +/* + * Print intent log header and statistics. + */ + +#include <stdio.h> +#include <stdlib.h> +#include <ctype.h> +#include <sys/zfs_context.h> +#include <sys/spa.h> +#include <sys/dmu.h> +#include <sys/stat.h> +#include <sys/resource.h> +#include <sys/zil.h> +#include <sys/zil_impl.h> + +extern uint8_t dump_opt[256]; + +static void +print_log_bp(const blkptr_t *bp, const char *prefix) +{ + char blkbuf[BP_SPRINTF_LEN]; + + sprintf_blkptr(blkbuf, BP_SPRINTF_LEN, bp); + (void) printf("%s%s\n", prefix, blkbuf); +} + +/* ARGSUSED */ +static void +zil_prt_rec_create(zilog_t *zilog, int txtype, lr_create_t *lr) +{ + time_t crtime = lr->lr_crtime[0]; + char *name = (char *)(lr + 1); + char *link = name + strlen(name) + 1; + + if (txtype == TX_SYMLINK) + (void) printf("\t\t\t%s -> %s\n", name, link); + else + (void) printf("\t\t\t%s\n", name); + + (void) printf("\t\t\t%s", ctime(&crtime)); + (void) printf("\t\t\tdoid %llu, foid %llu, mode %llo\n", + (u_longlong_t)lr->lr_doid, (u_longlong_t)lr->lr_foid, + (longlong_t)lr->lr_mode); + (void) printf("\t\t\tuid %llu, gid %llu, gen %llu, rdev 0x%llx\n", + (u_longlong_t)lr->lr_uid, (u_longlong_t)lr->lr_gid, + (u_longlong_t)lr->lr_gen, (u_longlong_t)lr->lr_rdev); +} + +/* ARGSUSED */ +static void +zil_prt_rec_remove(zilog_t *zilog, int txtype, lr_remove_t *lr) +{ + (void) printf("\t\t\tdoid %llu, name %s\n", + (u_longlong_t)lr->lr_doid, (char *)(lr + 1)); +} + +/* ARGSUSED */ +static void +zil_prt_rec_link(zilog_t *zilog, int txtype, lr_link_t *lr) +{ + (void) printf("\t\t\tdoid %llu, link_obj %llu, name %s\n", + (u_longlong_t)lr->lr_doid, (u_longlong_t)lr->lr_link_obj, + (char *)(lr + 1)); +} + +/* ARGSUSED */ +static void +zil_prt_rec_rename(zilog_t *zilog, int txtype, lr_rename_t *lr) +{ + char *snm = (char *)(lr + 1); + char *tnm = snm + strlen(snm) + 1; + + (void) printf("\t\t\tsdoid %llu, tdoid %llu\n", + (u_longlong_t)lr->lr_sdoid, (u_longlong_t)lr->lr_tdoid); + (void) printf("\t\t\tsrc %s tgt %s\n", snm, tnm); +} + +/* ARGSUSED */ +static void +zil_prt_rec_write(zilog_t *zilog, int txtype, lr_write_t *lr) +{ + char *data, *dlimit; + blkptr_t *bp = &lr->lr_blkptr; + char buf[SPA_MAXBLOCKSIZE]; + int verbose = MAX(dump_opt['d'], dump_opt['i']); + int error; + + (void) printf("\t\t\tfoid %llu, offset 0x%llx," + " length 0x%llx, blkoff 0x%llx\n", + (u_longlong_t)lr->lr_foid, (longlong_t)lr->lr_offset, + (u_longlong_t)lr->lr_length, (u_longlong_t)lr->lr_blkoff); + + if (verbose < 5) + return; + + if (lr->lr_common.lrc_reclen == sizeof (lr_write_t)) { + (void) printf("\t\t\thas blkptr, %s\n", + bp->blk_birth >= spa_first_txg(zilog->zl_spa) ? + "will claim" : "won't claim"); + print_log_bp(bp, "\t\t\t"); + if (bp->blk_birth == 0) { + bzero(buf, sizeof (buf)); + } else { + zbookmark_t zb; + + ASSERT3U(bp->blk_cksum.zc_word[ZIL_ZC_OBJSET], ==, + dmu_objset_id(zilog->zl_os)); + + zb.zb_objset = bp->blk_cksum.zc_word[ZIL_ZC_OBJSET]; + zb.zb_object = 0; + zb.zb_level = -1; + zb.zb_blkid = bp->blk_cksum.zc_word[ZIL_ZC_SEQ]; + + error = zio_wait(zio_read(NULL, zilog->zl_spa, + bp, buf, BP_GET_LSIZE(bp), NULL, NULL, + ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL, &zb)); + if (error) + return; + } + data = buf + lr->lr_blkoff; + } else { + data = (char *)(lr + 1); + } + + dlimit = data + MIN(lr->lr_length, + (verbose < 6 ? 20 : SPA_MAXBLOCKSIZE)); + + (void) printf("\t\t\t"); + while (data < dlimit) { + if (isprint(*data)) + (void) printf("%c ", *data); + else + (void) printf("%2X", *data); + data++; + } + (void) printf("\n"); +} + +/* ARGSUSED */ +static void +zil_prt_rec_truncate(zilog_t *zilog, int txtype, lr_truncate_t *lr) +{ + (void) printf("\t\t\tfoid %llu, offset 0x%llx, length 0x%llx\n", + (u_longlong_t)lr->lr_foid, (longlong_t)lr->lr_offset, + (u_longlong_t)lr->lr_length); +} + +/* ARGSUSED */ +static void +zil_prt_rec_setattr(zilog_t *zilog, int txtype, lr_setattr_t *lr) +{ + time_t atime = (time_t)lr->lr_atime[0]; + time_t mtime = (time_t)lr->lr_mtime[0]; + + (void) printf("\t\t\tfoid %llu, mask 0x%llx\n", + (u_longlong_t)lr->lr_foid, (u_longlong_t)lr->lr_mask); + + if (lr->lr_mask & AT_MODE) { + (void) printf("\t\t\tAT_MODE %llo\n", + (longlong_t)lr->lr_mode); + } + + if (lr->lr_mask & AT_UID) { + (void) printf("\t\t\tAT_UID %llu\n", + (u_longlong_t)lr->lr_uid); + } + + if (lr->lr_mask & AT_GID) { + (void) printf("\t\t\tAT_GID %llu\n", + (u_longlong_t)lr->lr_gid); + } + + if (lr->lr_mask & AT_SIZE) { + (void) printf("\t\t\tAT_SIZE %llu\n", + (u_longlong_t)lr->lr_size); + } + + if (lr->lr_mask & AT_ATIME) { + (void) printf("\t\t\tAT_ATIME %llu.%09llu %s", + (u_longlong_t)lr->lr_atime[0], + (u_longlong_t)lr->lr_atime[1], + ctime(&atime)); + } + + if (lr->lr_mask & AT_MTIME) { + (void) printf("\t\t\tAT_MTIME %llu.%09llu %s", + (u_longlong_t)lr->lr_mtime[0], + (u_longlong_t)lr->lr_mtime[1], + ctime(&mtime)); + } +} + +/* ARGSUSED */ +static void +zil_prt_rec_acl(zilog_t *zilog, int txtype, lr_acl_t *lr) +{ + (void) printf("\t\t\tfoid %llu, aclcnt %llu\n", + (u_longlong_t)lr->lr_foid, (u_longlong_t)lr->lr_aclcnt); +} + +typedef void (*zil_prt_rec_func_t)(); +typedef struct zil_rec_info { + zil_prt_rec_func_t zri_print; + char *zri_name; + uint64_t zri_count; +} zil_rec_info_t; + +static zil_rec_info_t zil_rec_info[TX_MAX_TYPE] = { + { NULL, "Total " }, + { zil_prt_rec_create, "TX_CREATE " }, + { zil_prt_rec_create, "TX_MKDIR " }, + { zil_prt_rec_create, "TX_MKXATTR " }, + { zil_prt_rec_create, "TX_SYMLINK " }, + { zil_prt_rec_remove, "TX_REMOVE " }, + { zil_prt_rec_remove, "TX_RMDIR " }, + { zil_prt_rec_link, "TX_LINK " }, + { zil_prt_rec_rename, "TX_RENAME " }, + { zil_prt_rec_write, "TX_WRITE " }, + { zil_prt_rec_truncate, "TX_TRUNCATE " }, + { zil_prt_rec_setattr, "TX_SETATTR " }, + { zil_prt_rec_acl, "TX_ACL_V0 " }, + { zil_prt_rec_acl, "TX_ACL_ACL " }, + { zil_prt_rec_create, "TX_CREATE_ACL " }, + { zil_prt_rec_create, "TX_CREATE_ATTR " }, + { zil_prt_rec_create, "TX_CREATE_ACL_ATTR " }, + { zil_prt_rec_create, "TX_MKDIR_ACL " }, + { zil_prt_rec_create, "TX_MKDIR_ATTR " }, + { zil_prt_rec_create, "TX_MKDIR_ACL_ATTR " }, +}; + +/* ARGSUSED */ +static void +print_log_record(zilog_t *zilog, lr_t *lr, void *arg, uint64_t claim_txg) +{ + int txtype; + int verbose = MAX(dump_opt['d'], dump_opt['i']); + + /* reduce size of txtype to strip off TX_CI bit */ + txtype = lr->lrc_txtype; + + ASSERT(txtype != 0 && (uint_t)txtype < TX_MAX_TYPE); + ASSERT(lr->lrc_txg); + + (void) printf("\t\t%s%s len %6llu, txg %llu, seq %llu\n", + (lr->lrc_txtype & TX_CI) ? "CI-" : "", + zil_rec_info[txtype].zri_name, + (u_longlong_t)lr->lrc_reclen, + (u_longlong_t)lr->lrc_txg, + (u_longlong_t)lr->lrc_seq); + + if (txtype && verbose >= 3) + zil_rec_info[txtype].zri_print(zilog, txtype, lr); + + zil_rec_info[txtype].zri_count++; + zil_rec_info[0].zri_count++; +} + +/* ARGSUSED */ +static void +print_log_block(zilog_t *zilog, blkptr_t *bp, void *arg, uint64_t claim_txg) +{ + char blkbuf[BP_SPRINTF_LEN]; + int verbose = MAX(dump_opt['d'], dump_opt['i']); + char *claim; + + if (verbose <= 3) + return; + + if (verbose >= 5) { + (void) strcpy(blkbuf, ", "); + sprintf_blkptr(blkbuf + strlen(blkbuf), + BP_SPRINTF_LEN - strlen(blkbuf), bp); + } else { + blkbuf[0] = '\0'; + } + + if (claim_txg != 0) + claim = "already claimed"; + else if (bp->blk_birth >= spa_first_txg(zilog->zl_spa)) + claim = "will claim"; + else + claim = "won't claim"; + + (void) printf("\tBlock seqno %llu, %s%s\n", + (u_longlong_t)bp->blk_cksum.zc_word[ZIL_ZC_SEQ], claim, blkbuf); +} + +static void +print_log_stats(int verbose) +{ + int i, w, p10; + + if (verbose > 3) + (void) printf("\n"); + + if (zil_rec_info[0].zri_count == 0) + return; + + for (w = 1, p10 = 10; zil_rec_info[0].zri_count >= p10; p10 *= 10) + w++; + + for (i = 0; i < TX_MAX_TYPE; i++) + if (zil_rec_info[i].zri_count || verbose >= 3) + (void) printf("\t\t%s %*llu\n", + zil_rec_info[i].zri_name, w, + (u_longlong_t)zil_rec_info[i].zri_count); + (void) printf("\n"); +} + +/* ARGSUSED */ +void +dump_intent_log(zilog_t *zilog) +{ + const zil_header_t *zh = zilog->zl_header; + int verbose = MAX(dump_opt['d'], dump_opt['i']); + int i; + + if (zh->zh_log.blk_birth == 0 || verbose < 2) + return; + + (void) printf("\n ZIL header: claim_txg %llu, seq %llu\n", + (u_longlong_t)zh->zh_claim_txg, (u_longlong_t)zh->zh_replay_seq); + + if (verbose >= 4) + print_log_bp(&zh->zh_log, "\n\tfirst block: "); + + for (i = 0; i < TX_MAX_TYPE; i++) + zil_rec_info[i].zri_count = 0; + + if (verbose >= 2) { + (void) printf("\n"); + (void) zil_parse(zilog, print_log_block, print_log_record, NULL, + zh->zh_claim_txg); + print_log_stats(verbose); + } +} diff --git a/cmd/zdump/zdump.c b/cmd/zdump/zdump.c new file mode 100644 index 000000000..45be5589d --- /dev/null +++ b/cmd/zdump/zdump.c @@ -0,0 +1,664 @@ +/* + * Copyright 2006 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#pragma ident "%Z%%M% %I% %E% SMI" + +/* + * zdump 7.24 + * Taken from elsie.nci.nih.gov to replace the existing Solaris zdump, + * which was based on an earlier version of the elsie code. + * + * For zdump 7.24, the following changes were made to the elsie code: + * locale/textdomain/messages to match existing Solaris style. + * Solaris verbose mode is documented to display the current time first. + * cstyle cleaned code. + * removed old locale/textdomain code. + */ + +static char elsieid[] = "@(#)zdump.c 7.74"; + +/* + * This code has been made independent of the rest of the time + * conversion package to increase confidence in the verification it provides. + * You can use this code to help in verifying other implementations. + */ + +#include "stdio.h" /* for stdout, stderr, perror */ +#include "string.h" /* for strcpy */ +#include "sys/types.h" /* for time_t */ +#include "time.h" /* for struct tm */ +#include "stdlib.h" /* for exit, malloc, atoi */ +#include "locale.h" /* for setlocale, textdomain */ +#include "libintl.h" +#include <ctype.h> +#include "tzfile.h" /* for defines */ +#include <limits.h> + +#ifndef ZDUMP_LO_YEAR +#define ZDUMP_LO_YEAR (-500) +#endif /* !defined ZDUMP_LO_YEAR */ + +#ifndef ZDUMP_HI_YEAR +#define ZDUMP_HI_YEAR 2500 +#endif /* !defined ZDUMP_HI_YEAR */ + +#ifndef MAX_STRING_LENGTH +#define MAX_STRING_LENGTH 1024 +#endif /* !defined MAX_STRING_LENGTH */ + +#ifndef TRUE +#define TRUE 1 +#endif /* !defined TRUE */ + +#ifndef FALSE +#define FALSE 0 +#endif /* !defined FALSE */ + +#ifndef isleap_sum +/* + * See tzfile.h for details on isleap_sum. + */ +#define isleap_sum(a, b) isleap((a) % 400 + (b) % 400) +#endif /* !defined isleap_sum */ + +#ifndef SECSPERDAY +#define SECSPERDAY ((long)SECSPERHOUR * HOURSPERDAY) +#endif +#define SECSPERNYEAR (SECSPERDAY * DAYSPERNYEAR) +#define SECSPERLYEAR (SECSPERNYEAR + SECSPERDAY) + +#ifndef GNUC_or_lint +#ifdef lint +#define GNUC_or_lint +#else /* !defined lint */ +#ifdef __GNUC__ +#define GNUC_or_lint +#endif /* defined __GNUC__ */ +#endif /* !defined lint */ +#endif /* !defined GNUC_or_lint */ + +#ifndef INITIALIZE +#ifdef GNUC_or_lint +#define INITIALIZE(x) ((x) = 0) +#else /* !defined GNUC_or_lint */ +#define INITIALIZE(x) +#endif /* !defined GNUC_or_lint */ +#endif /* !defined INITIALIZE */ + +static time_t absolute_min_time; +static time_t absolute_max_time; +static size_t longest; +static char *progname; +static int warned; + +static char *abbr(struct tm *); +static void abbrok(const char *, const char *); +static long delta(struct tm *, struct tm *); +static void dumptime(const struct tm *); +static time_t hunt(char *, time_t, time_t); +static void setabsolutes(void); +static void show(char *, time_t, int); +static void usage(void); +static const char *tformat(void); +static time_t yeartot(long y); + +#ifndef TYPECHECK +#define my_localtime localtime +#else /* !defined TYPECHECK */ +static struct tm * +my_localtime(tp) +time_t *tp; +{ + register struct tm *tmp; + + tmp = localtime(tp); + if (tp != NULL && tmp != NULL) { + struct tm tm; + register time_t t; + + tm = *tmp; + t = mktime(&tm); + if (t - *tp >= 1 || *tp - t >= 1) { + (void) fflush(stdout); + (void) fprintf(stderr, "\n%s: ", progname); + (void) fprintf(stderr, tformat(), *tp); + (void) fprintf(stderr, " ->"); + (void) fprintf(stderr, " year=%d", tmp->tm_year); + (void) fprintf(stderr, " mon=%d", tmp->tm_mon); + (void) fprintf(stderr, " mday=%d", tmp->tm_mday); + (void) fprintf(stderr, " hour=%d", tmp->tm_hour); + (void) fprintf(stderr, " min=%d", tmp->tm_min); + (void) fprintf(stderr, " sec=%d", tmp->tm_sec); + (void) fprintf(stderr, " isdst=%d", tmp->tm_isdst); + (void) fprintf(stderr, " -> "); + (void) fprintf(stderr, tformat(), t); + (void) fprintf(stderr, "\n"); + } + } + return (tmp); +} +#endif /* !defined TYPECHECK */ + +static void +abbrok(abbrp, zone) +const char * const abbrp; +const char * const zone; +{ + register const char *cp; + int error = 0; + + if (warned) + return; + cp = abbrp; + while (isascii(*cp) && isalpha((unsigned char)*cp)) + ++cp; + (void) fflush(stdout); + if (cp - abbrp == 0) { + /* + * TRANSLATION_NOTE + * The first %s prints for the program name (zdump), + * the second %s prints the timezone name, the third + * %s prints the timezone abbreviation (tzname[0] or + * tzname[1]). + */ + (void) fprintf(stderr, gettext("%s: warning: zone \"%s\" " + "abbreviation \"%s\" lacks alphabetic at start\n"), + progname, zone, abbrp); + error = 1; + } else if (cp - abbrp < 3) { + (void) fprintf(stderr, gettext("%s: warning: zone \"%s\" " + "abbreviation \"%s\" has fewer than 3 alphabetics\n"), + progname, zone, abbrp); + error = 1; + } else if (cp - abbrp > 6) { + (void) fprintf(stderr, gettext("%s: warning: zone \"%s\" " + "abbreviation \"%s\" has more than 6 alphabetics\n"), + progname, zone, abbrp); + error = 1; + } + if (error == 0 && (*cp == '+' || *cp == '-')) { + ++cp; + if (isascii(*cp) && isdigit((unsigned char)*cp)) + if (*cp++ == '1' && *cp >= '0' && *cp <= '4') + ++cp; + if (*cp != '\0') { + (void) fprintf(stderr, gettext("%s: warning: " + "zone \"%s\" abbreviation \"%s\" differs from " + "POSIX standard\n"), progname, zone, abbrp); + error = 1; + } + } + if (error) + warned = TRUE; +} + +int +main(argc, argv) +int argc; +char *argv[]; +{ + register int i; + register int c; + register int vflag; + register char *cutarg; + register long cutloyear = ZDUMP_LO_YEAR; + register long cuthiyear = ZDUMP_HI_YEAR; + register time_t cutlotime; + register time_t cuthitime; + time_t now; + time_t t; + time_t newt; + struct tm tm; + struct tm newtm; + register struct tm *tmp; + register struct tm *newtmp; + + INITIALIZE(cutlotime); + INITIALIZE(cuthitime); + + (void) setlocale(LC_ALL, ""); +#if !defined(TEXT_DOMAIN) /* Should be defined by cc -D */ +#define TEXT_DOMAIN "SYS_TEST" /* Use this only if it weren't */ +#endif + (void) textdomain(TEXT_DOMAIN); + + progname = argv[0]; + for (i = 1; i < argc; ++i) + if (strcmp(argv[i], "--version") == 0) { + (void) printf("%s\n", elsieid); + exit(EXIT_SUCCESS); + } + vflag = 0; + cutarg = NULL; + while ((c = getopt(argc, argv, "c:v")) == 'c' || c == 'v') + if (c == 'v') + vflag = 1; + else cutarg = optarg; + if (c != EOF || + (optind == argc - 1 && strcmp(argv[optind], "=") == 0)) { + usage(); + /* NOTREACHED */ + } + if (vflag) { + if (cutarg != NULL) { + long lo; + long hi; + char dummy; + + if (sscanf(cutarg, "%ld%c", &hi, &dummy) == 1) { + cuthiyear = hi; + } else if (sscanf(cutarg, "%ld,%ld%c", + &lo, &hi, &dummy) == 2) { + cutloyear = lo; + cuthiyear = hi; + } else { +(void) fprintf(stderr, gettext("%s: wild -c argument %s\n"), + progname, cutarg); + exit(EXIT_FAILURE); + } + } + setabsolutes(); + cutlotime = yeartot(cutloyear); + cuthitime = yeartot(cuthiyear); + } + (void) time(&now); + longest = 0; + for (i = optind; i < argc; ++i) + if (strlen(argv[i]) > longest) + longest = strlen(argv[i]); + + for (i = optind; i < argc; ++i) { + static char buf[MAX_STRING_LENGTH]; + static char *tzp = NULL; + + (void) unsetenv("TZ"); + if (tzp != NULL) + free(tzp); + if ((tzp = malloc(3 + strlen(argv[i]) + 1)) == NULL) { + perror(progname); + exit(EXIT_FAILURE); + } + (void) strcpy(tzp, "TZ="); + (void) strcat(tzp, argv[i]); + if (putenv(tzp) != 0) { + perror(progname); + exit(EXIT_FAILURE); + } + if (!vflag) { + show(argv[i], now, FALSE); + continue; + } + +#if defined(sun) + /* + * We show the current time first, probably because we froze + * the behavior of zdump some time ago and then it got + * changed. + */ + show(argv[i], now, TRUE); +#endif + warned = FALSE; + t = absolute_min_time; + show(argv[i], t, TRUE); + t += SECSPERHOUR * HOURSPERDAY; + show(argv[i], t, TRUE); + if (t < cutlotime) + t = cutlotime; + tmp = my_localtime(&t); + if (tmp != NULL) { + tm = *tmp; + (void) strncpy(buf, abbr(&tm), sizeof (buf) - 1); + } + for (;;) { + if (t >= cuthitime) + break; + /* check if newt will overrun maximum time_t value */ + if (t > LONG_MAX - (SECSPERHOUR * 12)) + break; + newt = t + SECSPERHOUR * 12; + if (newt >= cuthitime) + break; + newtmp = localtime(&newt); + if (newtmp != NULL) + newtm = *newtmp; + if ((tmp == NULL || newtmp == NULL) ? (tmp != newtmp) : + (delta(&newtm, &tm) != (newt - t) || + newtm.tm_isdst != tm.tm_isdst || + strcmp(abbr(&newtm), buf) != 0)) { + newt = hunt(argv[i], t, newt); + newtmp = localtime(&newt); + if (newtmp != NULL) { + newtm = *newtmp; + (void) strncpy(buf, + abbr(&newtm), + sizeof (buf) - 1); + } + } + t = newt; + tm = newtm; + tmp = newtmp; + } + t = absolute_max_time; +#if defined(sun) + show(argv[i], t, TRUE); + t -= SECSPERHOUR * HOURSPERDAY; + show(argv[i], t, TRUE); +#else /* !defined(sun) */ + t -= SECSPERHOUR * HOURSPERDAY; + show(argv[i], t, TRUE); + t += SECSPERHOUR * HOURSPERDAY; + show(argv[i], t, TRUE); +#endif /* !defined(sun) */ + } + if (fflush(stdout) || ferror(stdout)) { + (void) fprintf(stderr, "%s: ", progname); + (void) perror(gettext("Error writing standard output")); + exit(EXIT_FAILURE); + } + return (EXIT_SUCCESS); +} + +static void +setabsolutes() +{ +#if defined(sun) + absolute_min_time = LONG_MIN; + absolute_max_time = LONG_MAX; +#else + if (0.5 == (time_t)0.5) { + /* + * time_t is floating. + */ + if (sizeof (time_t) == sizeof (float)) { + absolute_min_time = (time_t)-FLT_MAX; + absolute_max_time = (time_t)FLT_MAX; + } else if (sizeof (time_t) == sizeof (double)) { + absolute_min_time = (time_t)-DBL_MAX; + absolute_max_time = (time_t)DBL_MAX; + } else { + (void) fprintf(stderr, gettext("%s: use of -v on " + "system with floating time_t other than float " + "or double\n"), progname); + exit(EXIT_FAILURE); + } + } else + /*CONSTANTCONDITION*/ + if (0 > (time_t)-1) { + /* + * time_t is signed. + */ + register time_t hibit; + + for (hibit = 1; (hibit * 2) != 0; hibit *= 2) + continue; + absolute_min_time = hibit; + absolute_max_time = -(hibit + 1); + } else { + /* + * time_t is unsigned. + */ + absolute_min_time = 0; + absolute_max_time = absolute_min_time - 1; + } +#endif +} + +static time_t +yeartot(y) +const long y; +{ + register long myy; + register long seconds; + register time_t t; + + myy = EPOCH_YEAR; + t = 0; + while (myy != y) { + if (myy < y) { + seconds = isleap(myy) ? SECSPERLYEAR : SECSPERNYEAR; + ++myy; + if (t > absolute_max_time - seconds) { + t = absolute_max_time; + break; + } + t += seconds; + } else { + --myy; + seconds = isleap(myy) ? SECSPERLYEAR : SECSPERNYEAR; + if (t < absolute_min_time + seconds) { + t = absolute_min_time; + break; + } + t -= seconds; + } + } + return (t); +} + +static time_t +hunt(name, lot, hit) +char *name; +time_t lot; +time_t hit; +{ + time_t t; + long diff; + struct tm lotm; + register struct tm *lotmp; + struct tm tm; + register struct tm *tmp; + char loab[MAX_STRING_LENGTH]; + + lotmp = my_localtime(&lot); + if (lotmp != NULL) { + lotm = *lotmp; + (void) strncpy(loab, abbr(&lotm), sizeof (loab) - 1); + } + for (;;) { + diff = (long)(hit - lot); + if (diff < 2) + break; + t = lot; + t += diff / 2; + if (t <= lot) + ++t; + else if (t >= hit) + --t; + tmp = my_localtime(&t); + if (tmp != NULL) + tm = *tmp; + if ((lotmp == NULL || tmp == NULL) ? (lotmp == tmp) : + (delta(&tm, &lotm) == (t - lot) && + tm.tm_isdst == lotm.tm_isdst && + strcmp(abbr(&tm), loab) == 0)) { + lot = t; + lotm = tm; + lotmp = tmp; + } else hit = t; + } + show(name, lot, TRUE); + show(name, hit, TRUE); + return (hit); +} + +/* + * Thanks to Paul Eggert for logic used in delta. + */ + +static long +delta(newp, oldp) +struct tm *newp; +struct tm *oldp; +{ + register long result; + register int tmy; + + if (newp->tm_year < oldp->tm_year) + return (-delta(oldp, newp)); + result = 0; + for (tmy = oldp->tm_year; tmy < newp->tm_year; ++tmy) + result += DAYSPERNYEAR + isleap_sum(tmy, TM_YEAR_BASE); + result += newp->tm_yday - oldp->tm_yday; + result *= HOURSPERDAY; + result += newp->tm_hour - oldp->tm_hour; + result *= MINSPERHOUR; + result += newp->tm_min - oldp->tm_min; + result *= SECSPERMIN; + result += newp->tm_sec - oldp->tm_sec; + return (result); +} + +static void +show(zone, t, v) +char *zone; +time_t t; +int v; +{ + register struct tm *tmp; + + (void) printf("%-*s ", (int)longest, zone); + if (v) { + tmp = gmtime(&t); + if (tmp == NULL) { + (void) printf(tformat(), t); + } else { + dumptime(tmp); + (void) printf(" UTC"); + } + (void) printf(" = "); + } + tmp = my_localtime(&t); + dumptime(tmp); + if (tmp != NULL) { + if (*abbr(tmp) != '\0') + (void) printf(" %s", abbr(tmp)); + if (v) { + (void) printf(" isdst=%d", tmp->tm_isdst); +#ifdef TM_GMTOFF + (void) printf(" gmtoff=%ld", tmp->TM_GMTOFF); +#endif /* defined TM_GMTOFF */ + } + } + (void) printf("\n"); + if (tmp != NULL && *abbr(tmp) != '\0') + abbrok(abbr(tmp), zone); +} + +static char * +abbr(tmp) +struct tm *tmp; +{ + register char *result; + static char nada; + + if (tmp->tm_isdst != 0 && tmp->tm_isdst != 1) + return (&nada); + result = tzname[tmp->tm_isdst]; + return ((result == NULL) ? &nada : result); +} + +/* + * The code below can fail on certain theoretical systems; + * it works on all known real-world systems as of 2004-12-30. + */ + +static const char * +tformat() +{ +#if defined(sun) + /* time_t is signed long */ + return ("%ld"); +#else + /*CONSTANTCONDITION*/ + if (0.5 == (time_t)0.5) { /* floating */ + /*CONSTANTCONDITION*/ + if (sizeof (time_t) > sizeof (double)) + return ("%Lg"); + return ("%g"); + } + /*CONSTANTCONDITION*/ + if (0 > (time_t)-1) { /* signed */ + /*CONSTANTCONDITION*/ + if (sizeof (time_t) > sizeof (long)) + return ("%lld"); + /*CONSTANTCONDITION*/ + if (sizeof (time_t) > sizeof (int)) + return ("%ld"); + return ("%d"); + } + /*CONSTANTCONDITION*/ + if (sizeof (time_t) > sizeof (unsigned long)) + return ("%llu"); + /*CONSTANTCONDITION*/ + if (sizeof (time_t) > sizeof (unsigned int)) + return ("%lu"); + return ("%u"); +#endif +} + +static void +dumptime(timeptr) +register const struct tm *timeptr; +{ + static const char wday_name[][3] = { + "Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat" + }; + static const char mon_name[][3] = { + "Jan", "Feb", "Mar", "Apr", "May", "Jun", + "Jul", "Aug", "Sep", "Oct", "Nov", "Dec" + }; + register const char *wn; + register const char *mn; + register int lead; + register int trail; + + if (timeptr == NULL) { + (void) printf("NULL"); + return; + } + /* + * The packaged versions of localtime and gmtime never put out-of-range + * values in tm_wday or tm_mon, but since this code might be compiled + * with other (perhaps experimental) versions, paranoia is in order. + */ + if (timeptr->tm_wday < 0 || timeptr->tm_wday >= + (int)(sizeof (wday_name) / sizeof (wday_name[0]))) + wn = "???"; + else wn = wday_name[timeptr->tm_wday]; + if (timeptr->tm_mon < 0 || timeptr->tm_mon >= + (int)(sizeof (mon_name) / sizeof (mon_name[0]))) + mn = "???"; + else mn = mon_name[timeptr->tm_mon]; + (void) printf("%.3s %.3s%3d %.2d:%.2d:%.2d ", + wn, mn, + timeptr->tm_mday, timeptr->tm_hour, + timeptr->tm_min, timeptr->tm_sec); +#define DIVISOR 10 + trail = timeptr->tm_year % DIVISOR + TM_YEAR_BASE % DIVISOR; + lead = timeptr->tm_year / DIVISOR + TM_YEAR_BASE / DIVISOR + + trail / DIVISOR; + trail %= DIVISOR; + if (trail < 0 && lead > 0) { + trail += DIVISOR; + --lead; + } else if (lead < 0 && trail > 0) { + trail -= DIVISOR; + ++lead; + } + if (lead == 0) + (void) printf("%d", trail); + else + (void) printf("%d%d", lead, ((trail < 0) ? -trail : trail)); +} + +static void +usage() +{ + (void) fprintf(stderr, gettext( + "%s: [ --version ] [ -v ] [ -c [loyear,]hiyear ] zonename ...\n"), + progname); + exit(EXIT_FAILURE); + /* NOTREACHED */ +} diff --git a/cmd/zfs/zfs_iter.c b/cmd/zfs/zfs_iter.c new file mode 100644 index 000000000..a22370a02 --- /dev/null +++ b/cmd/zfs/zfs_iter.c @@ -0,0 +1,420 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ +/* + * Copyright 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#include <libintl.h> +#include <libuutil.h> +#include <stddef.h> +#include <stdio.h> +#include <stdlib.h> +#include <strings.h> + +#include <libzfs.h> + +#include "zfs_util.h" +#include "zfs_iter.h" + +/* + * This is a private interface used to gather up all the datasets specified on + * the command line so that we can iterate over them in order. + * + * First, we iterate over all filesystems, gathering them together into an + * AVL tree. We report errors for any explicitly specified datasets + * that we couldn't open. + * + * When finished, we have an AVL tree of ZFS handles. We go through and execute + * the provided callback for each one, passing whatever data the user supplied. + */ + +typedef struct zfs_node { + zfs_handle_t *zn_handle; + uu_avl_node_t zn_avlnode; +} zfs_node_t; + +typedef struct callback_data { + uu_avl_t *cb_avl; + int cb_flags; + zfs_type_t cb_types; + zfs_sort_column_t *cb_sortcol; + zprop_list_t **cb_proplist; +} callback_data_t; + +uu_avl_pool_t *avl_pool; + +/* + * Include snaps if they were requested or if this a zfs list where types + * were not specified and the "listsnapshots" property is set on this pool. + */ +static int +zfs_include_snapshots(zfs_handle_t *zhp, callback_data_t *cb) +{ + zpool_handle_t *zph; + + if ((cb->cb_flags & ZFS_ITER_PROP_LISTSNAPS) == 0) + return (cb->cb_types & ZFS_TYPE_SNAPSHOT); + + zph = zfs_get_pool_handle(zhp); + return (zpool_get_prop_int(zph, ZPOOL_PROP_LISTSNAPS, NULL)); +} + +/* + * Called for each dataset. If the object is of an appropriate type, + * add it to the avl tree and recurse over any children as necessary. + */ +static int +zfs_callback(zfs_handle_t *zhp, void *data) +{ + callback_data_t *cb = data; + int dontclose = 0; + int include_snaps = zfs_include_snapshots(zhp, cb); + + if ((zfs_get_type(zhp) & cb->cb_types) || + ((zfs_get_type(zhp) == ZFS_TYPE_SNAPSHOT) && include_snaps)) { + uu_avl_index_t idx; + zfs_node_t *node = safe_malloc(sizeof (zfs_node_t)); + + node->zn_handle = zhp; + uu_avl_node_init(node, &node->zn_avlnode, avl_pool); + if (uu_avl_find(cb->cb_avl, node, cb->cb_sortcol, + &idx) == NULL) { + if (cb->cb_proplist && + zfs_expand_proplist(zhp, cb->cb_proplist) != 0) { + free(node); + return (-1); + } + uu_avl_insert(cb->cb_avl, node, idx); + dontclose = 1; + } else { + free(node); + } + } + + /* + * Recurse if necessary. + */ + if (cb->cb_flags & ZFS_ITER_RECURSE) { + if (zfs_get_type(zhp) == ZFS_TYPE_FILESYSTEM) + (void) zfs_iter_filesystems(zhp, zfs_callback, data); + if ((zfs_get_type(zhp) != ZFS_TYPE_SNAPSHOT) && include_snaps) + (void) zfs_iter_snapshots(zhp, zfs_callback, data); + } + + if (!dontclose) + zfs_close(zhp); + + return (0); +} + +int +zfs_add_sort_column(zfs_sort_column_t **sc, const char *name, + boolean_t reverse) +{ + zfs_sort_column_t *col; + zfs_prop_t prop; + + if ((prop = zfs_name_to_prop(name)) == ZPROP_INVAL && + !zfs_prop_user(name)) + return (-1); + + col = safe_malloc(sizeof (zfs_sort_column_t)); + + col->sc_prop = prop; + col->sc_reverse = reverse; + if (prop == ZPROP_INVAL) { + col->sc_user_prop = safe_malloc(strlen(name) + 1); + (void) strcpy(col->sc_user_prop, name); + } + + if (*sc == NULL) { + col->sc_last = col; + *sc = col; + } else { + (*sc)->sc_last->sc_next = col; + (*sc)->sc_last = col; + } + + return (0); +} + +void +zfs_free_sort_columns(zfs_sort_column_t *sc) +{ + zfs_sort_column_t *col; + + while (sc != NULL) { + col = sc->sc_next; + free(sc->sc_user_prop); + free(sc); + sc = col; + } +} + +/* ARGSUSED */ +static int +zfs_compare(const void *larg, const void *rarg, void *unused) +{ + zfs_handle_t *l = ((zfs_node_t *)larg)->zn_handle; + zfs_handle_t *r = ((zfs_node_t *)rarg)->zn_handle; + const char *lname = zfs_get_name(l); + const char *rname = zfs_get_name(r); + char *lat, *rat; + uint64_t lcreate, rcreate; + int ret; + + lat = (char *)strchr(lname, '@'); + rat = (char *)strchr(rname, '@'); + + if (lat != NULL) + *lat = '\0'; + if (rat != NULL) + *rat = '\0'; + + ret = strcmp(lname, rname); + if (ret == 0) { + /* + * If we're comparing a dataset to one of its snapshots, we + * always make the full dataset first. + */ + if (lat == NULL) { + ret = -1; + } else if (rat == NULL) { + ret = 1; + } else { + /* + * If we have two snapshots from the same dataset, then + * we want to sort them according to creation time. We + * use the hidden CREATETXG property to get an absolute + * ordering of snapshots. + */ + lcreate = zfs_prop_get_int(l, ZFS_PROP_CREATETXG); + rcreate = zfs_prop_get_int(r, ZFS_PROP_CREATETXG); + + if (lcreate < rcreate) + ret = -1; + else if (lcreate > rcreate) + ret = 1; + } + } + + if (lat != NULL) + *lat = '@'; + if (rat != NULL) + *rat = '@'; + + return (ret); +} + +/* + * Sort datasets by specified columns. + * + * o Numeric types sort in ascending order. + * o String types sort in alphabetical order. + * o Types inappropriate for a row sort that row to the literal + * bottom, regardless of the specified ordering. + * + * If no sort columns are specified, or two datasets compare equally + * across all specified columns, they are sorted alphabetically by name + * with snapshots grouped under their parents. + */ +static int +zfs_sort(const void *larg, const void *rarg, void *data) +{ + zfs_handle_t *l = ((zfs_node_t *)larg)->zn_handle; + zfs_handle_t *r = ((zfs_node_t *)rarg)->zn_handle; + zfs_sort_column_t *sc = (zfs_sort_column_t *)data; + zfs_sort_column_t *psc; + + for (psc = sc; psc != NULL; psc = psc->sc_next) { + char lbuf[ZFS_MAXPROPLEN], rbuf[ZFS_MAXPROPLEN]; + char *lstr, *rstr; + uint64_t lnum, rnum; + boolean_t lvalid, rvalid; + int ret = 0; + + /* + * We group the checks below the generic code. If 'lstr' and + * 'rstr' are non-NULL, then we do a string based comparison. + * Otherwise, we compare 'lnum' and 'rnum'. + */ + lstr = rstr = NULL; + if (psc->sc_prop == ZPROP_INVAL) { + nvlist_t *luser, *ruser; + nvlist_t *lval, *rval; + + luser = zfs_get_user_props(l); + ruser = zfs_get_user_props(r); + + lvalid = (nvlist_lookup_nvlist(luser, + psc->sc_user_prop, &lval) == 0); + rvalid = (nvlist_lookup_nvlist(ruser, + psc->sc_user_prop, &rval) == 0); + + if (lvalid) + verify(nvlist_lookup_string(lval, + ZPROP_VALUE, &lstr) == 0); + if (rvalid) + verify(nvlist_lookup_string(rval, + ZPROP_VALUE, &rstr) == 0); + + } else if (zfs_prop_is_string(psc->sc_prop)) { + lvalid = (zfs_prop_get(l, psc->sc_prop, lbuf, + sizeof (lbuf), NULL, NULL, 0, B_TRUE) == 0); + rvalid = (zfs_prop_get(r, psc->sc_prop, rbuf, + sizeof (rbuf), NULL, NULL, 0, B_TRUE) == 0); + + lstr = lbuf; + rstr = rbuf; + } else { + lvalid = zfs_prop_valid_for_type(psc->sc_prop, + zfs_get_type(l)); + rvalid = zfs_prop_valid_for_type(psc->sc_prop, + zfs_get_type(r)); + + if (lvalid) + (void) zfs_prop_get_numeric(l, psc->sc_prop, + &lnum, NULL, NULL, 0); + if (rvalid) + (void) zfs_prop_get_numeric(r, psc->sc_prop, + &rnum, NULL, NULL, 0); + } + + if (!lvalid && !rvalid) + continue; + else if (!lvalid) + return (1); + else if (!rvalid) + return (-1); + + if (lstr) + ret = strcmp(lstr, rstr); + else if (lnum < rnum) + ret = -1; + else if (lnum > rnum) + ret = 1; + + if (ret != 0) { + if (psc->sc_reverse == B_TRUE) + ret = (ret < 0) ? 1 : -1; + return (ret); + } + } + + return (zfs_compare(larg, rarg, NULL)); +} + +int +zfs_for_each(int argc, char **argv, int flags, zfs_type_t types, + zfs_sort_column_t *sortcol, zprop_list_t **proplist, + zfs_iter_f callback, void *data) +{ + callback_data_t cb; + int ret = 0; + zfs_node_t *node; + uu_avl_walk_t *walk; + + avl_pool = uu_avl_pool_create("zfs_pool", sizeof (zfs_node_t), + offsetof(zfs_node_t, zn_avlnode), zfs_sort, UU_DEFAULT); + + if (avl_pool == NULL) { + (void) fprintf(stderr, + gettext("internal error: out of memory\n")); + exit(1); + } + + cb.cb_sortcol = sortcol; + cb.cb_flags = flags; + cb.cb_proplist = proplist; + cb.cb_types = types; + if ((cb.cb_avl = uu_avl_create(avl_pool, NULL, UU_DEFAULT)) == NULL) { + (void) fprintf(stderr, + gettext("internal error: out of memory\n")); + exit(1); + } + + if (argc == 0) { + /* + * If given no arguments, iterate over all datasets. + */ + cb.cb_flags |= ZFS_ITER_RECURSE; + ret = zfs_iter_root(g_zfs, zfs_callback, &cb); + } else { + int i; + zfs_handle_t *zhp; + zfs_type_t argtype; + + /* + * If we're recursive, then we always allow filesystems as + * arguments. If we also are interested in snapshots, then we + * can take volumes as well. + */ + argtype = types; + if (flags & ZFS_ITER_RECURSE) { + argtype |= ZFS_TYPE_FILESYSTEM; + if (types & ZFS_TYPE_SNAPSHOT) + argtype |= ZFS_TYPE_VOLUME; + } + + for (i = 0; i < argc; i++) { + if (flags & ZFS_ITER_ARGS_CAN_BE_PATHS) { + zhp = zfs_path_to_zhandle(g_zfs, argv[i], + argtype); + } else { + zhp = zfs_open(g_zfs, argv[i], argtype); + } + if (zhp != NULL) + ret |= zfs_callback(zhp, &cb); + else + ret = 1; + } + } + + /* + * At this point we've got our AVL tree full of zfs handles, so iterate + * over each one and execute the real user callback. + */ + for (node = uu_avl_first(cb.cb_avl); node != NULL; + node = uu_avl_next(cb.cb_avl, node)) + ret |= callback(node->zn_handle, data); + + /* + * Finally, clean up the AVL tree. + */ + if ((walk = uu_avl_walk_start(cb.cb_avl, UU_WALK_ROBUST)) == NULL) { + (void) fprintf(stderr, + gettext("internal error: out of memory")); + exit(1); + } + + while ((node = uu_avl_walk_next(walk)) != NULL) { + uu_avl_remove(cb.cb_avl, node); + zfs_close(node->zn_handle); + free(node); + } + + uu_avl_walk_end(walk); + uu_avl_destroy(cb.cb_avl); + uu_avl_pool_destroy(avl_pool); + + return (ret); +} diff --git a/cmd/zfs/zfs_iter.h b/cmd/zfs/zfs_iter.h new file mode 100644 index 000000000..76a11085a --- /dev/null +++ b/cmd/zfs/zfs_iter.h @@ -0,0 +1,54 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ +/* + * Copyright 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#ifndef ZFS_ITER_H +#define ZFS_ITER_H + +#ifdef __cplusplus +extern "C" { +#endif + +typedef struct zfs_sort_column { + struct zfs_sort_column *sc_next; + struct zfs_sort_column *sc_last; + zfs_prop_t sc_prop; + char *sc_user_prop; + boolean_t sc_reverse; +} zfs_sort_column_t; + +#define ZFS_ITER_RECURSE (1 << 0) +#define ZFS_ITER_ARGS_CAN_BE_PATHS (1 << 1) +#define ZFS_ITER_PROP_LISTSNAPS (1 << 2) + +int zfs_for_each(int, char **, int options, zfs_type_t, + zfs_sort_column_t *, zprop_list_t **, zfs_iter_f, void *); +int zfs_add_sort_column(zfs_sort_column_t **, const char *, boolean_t); +void zfs_free_sort_columns(zfs_sort_column_t *); + +#ifdef __cplusplus +} +#endif + +#endif /* ZFS_ITER_H */ diff --git a/cmd/zfs/zfs_main.c b/cmd/zfs/zfs_main.c new file mode 100644 index 000000000..f9480ceba --- /dev/null +++ b/cmd/zfs/zfs_main.c @@ -0,0 +1,4229 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ + +/* + * Copyright 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#include <assert.h> +#include <ctype.h> +#include <errno.h> +#include <libgen.h> +#include <libintl.h> +#include <libuutil.h> +#include <libnvpair.h> +#include <locale.h> +#include <stddef.h> +#include <stdio.h> +#include <stdlib.h> +#include <strings.h> +#include <unistd.h> +#include <fcntl.h> +#include <zone.h> +#include <sys/mkdev.h> +#include <sys/mntent.h> +#include <sys/mnttab.h> +#include <sys/mount.h> +#include <sys/stat.h> +#include <sys/avl.h> + +#include <libzfs.h> +#include <libuutil.h> + +#include "zfs_iter.h" +#include "zfs_util.h" + +libzfs_handle_t *g_zfs; + +static FILE *mnttab_file; +static char history_str[HIS_MAX_RECORD_LEN]; + +static int zfs_do_clone(int argc, char **argv); +static int zfs_do_create(int argc, char **argv); +static int zfs_do_destroy(int argc, char **argv); +static int zfs_do_get(int argc, char **argv); +static int zfs_do_inherit(int argc, char **argv); +static int zfs_do_list(int argc, char **argv); +static int zfs_do_mount(int argc, char **argv); +static int zfs_do_rename(int argc, char **argv); +static int zfs_do_rollback(int argc, char **argv); +static int zfs_do_set(int argc, char **argv); +static int zfs_do_upgrade(int argc, char **argv); +static int zfs_do_snapshot(int argc, char **argv); +static int zfs_do_unmount(int argc, char **argv); +static int zfs_do_share(int argc, char **argv); +static int zfs_do_unshare(int argc, char **argv); +static int zfs_do_send(int argc, char **argv); +static int zfs_do_receive(int argc, char **argv); +static int zfs_do_promote(int argc, char **argv); +static int zfs_do_allow(int argc, char **argv); +static int zfs_do_unallow(int argc, char **argv); + +/* + * Enable a reasonable set of defaults for libumem debugging on DEBUG builds. + */ + +#ifdef DEBUG +const char * +_umem_debug_init(void) +{ + return ("default,verbose"); /* $UMEM_DEBUG setting */ +} + +const char * +_umem_logging_init(void) +{ + return ("fail,contents"); /* $UMEM_LOGGING setting */ +} +#endif + +typedef enum { + HELP_CLONE, + HELP_CREATE, + HELP_DESTROY, + HELP_GET, + HELP_INHERIT, + HELP_UPGRADE, + HELP_LIST, + HELP_MOUNT, + HELP_PROMOTE, + HELP_RECEIVE, + HELP_RENAME, + HELP_ROLLBACK, + HELP_SEND, + HELP_SET, + HELP_SHARE, + HELP_SNAPSHOT, + HELP_UNMOUNT, + HELP_UNSHARE, + HELP_ALLOW, + HELP_UNALLOW +} zfs_help_t; + +typedef struct zfs_command { + const char *name; + int (*func)(int argc, char **argv); + zfs_help_t usage; +} zfs_command_t; + +/* + * Master command table. Each ZFS command has a name, associated function, and + * usage message. The usage messages need to be internationalized, so we have + * to have a function to return the usage message based on a command index. + * + * These commands are organized according to how they are displayed in the usage + * message. An empty command (one with a NULL name) indicates an empty line in + * the generic usage message. + */ +static zfs_command_t command_table[] = { + { "create", zfs_do_create, HELP_CREATE }, + { "destroy", zfs_do_destroy, HELP_DESTROY }, + { NULL }, + { "snapshot", zfs_do_snapshot, HELP_SNAPSHOT }, + { "rollback", zfs_do_rollback, HELP_ROLLBACK }, + { "clone", zfs_do_clone, HELP_CLONE }, + { "promote", zfs_do_promote, HELP_PROMOTE }, + { "rename", zfs_do_rename, HELP_RENAME }, + { NULL }, + { "list", zfs_do_list, HELP_LIST }, + { NULL }, + { "set", zfs_do_set, HELP_SET }, + { "get", zfs_do_get, HELP_GET }, + { "inherit", zfs_do_inherit, HELP_INHERIT }, + { "upgrade", zfs_do_upgrade, HELP_UPGRADE }, + { NULL }, + { "mount", zfs_do_mount, HELP_MOUNT }, + { "unmount", zfs_do_unmount, HELP_UNMOUNT }, + { "share", zfs_do_share, HELP_SHARE }, + { "unshare", zfs_do_unshare, HELP_UNSHARE }, + { NULL }, + { "send", zfs_do_send, HELP_SEND }, + { "receive", zfs_do_receive, HELP_RECEIVE }, + { NULL }, + { "allow", zfs_do_allow, HELP_ALLOW }, + { NULL }, + { "unallow", zfs_do_unallow, HELP_UNALLOW }, +}; + +#define NCOMMAND (sizeof (command_table) / sizeof (command_table[0])) + +zfs_command_t *current_command; + +static const char * +get_usage(zfs_help_t idx) +{ + switch (idx) { + case HELP_CLONE: + return (gettext("\tclone [-p] [-o property=value] ... " + "<snapshot> <filesystem|volume>\n")); + case HELP_CREATE: + return (gettext("\tcreate [-p] [-o property=value] ... " + "<filesystem>\n" + "\tcreate [-ps] [-b blocksize] [-o property=value] ... " + "-V <size> <volume>\n")); + case HELP_DESTROY: + return (gettext("\tdestroy [-rRf] " + "<filesystem|volume|snapshot>\n")); + case HELP_GET: + return (gettext("\tget [-rHp] [-o field[,...]] " + "[-s source[,...]]\n" + "\t <\"all\" | property[,...]> " + "[filesystem|volume|snapshot] ...\n")); + case HELP_INHERIT: + return (gettext("\tinherit [-r] <property> " + "<filesystem|volume|snapshot> ...\n")); + case HELP_UPGRADE: + return (gettext("\tupgrade [-v]\n" + "\tupgrade [-r] [-V version] <-a | filesystem ...>\n")); + case HELP_LIST: + return (gettext("\tlist [-rH] [-o property[,...]] " + "[-t type[,...]] [-s property] ...\n" + "\t [-S property] ... " + "[filesystem|volume|snapshot] ...\n")); + case HELP_MOUNT: + return (gettext("\tmount\n" + "\tmount [-vO] [-o opts] <-a | filesystem>\n")); + case HELP_PROMOTE: + return (gettext("\tpromote <clone-filesystem>\n")); + case HELP_RECEIVE: + return (gettext("\treceive [-vnF] <filesystem|volume|" + "snapshot>\n" + "\treceive [-vnF] -d <filesystem>\n")); + case HELP_RENAME: + return (gettext("\trename <filesystem|volume|snapshot> " + "<filesystem|volume|snapshot>\n" + "\trename -p <filesystem|volume> <filesystem|volume>\n" + "\trename -r <snapshot> <snapshot>")); + case HELP_ROLLBACK: + return (gettext("\trollback [-rRf] <snapshot>\n")); + case HELP_SEND: + return (gettext("\tsend [-R] [-[iI] snapshot] <snapshot>\n")); + case HELP_SET: + return (gettext("\tset <property=value> " + "<filesystem|volume|snapshot> ...\n")); + case HELP_SHARE: + return (gettext("\tshare <-a | filesystem>\n")); + case HELP_SNAPSHOT: + return (gettext("\tsnapshot [-r] [-o property=value] ... " + "<filesystem@snapname|volume@snapname>\n")); + case HELP_UNMOUNT: + return (gettext("\tunmount [-f] " + "<-a | filesystem|mountpoint>\n")); + case HELP_UNSHARE: + return (gettext("\tunshare [-f] " + "<-a | filesystem|mountpoint>\n")); + case HELP_ALLOW: + return (gettext("\tallow [-ldug] " + "<\"everyone\"|user|group>[,...] <perm|@setname>[,...]\n" + "\t <filesystem|volume>\n" + "\tallow [-ld] -e <perm|@setname>[,...] " + "<filesystem|volume>\n" + "\tallow -c <perm|@setname>[,...] <filesystem|volume>\n" + "\tallow -s @setname <perm|@setname>[,...] " + "<filesystem|volume>\n")); + case HELP_UNALLOW: + return (gettext("\tunallow [-rldug] " + "<\"everyone\"|user|group>[,...]\n" + "\t [<perm|@setname>[,...]] <filesystem|volume>\n" + "\tunallow [-rld] -e [<perm|@setname>[,...]] " + "<filesystem|volume>\n" + "\tunallow [-r] -c [<perm|@setname>[,...]] " + "<filesystem|volume>\n" + "\tunallow [-r] -s @setname [<perm|@setname>[,...]] " + "<filesystem|volume>\n")); + } + + abort(); + /* NOTREACHED */ +} + +/* + * Utility function to guarantee malloc() success. + */ +void * +safe_malloc(size_t size) +{ + void *data; + + if ((data = calloc(1, size)) == NULL) { + (void) fprintf(stderr, "internal error: out of memory\n"); + exit(1); + } + + return (data); +} + +/* + * Callback routine that will print out information for each of + * the properties. + */ +static int +usage_prop_cb(int prop, void *cb) +{ + FILE *fp = cb; + + (void) fprintf(fp, "\t%-15s ", zfs_prop_to_name(prop)); + + if (zfs_prop_readonly(prop)) + (void) fprintf(fp, " NO "); + else + (void) fprintf(fp, "YES "); + + if (zfs_prop_inheritable(prop)) + (void) fprintf(fp, " YES "); + else + (void) fprintf(fp, " NO "); + + if (zfs_prop_values(prop) == NULL) + (void) fprintf(fp, "-\n"); + else + (void) fprintf(fp, "%s\n", zfs_prop_values(prop)); + + return (ZPROP_CONT); +} + +/* + * Display usage message. If we're inside a command, display only the usage for + * that command. Otherwise, iterate over the entire command table and display + * a complete usage message. + */ +static void +usage(boolean_t requested) +{ + int i; + boolean_t show_properties = B_FALSE; + boolean_t show_permissions = B_FALSE; + FILE *fp = requested ? stdout : stderr; + + if (current_command == NULL) { + + (void) fprintf(fp, gettext("usage: zfs command args ...\n")); + (void) fprintf(fp, + gettext("where 'command' is one of the following:\n\n")); + + for (i = 0; i < NCOMMAND; i++) { + if (command_table[i].name == NULL) + (void) fprintf(fp, "\n"); + else + (void) fprintf(fp, "%s", + get_usage(command_table[i].usage)); + } + + (void) fprintf(fp, gettext("\nEach dataset is of the form: " + "pool/[dataset/]*dataset[@name]\n")); + } else { + (void) fprintf(fp, gettext("usage:\n")); + (void) fprintf(fp, "%s", get_usage(current_command->usage)); + } + + if (current_command != NULL && + (strcmp(current_command->name, "set") == 0 || + strcmp(current_command->name, "get") == 0 || + strcmp(current_command->name, "inherit") == 0 || + strcmp(current_command->name, "list") == 0)) + show_properties = B_TRUE; + + if (current_command != NULL && + (strcmp(current_command->name, "allow") == 0 || + strcmp(current_command->name, "unallow") == 0)) + show_permissions = B_TRUE; + + if (show_properties) { + + (void) fprintf(fp, + gettext("\nThe following properties are supported:\n")); + + (void) fprintf(fp, "\n\t%-14s %s %s %s\n\n", + "PROPERTY", "EDIT", "INHERIT", "VALUES"); + + /* Iterate over all properties */ + (void) zprop_iter(usage_prop_cb, fp, B_FALSE, B_TRUE, + ZFS_TYPE_DATASET); + + (void) fprintf(fp, gettext("\nSizes are specified in bytes " + "with standard units such as K, M, G, etc.\n")); + (void) fprintf(fp, gettext("\nUser-defined properties can " + "be specified by using a name containing a colon (:).\n")); + + } else if (show_permissions) { + (void) fprintf(fp, + gettext("\nThe following permissions are supported:\n")); + + zfs_deleg_permissions(); + } else { + /* + * TRANSLATION NOTE: + * "zfs set|get" must not be localised this is the + * command name and arguments. + */ + + (void) fprintf(fp, + gettext("\nFor the property list, run: zfs set|get\n")); + + (void) fprintf(fp, + gettext("\nFor the delegated permission list, run:" + " zfs allow|unallow\n")); + } + + /* + * See comments at end of main(). + */ + if (getenv("ZFS_ABORT") != NULL) { + (void) printf("dumping core by request\n"); + abort(); + } + + exit(requested ? 0 : 2); +} + +static int +parseprop(nvlist_t *props) +{ + char *propname = optarg; + char *propval, *strval; + + if ((propval = strchr(propname, '=')) == NULL) { + (void) fprintf(stderr, gettext("missing " + "'=' for -o option\n")); + return (-1); + } + *propval = '\0'; + propval++; + if (nvlist_lookup_string(props, propname, &strval) == 0) { + (void) fprintf(stderr, gettext("property '%s' " + "specified multiple times\n"), propname); + return (-1); + } + if (nvlist_add_string(props, propname, propval) != 0) { + (void) fprintf(stderr, gettext("internal " + "error: out of memory\n")); + return (-1); + } + return (0); + +} + +/* + * zfs clone [-p] [-o prop=value] ... <snap> <fs | vol> + * + * Given an existing dataset, create a writable copy whose initial contents + * are the same as the source. The newly created dataset maintains a + * dependency on the original; the original cannot be destroyed so long as + * the clone exists. + * + * The '-p' flag creates all the non-existing ancestors of the target first. + */ +static int +zfs_do_clone(int argc, char **argv) +{ + zfs_handle_t *zhp = NULL; + boolean_t parents = B_FALSE; + nvlist_t *props; + int ret; + int c; + + if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) { + (void) fprintf(stderr, gettext("internal error: " + "out of memory\n")); + return (1); + } + + /* check options */ + while ((c = getopt(argc, argv, "o:p")) != -1) { + switch (c) { + case 'o': + if (parseprop(props)) + return (1); + break; + case 'p': + parents = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + goto usage; + } + } + + argc -= optind; + argv += optind; + + /* check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing source dataset " + "argument\n")); + goto usage; + } + if (argc < 2) { + (void) fprintf(stderr, gettext("missing target dataset " + "argument\n")); + goto usage; + } + if (argc > 2) { + (void) fprintf(stderr, gettext("too many arguments\n")); + goto usage; + } + + /* open the source dataset */ + if ((zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_SNAPSHOT)) == NULL) + return (1); + + if (parents && zfs_name_valid(argv[1], ZFS_TYPE_FILESYSTEM | + ZFS_TYPE_VOLUME)) { + /* + * Now create the ancestors of the target dataset. If the + * target already exists and '-p' option was used we should not + * complain. + */ + if (zfs_dataset_exists(g_zfs, argv[1], ZFS_TYPE_FILESYSTEM | + ZFS_TYPE_VOLUME)) + return (0); + if (zfs_create_ancestors(g_zfs, argv[1]) != 0) + return (1); + } + + /* pass to libzfs */ + ret = zfs_clone(zhp, argv[1], props); + + /* create the mountpoint if necessary */ + if (ret == 0) { + zfs_handle_t *clone; + + clone = zfs_open(g_zfs, argv[1], ZFS_TYPE_DATASET); + if (clone != NULL) { + if ((ret = zfs_mount(clone, NULL, 0)) == 0) + ret = zfs_share(clone); + zfs_close(clone); + } + } + + zfs_close(zhp); + nvlist_free(props); + + return (!!ret); + +usage: + if (zhp) + zfs_close(zhp); + nvlist_free(props); + usage(B_FALSE); + return (-1); +} + +/* + * zfs create [-p] [-o prop=value] ... fs + * zfs create [-ps] [-b blocksize] [-o prop=value] ... -V vol size + * + * Create a new dataset. This command can be used to create filesystems + * and volumes. Snapshot creation is handled by 'zfs snapshot'. + * For volumes, the user must specify a size to be used. + * + * The '-s' flag applies only to volumes, and indicates that we should not try + * to set the reservation for this volume. By default we set a reservation + * equal to the size for any volume. For pools with SPA_VERSION >= + * SPA_VERSION_REFRESERVATION, we set a refreservation instead. + * + * The '-p' flag creates all the non-existing ancestors of the target first. + */ +static int +zfs_do_create(int argc, char **argv) +{ + zfs_type_t type = ZFS_TYPE_FILESYSTEM; + zfs_handle_t *zhp = NULL; + uint64_t volsize; + int c; + boolean_t noreserve = B_FALSE; + boolean_t bflag = B_FALSE; + boolean_t parents = B_FALSE; + int ret = 1; + nvlist_t *props; + uint64_t intval; + int canmount; + + if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) { + (void) fprintf(stderr, gettext("internal error: " + "out of memory\n")); + return (1); + } + + /* check options */ + while ((c = getopt(argc, argv, ":V:b:so:p")) != -1) { + switch (c) { + case 'V': + type = ZFS_TYPE_VOLUME; + if (zfs_nicestrtonum(g_zfs, optarg, &intval) != 0) { + (void) fprintf(stderr, gettext("bad volume " + "size '%s': %s\n"), optarg, + libzfs_error_description(g_zfs)); + goto error; + } + + if (nvlist_add_uint64(props, + zfs_prop_to_name(ZFS_PROP_VOLSIZE), + intval) != 0) { + (void) fprintf(stderr, gettext("internal " + "error: out of memory\n")); + goto error; + } + volsize = intval; + break; + case 'p': + parents = B_TRUE; + break; + case 'b': + bflag = B_TRUE; + if (zfs_nicestrtonum(g_zfs, optarg, &intval) != 0) { + (void) fprintf(stderr, gettext("bad volume " + "block size '%s': %s\n"), optarg, + libzfs_error_description(g_zfs)); + goto error; + } + + if (nvlist_add_uint64(props, + zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE), + intval) != 0) { + (void) fprintf(stderr, gettext("internal " + "error: out of memory\n")); + goto error; + } + break; + case 'o': + if (parseprop(props)) + goto error; + break; + case 's': + noreserve = B_TRUE; + break; + case ':': + (void) fprintf(stderr, gettext("missing size " + "argument\n")); + goto badusage; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + goto badusage; + } + } + + if ((bflag || noreserve) && type != ZFS_TYPE_VOLUME) { + (void) fprintf(stderr, gettext("'-s' and '-b' can only be " + "used when creating a volume\n")); + goto badusage; + } + + argc -= optind; + argv += optind; + + /* check number of arguments */ + if (argc == 0) { + (void) fprintf(stderr, gettext("missing %s argument\n"), + zfs_type_to_name(type)); + goto badusage; + } + if (argc > 1) { + (void) fprintf(stderr, gettext("too many arguments\n")); + goto badusage; + } + + if (type == ZFS_TYPE_VOLUME && !noreserve) { + zpool_handle_t *zpool_handle; + uint64_t spa_version; + char *p; + zfs_prop_t resv_prop; + char *strval; + + if (p = strchr(argv[0], '/')) + *p = '\0'; + zpool_handle = zpool_open(g_zfs, argv[0]); + if (p != NULL) + *p = '/'; + if (zpool_handle == NULL) + goto error; + spa_version = zpool_get_prop_int(zpool_handle, + ZPOOL_PROP_VERSION, NULL); + zpool_close(zpool_handle); + if (spa_version >= SPA_VERSION_REFRESERVATION) + resv_prop = ZFS_PROP_REFRESERVATION; + else + resv_prop = ZFS_PROP_RESERVATION; + + if (nvlist_lookup_string(props, zfs_prop_to_name(resv_prop), + &strval) != 0) { + if (nvlist_add_uint64(props, + zfs_prop_to_name(resv_prop), volsize) != 0) { + (void) fprintf(stderr, gettext("internal " + "error: out of memory\n")); + nvlist_free(props); + return (1); + } + } + } + + if (parents && zfs_name_valid(argv[0], type)) { + /* + * Now create the ancestors of target dataset. If the target + * already exists and '-p' option was used we should not + * complain. + */ + if (zfs_dataset_exists(g_zfs, argv[0], type)) { + ret = 0; + goto error; + } + if (zfs_create_ancestors(g_zfs, argv[0]) != 0) + goto error; + } + + /* pass to libzfs */ + if (zfs_create(g_zfs, argv[0], type, props) != 0) + goto error; + + if ((zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_DATASET)) == NULL) + goto error; + /* + * if the user doesn't want the dataset automatically mounted, + * then skip the mount/share step + */ + + canmount = zfs_prop_get_int(zhp, ZFS_PROP_CANMOUNT); + + /* + * Mount and/or share the new filesystem as appropriate. We provide a + * verbose error message to let the user know that their filesystem was + * in fact created, even if we failed to mount or share it. + */ + ret = 0; + if (canmount == ZFS_CANMOUNT_ON) { + if (zfs_mount(zhp, NULL, 0) != 0) { + (void) fprintf(stderr, gettext("filesystem " + "successfully created, but not mounted\n")); + ret = 1; + } else if (zfs_share(zhp) != 0) { + (void) fprintf(stderr, gettext("filesystem " + "successfully created, but not shared\n")); + ret = 1; + } + } + +error: + if (zhp) + zfs_close(zhp); + nvlist_free(props); + return (ret); +badusage: + nvlist_free(props); + usage(B_FALSE); + return (2); +} + +/* + * zfs destroy [-rf] <fs, snap, vol> + * + * -r Recursively destroy all children + * -R Recursively destroy all dependents, including clones + * -f Force unmounting of any dependents + * + * Destroys the given dataset. By default, it will unmount any filesystems, + * and refuse to destroy a dataset that has any dependents. A dependent can + * either be a child, or a clone of a child. + */ +typedef struct destroy_cbdata { + boolean_t cb_first; + int cb_force; + int cb_recurse; + int cb_error; + int cb_needforce; + int cb_doclones; + boolean_t cb_closezhp; + zfs_handle_t *cb_target; + char *cb_snapname; +} destroy_cbdata_t; + +/* + * Check for any dependents based on the '-r' or '-R' flags. + */ +static int +destroy_check_dependent(zfs_handle_t *zhp, void *data) +{ + destroy_cbdata_t *cbp = data; + const char *tname = zfs_get_name(cbp->cb_target); + const char *name = zfs_get_name(zhp); + + if (strncmp(tname, name, strlen(tname)) == 0 && + (name[strlen(tname)] == '/' || name[strlen(tname)] == '@')) { + /* + * This is a direct descendant, not a clone somewhere else in + * the hierarchy. + */ + if (cbp->cb_recurse) + goto out; + + if (cbp->cb_first) { + (void) fprintf(stderr, gettext("cannot destroy '%s': " + "%s has children\n"), + zfs_get_name(cbp->cb_target), + zfs_type_to_name(zfs_get_type(cbp->cb_target))); + (void) fprintf(stderr, gettext("use '-r' to destroy " + "the following datasets:\n")); + cbp->cb_first = B_FALSE; + cbp->cb_error = 1; + } + + (void) fprintf(stderr, "%s\n", zfs_get_name(zhp)); + } else { + /* + * This is a clone. We only want to report this if the '-r' + * wasn't specified, or the target is a snapshot. + */ + if (!cbp->cb_recurse && + zfs_get_type(cbp->cb_target) != ZFS_TYPE_SNAPSHOT) + goto out; + + if (cbp->cb_first) { + (void) fprintf(stderr, gettext("cannot destroy '%s': " + "%s has dependent clones\n"), + zfs_get_name(cbp->cb_target), + zfs_type_to_name(zfs_get_type(cbp->cb_target))); + (void) fprintf(stderr, gettext("use '-R' to destroy " + "the following datasets:\n")); + cbp->cb_first = B_FALSE; + cbp->cb_error = 1; + } + + (void) fprintf(stderr, "%s\n", zfs_get_name(zhp)); + } + +out: + zfs_close(zhp); + return (0); +} + +static int +destroy_callback(zfs_handle_t *zhp, void *data) +{ + destroy_cbdata_t *cbp = data; + + /* + * Ignore pools (which we've already flagged as an error before getting + * here. + */ + if (strchr(zfs_get_name(zhp), '/') == NULL && + zfs_get_type(zhp) == ZFS_TYPE_FILESYSTEM) { + zfs_close(zhp); + return (0); + } + + /* + * Bail out on the first error. + */ + if (zfs_unmount(zhp, NULL, cbp->cb_force ? MS_FORCE : 0) != 0 || + zfs_destroy(zhp) != 0) { + zfs_close(zhp); + return (-1); + } + + zfs_close(zhp); + return (0); +} + +static int +destroy_snap_clones(zfs_handle_t *zhp, void *arg) +{ + destroy_cbdata_t *cbp = arg; + char thissnap[MAXPATHLEN]; + zfs_handle_t *szhp; + boolean_t closezhp = cbp->cb_closezhp; + int rv; + + (void) snprintf(thissnap, sizeof (thissnap), + "%s@%s", zfs_get_name(zhp), cbp->cb_snapname); + + libzfs_print_on_error(g_zfs, B_FALSE); + szhp = zfs_open(g_zfs, thissnap, ZFS_TYPE_SNAPSHOT); + libzfs_print_on_error(g_zfs, B_TRUE); + if (szhp) { + /* + * Destroy any clones of this snapshot + */ + if (zfs_iter_dependents(szhp, B_FALSE, destroy_callback, + cbp) != 0) { + zfs_close(szhp); + if (closezhp) + zfs_close(zhp); + return (-1); + } + zfs_close(szhp); + } + + cbp->cb_closezhp = B_TRUE; + rv = zfs_iter_filesystems(zhp, destroy_snap_clones, arg); + if (closezhp) + zfs_close(zhp); + return (rv); +} + +static int +zfs_do_destroy(int argc, char **argv) +{ + destroy_cbdata_t cb = { 0 }; + int c; + zfs_handle_t *zhp; + char *cp; + + /* check options */ + while ((c = getopt(argc, argv, "frR")) != -1) { + switch (c) { + case 'f': + cb.cb_force = 1; + break; + case 'r': + cb.cb_recurse = 1; + break; + case 'R': + cb.cb_recurse = 1; + cb.cb_doclones = 1; + break; + case '?': + default: + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* check number of arguments */ + if (argc == 0) { + (void) fprintf(stderr, gettext("missing path argument\n")); + usage(B_FALSE); + } + if (argc > 1) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + /* + * If we are doing recursive destroy of a snapshot, then the + * named snapshot may not exist. Go straight to libzfs. + */ + if (cb.cb_recurse && (cp = strchr(argv[0], '@'))) { + int ret; + + *cp = '\0'; + if ((zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_DATASET)) == NULL) + return (1); + *cp = '@'; + cp++; + + if (cb.cb_doclones) { + cb.cb_snapname = cp; + if (destroy_snap_clones(zhp, &cb) != 0) { + zfs_close(zhp); + return (1); + } + } + + ret = zfs_destroy_snaps(zhp, cp); + zfs_close(zhp); + if (ret) { + (void) fprintf(stderr, + gettext("no snapshots destroyed\n")); + } + return (ret != 0); + } + + + /* Open the given dataset */ + if ((zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_DATASET)) == NULL) + return (1); + + cb.cb_target = zhp; + + /* + * Perform an explicit check for pools before going any further. + */ + if (!cb.cb_recurse && strchr(zfs_get_name(zhp), '/') == NULL && + zfs_get_type(zhp) == ZFS_TYPE_FILESYSTEM) { + (void) fprintf(stderr, gettext("cannot destroy '%s': " + "operation does not apply to pools\n"), + zfs_get_name(zhp)); + (void) fprintf(stderr, gettext("use 'zfs destroy -r " + "%s' to destroy all datasets in the pool\n"), + zfs_get_name(zhp)); + (void) fprintf(stderr, gettext("use 'zpool destroy %s' " + "to destroy the pool itself\n"), zfs_get_name(zhp)); + zfs_close(zhp); + return (1); + } + + /* + * Check for any dependents and/or clones. + */ + cb.cb_first = B_TRUE; + if (!cb.cb_doclones && + zfs_iter_dependents(zhp, B_TRUE, destroy_check_dependent, + &cb) != 0) { + zfs_close(zhp); + return (1); + } + + if (cb.cb_error || + zfs_iter_dependents(zhp, B_FALSE, destroy_callback, &cb) != 0) { + zfs_close(zhp); + return (1); + } + + /* + * Do the real thing. The callback will close the handle regardless of + * whether it succeeds or not. + */ + + if (destroy_callback(zhp, &cb) != 0) + return (1); + + + return (0); +} + +/* + * zfs get [-rHp] [-o field[,field]...] [-s source[,source]...] + * < all | property[,property]... > < fs | snap | vol > ... + * + * -r recurse over any child datasets + * -H scripted mode. Headers are stripped, and fields are separated + * by tabs instead of spaces. + * -o Set of fields to display. One of "name,property,value,source". + * Default is all four. + * -s Set of sources to allow. One of + * "local,default,inherited,temporary,none". Default is all + * five. + * -p Display values in parsable (literal) format. + * + * Prints properties for the given datasets. The user can control which + * columns to display as well as which property types to allow. + */ + +/* + * Invoked to display the properties for a single dataset. + */ +static int +get_callback(zfs_handle_t *zhp, void *data) +{ + char buf[ZFS_MAXPROPLEN]; + zprop_source_t sourcetype; + char source[ZFS_MAXNAMELEN]; + zprop_get_cbdata_t *cbp = data; + nvlist_t *userprop = zfs_get_user_props(zhp); + zprop_list_t *pl = cbp->cb_proplist; + nvlist_t *propval; + char *strval; + char *sourceval; + + for (; pl != NULL; pl = pl->pl_next) { + /* + * Skip the special fake placeholder. This will also skip over + * the name property when 'all' is specified. + */ + if (pl->pl_prop == ZFS_PROP_NAME && + pl == cbp->cb_proplist) + continue; + + if (pl->pl_prop != ZPROP_INVAL) { + if (zfs_prop_get(zhp, pl->pl_prop, buf, + sizeof (buf), &sourcetype, source, + sizeof (source), + cbp->cb_literal) != 0) { + if (pl->pl_all) + continue; + if (!zfs_prop_valid_for_type(pl->pl_prop, + ZFS_TYPE_DATASET)) { + (void) fprintf(stderr, + gettext("No such property '%s'\n"), + zfs_prop_to_name(pl->pl_prop)); + continue; + } + sourcetype = ZPROP_SRC_NONE; + (void) strlcpy(buf, "-", sizeof (buf)); + } + + zprop_print_one_property(zfs_get_name(zhp), cbp, + zfs_prop_to_name(pl->pl_prop), + buf, sourcetype, source); + } else { + if (nvlist_lookup_nvlist(userprop, + pl->pl_user_prop, &propval) != 0) { + if (pl->pl_all) + continue; + sourcetype = ZPROP_SRC_NONE; + strval = "-"; + } else { + verify(nvlist_lookup_string(propval, + ZPROP_VALUE, &strval) == 0); + verify(nvlist_lookup_string(propval, + ZPROP_SOURCE, &sourceval) == 0); + + if (strcmp(sourceval, + zfs_get_name(zhp)) == 0) { + sourcetype = ZPROP_SRC_LOCAL; + } else { + sourcetype = ZPROP_SRC_INHERITED; + (void) strlcpy(source, + sourceval, sizeof (source)); + } + } + + zprop_print_one_property(zfs_get_name(zhp), cbp, + pl->pl_user_prop, strval, sourcetype, + source); + } + } + + return (0); +} + +static int +zfs_do_get(int argc, char **argv) +{ + zprop_get_cbdata_t cb = { 0 }; + int i, c, flags = 0; + char *value, *fields; + int ret; + zprop_list_t fake_name = { 0 }; + + /* + * Set up default columns and sources. + */ + cb.cb_sources = ZPROP_SRC_ALL; + cb.cb_columns[0] = GET_COL_NAME; + cb.cb_columns[1] = GET_COL_PROPERTY; + cb.cb_columns[2] = GET_COL_VALUE; + cb.cb_columns[3] = GET_COL_SOURCE; + cb.cb_type = ZFS_TYPE_DATASET; + + /* check options */ + while ((c = getopt(argc, argv, ":o:s:rHp")) != -1) { + switch (c) { + case 'p': + cb.cb_literal = B_TRUE; + break; + case 'r': + flags |= ZFS_ITER_RECURSE; + break; + case 'H': + cb.cb_scripted = B_TRUE; + break; + case ':': + (void) fprintf(stderr, gettext("missing argument for " + "'%c' option\n"), optopt); + usage(B_FALSE); + break; + case 'o': + /* + * Process the set of columns to display. We zero out + * the structure to give us a blank slate. + */ + bzero(&cb.cb_columns, sizeof (cb.cb_columns)); + i = 0; + while (*optarg != '\0') { + static char *col_subopts[] = + { "name", "property", "value", "source", + NULL }; + + if (i == 4) { + (void) fprintf(stderr, gettext("too " + "many fields given to -o " + "option\n")); + usage(B_FALSE); + } + + switch (getsubopt(&optarg, col_subopts, + &value)) { + case 0: + cb.cb_columns[i++] = GET_COL_NAME; + break; + case 1: + cb.cb_columns[i++] = GET_COL_PROPERTY; + break; + case 2: + cb.cb_columns[i++] = GET_COL_VALUE; + break; + case 3: + cb.cb_columns[i++] = GET_COL_SOURCE; + break; + default: + (void) fprintf(stderr, + gettext("invalid column name " + "'%s'\n"), value); + usage(B_FALSE); + } + } + break; + + case 's': + cb.cb_sources = 0; + while (*optarg != '\0') { + static char *source_subopts[] = { + "local", "default", "inherited", + "temporary", "none", NULL }; + + switch (getsubopt(&optarg, source_subopts, + &value)) { + case 0: + cb.cb_sources |= ZPROP_SRC_LOCAL; + break; + case 1: + cb.cb_sources |= ZPROP_SRC_DEFAULT; + break; + case 2: + cb.cb_sources |= ZPROP_SRC_INHERITED; + break; + case 3: + cb.cb_sources |= ZPROP_SRC_TEMPORARY; + break; + case 4: + cb.cb_sources |= ZPROP_SRC_NONE; + break; + default: + (void) fprintf(stderr, + gettext("invalid source " + "'%s'\n"), value); + usage(B_FALSE); + } + } + break; + + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + if (argc < 1) { + (void) fprintf(stderr, gettext("missing property " + "argument\n")); + usage(B_FALSE); + } + + fields = argv[0]; + + if (zprop_get_list(g_zfs, fields, &cb.cb_proplist, ZFS_TYPE_DATASET) + != 0) + usage(B_FALSE); + + argc--; + argv++; + + /* + * As part of zfs_expand_proplist(), we keep track of the maximum column + * width for each property. For the 'NAME' (and 'SOURCE') columns, we + * need to know the maximum name length. However, the user likely did + * not specify 'name' as one of the properties to fetch, so we need to + * make sure we always include at least this property for + * print_get_headers() to work properly. + */ + if (cb.cb_proplist != NULL) { + fake_name.pl_prop = ZFS_PROP_NAME; + fake_name.pl_width = strlen(gettext("NAME")); + fake_name.pl_next = cb.cb_proplist; + cb.cb_proplist = &fake_name; + } + + cb.cb_first = B_TRUE; + + /* run for each object */ + ret = zfs_for_each(argc, argv, flags, ZFS_TYPE_DATASET, NULL, + &cb.cb_proplist, get_callback, &cb); + + if (cb.cb_proplist == &fake_name) + zprop_free_list(fake_name.pl_next); + else + zprop_free_list(cb.cb_proplist); + + return (ret); +} + +/* + * inherit [-r] <property> <fs|vol> ... + * + * -r Recurse over all children + * + * For each dataset specified on the command line, inherit the given property + * from its parent. Inheriting a property at the pool level will cause it to + * use the default value. The '-r' flag will recurse over all children, and is + * useful for setting a property on a hierarchy-wide basis, regardless of any + * local modifications for each dataset. + */ + +static int +inherit_recurse_cb(zfs_handle_t *zhp, void *data) +{ + char *propname = data; + zfs_prop_t prop = zfs_name_to_prop(propname); + + /* + * If we're doing it recursively, then ignore properties that + * are not valid for this type of dataset. + */ + if (prop != ZPROP_INVAL && + !zfs_prop_valid_for_type(prop, zfs_get_type(zhp))) + return (0); + + return (zfs_prop_inherit(zhp, propname) != 0); +} + +static int +inherit_cb(zfs_handle_t *zhp, void *data) +{ + char *propname = data; + + return (zfs_prop_inherit(zhp, propname) != 0); +} + +static int +zfs_do_inherit(int argc, char **argv) +{ + int c; + zfs_prop_t prop; + char *propname; + int ret; + int flags = 0; + + /* check options */ + while ((c = getopt(argc, argv, "r")) != -1) { + switch (c) { + case 'r': + flags |= ZFS_ITER_RECURSE; + break; + case '?': + default: + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing property argument\n")); + usage(B_FALSE); + } + if (argc < 2) { + (void) fprintf(stderr, gettext("missing dataset argument\n")); + usage(B_FALSE); + } + + propname = argv[0]; + argc--; + argv++; + + if ((prop = zfs_name_to_prop(propname)) != ZPROP_INVAL) { + if (zfs_prop_readonly(prop)) { + (void) fprintf(stderr, gettext( + "%s property is read-only\n"), + propname); + return (1); + } + if (!zfs_prop_inheritable(prop)) { + (void) fprintf(stderr, gettext("'%s' property cannot " + "be inherited\n"), propname); + if (prop == ZFS_PROP_QUOTA || + prop == ZFS_PROP_RESERVATION || + prop == ZFS_PROP_REFQUOTA || + prop == ZFS_PROP_REFRESERVATION) + (void) fprintf(stderr, gettext("use 'zfs set " + "%s=none' to clear\n"), propname); + return (1); + } + } else if (!zfs_prop_user(propname)) { + (void) fprintf(stderr, gettext("invalid property '%s'\n"), + propname); + usage(B_FALSE); + } + + if (flags & ZFS_ITER_RECURSE) { + ret = zfs_for_each(argc, argv, flags, ZFS_TYPE_DATASET, + NULL, NULL, inherit_recurse_cb, propname); + } else { + ret = zfs_for_each(argc, argv, flags, ZFS_TYPE_DATASET, + NULL, NULL, inherit_cb, propname); + } + + return (ret); +} + +typedef struct upgrade_cbdata { + uint64_t cb_numupgraded; + uint64_t cb_numsamegraded; + uint64_t cb_numfailed; + uint64_t cb_version; + boolean_t cb_newer; + boolean_t cb_foundone; + char cb_lastfs[ZFS_MAXNAMELEN]; +} upgrade_cbdata_t; + +static int +same_pool(zfs_handle_t *zhp, const char *name) +{ + int len1 = strcspn(name, "/@"); + const char *zhname = zfs_get_name(zhp); + int len2 = strcspn(zhname, "/@"); + + if (len1 != len2) + return (B_FALSE); + return (strncmp(name, zhname, len1) == 0); +} + +static int +upgrade_list_callback(zfs_handle_t *zhp, void *data) +{ + upgrade_cbdata_t *cb = data; + int version = zfs_prop_get_int(zhp, ZFS_PROP_VERSION); + + /* list if it's old/new */ + if ((!cb->cb_newer && version < ZPL_VERSION) || + (cb->cb_newer && version > ZPL_VERSION)) { + char *str; + if (cb->cb_newer) { + str = gettext("The following filesystems are " + "formatted using a newer software version and\n" + "cannot be accessed on the current system.\n\n"); + } else { + str = gettext("The following filesystems are " + "out of date, and can be upgraded. After being\n" + "upgraded, these filesystems (and any 'zfs send' " + "streams generated from\n" + "subsequent snapshots) will no longer be " + "accessible by older software versions.\n\n"); + } + + if (!cb->cb_foundone) { + (void) puts(str); + (void) printf(gettext("VER FILESYSTEM\n")); + (void) printf(gettext("--- ------------\n")); + cb->cb_foundone = B_TRUE; + } + + (void) printf("%2u %s\n", version, zfs_get_name(zhp)); + } + + return (0); +} + +static int +upgrade_set_callback(zfs_handle_t *zhp, void *data) +{ + upgrade_cbdata_t *cb = data; + int version = zfs_prop_get_int(zhp, ZFS_PROP_VERSION); + + if (cb->cb_version >= ZPL_VERSION_FUID) { + int spa_version; + + if (zfs_spa_version(zhp, &spa_version) < 0) + return (-1); + + if (spa_version < SPA_VERSION_FUID) { + /* can't upgrade */ + (void) printf(gettext("%s: can not be upgraded; " + "the pool version needs to first be upgraded\nto " + "version %d\n\n"), + zfs_get_name(zhp), SPA_VERSION_FUID); + cb->cb_numfailed++; + return (0); + } + } + + /* upgrade */ + if (version < cb->cb_version) { + char verstr[16]; + (void) snprintf(verstr, sizeof (verstr), + "%llu", cb->cb_version); + if (cb->cb_lastfs[0] && !same_pool(zhp, cb->cb_lastfs)) { + /* + * If they did "zfs upgrade -a", then we could + * be doing ioctls to different pools. We need + * to log this history once to each pool. + */ + verify(zpool_stage_history(g_zfs, history_str) == 0); + } + if (zfs_prop_set(zhp, "version", verstr) == 0) + cb->cb_numupgraded++; + else + cb->cb_numfailed++; + (void) strcpy(cb->cb_lastfs, zfs_get_name(zhp)); + } else if (version > cb->cb_version) { + /* can't downgrade */ + (void) printf(gettext("%s: can not be downgraded; " + "it is already at version %u\n"), + zfs_get_name(zhp), version); + cb->cb_numfailed++; + } else { + cb->cb_numsamegraded++; + } + return (0); +} + +/* + * zfs upgrade + * zfs upgrade -v + * zfs upgrade [-r] [-V <version>] <-a | filesystem> + */ +static int +zfs_do_upgrade(int argc, char **argv) +{ + boolean_t all = B_FALSE; + boolean_t showversions = B_FALSE; + int ret; + upgrade_cbdata_t cb = { 0 }; + char c; + int flags = ZFS_ITER_ARGS_CAN_BE_PATHS; + + /* check options */ + while ((c = getopt(argc, argv, "rvV:a")) != -1) { + switch (c) { + case 'r': + flags |= ZFS_ITER_RECURSE; + break; + case 'v': + showversions = B_TRUE; + break; + case 'V': + if (zfs_prop_string_to_index(ZFS_PROP_VERSION, + optarg, &cb.cb_version) != 0) { + (void) fprintf(stderr, + gettext("invalid version %s\n"), optarg); + usage(B_FALSE); + } + break; + case 'a': + all = B_TRUE; + break; + case '?': + default: + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + if ((!all && !argc) && ((flags & ZFS_ITER_RECURSE) | cb.cb_version)) + usage(B_FALSE); + if (showversions && (flags & ZFS_ITER_RECURSE || all || + cb.cb_version || argc)) + usage(B_FALSE); + if ((all || argc) && (showversions)) + usage(B_FALSE); + if (all && argc) + usage(B_FALSE); + + if (showversions) { + /* Show info on available versions. */ + (void) printf(gettext("The following filesystem versions are " + "supported:\n\n")); + (void) printf(gettext("VER DESCRIPTION\n")); + (void) printf("--- -----------------------------------------" + "---------------\n"); + (void) printf(gettext(" 1 Initial ZFS filesystem version\n")); + (void) printf(gettext(" 2 Enhanced directory entries\n")); + (void) printf(gettext(" 3 Case insensitive and File system " + "unique identifer (FUID)\n")); + (void) printf(gettext("\nFor more information on a particular " + "version, including supported releases, see:\n\n")); + (void) printf("http://www.opensolaris.org/os/community/zfs/" + "version/zpl/N\n\n"); + (void) printf(gettext("Where 'N' is the version number.\n")); + ret = 0; + } else if (argc || all) { + /* Upgrade filesystems */ + if (cb.cb_version == 0) + cb.cb_version = ZPL_VERSION; + ret = zfs_for_each(argc, argv, flags, ZFS_TYPE_FILESYSTEM, + NULL, NULL, upgrade_set_callback, &cb); + (void) printf(gettext("%llu filesystems upgraded\n"), + cb.cb_numupgraded); + if (cb.cb_numsamegraded) { + (void) printf(gettext("%llu filesystems already at " + "this version\n"), + cb.cb_numsamegraded); + } + if (cb.cb_numfailed != 0) + ret = 1; + } else { + /* List old-version filesytems */ + boolean_t found; + (void) printf(gettext("This system is currently running " + "ZFS filesystem version %llu.\n\n"), ZPL_VERSION); + + flags |= ZFS_ITER_RECURSE; + ret = zfs_for_each(0, NULL, flags, ZFS_TYPE_FILESYSTEM, + NULL, NULL, upgrade_list_callback, &cb); + + found = cb.cb_foundone; + cb.cb_foundone = B_FALSE; + cb.cb_newer = B_TRUE; + + ret = zfs_for_each(0, NULL, flags, ZFS_TYPE_FILESYSTEM, + NULL, NULL, upgrade_list_callback, &cb); + + if (!cb.cb_foundone && !found) { + (void) printf(gettext("All filesystems are " + "formatted with the current version.\n")); + } + } + + return (ret); +} + +/* + * list [-rH] [-o property[,property]...] [-t type[,type]...] + * [-s property [-s property]...] [-S property [-S property]...] + * <dataset> ... + * + * -r Recurse over all children + * -H Scripted mode; elide headers and separate columns by tabs + * -o Control which fields to display. + * -t Control which object types to display. + * -s Specify sort columns, descending order. + * -S Specify sort columns, ascending order. + * + * When given no arguments, lists all filesystems in the system. + * Otherwise, list the specified datasets, optionally recursing down them if + * '-r' is specified. + */ +typedef struct list_cbdata { + boolean_t cb_first; + boolean_t cb_scripted; + zprop_list_t *cb_proplist; +} list_cbdata_t; + +/* + * Given a list of columns to display, output appropriate headers for each one. + */ +static void +print_header(zprop_list_t *pl) +{ + char headerbuf[ZFS_MAXPROPLEN]; + const char *header; + int i; + boolean_t first = B_TRUE; + boolean_t right_justify; + + for (; pl != NULL; pl = pl->pl_next) { + if (!first) { + (void) printf(" "); + } else { + first = B_FALSE; + } + + right_justify = B_FALSE; + if (pl->pl_prop != ZPROP_INVAL) { + header = zfs_prop_column_name(pl->pl_prop); + right_justify = zfs_prop_align_right(pl->pl_prop); + } else { + for (i = 0; pl->pl_user_prop[i] != '\0'; i++) + headerbuf[i] = toupper(pl->pl_user_prop[i]); + headerbuf[i] = '\0'; + header = headerbuf; + } + + if (pl->pl_next == NULL && !right_justify) + (void) printf("%s", header); + else if (right_justify) + (void) printf("%*s", pl->pl_width, header); + else + (void) printf("%-*s", pl->pl_width, header); + } + + (void) printf("\n"); +} + +/* + * Given a dataset and a list of fields, print out all the properties according + * to the described layout. + */ +static void +print_dataset(zfs_handle_t *zhp, zprop_list_t *pl, boolean_t scripted) +{ + boolean_t first = B_TRUE; + char property[ZFS_MAXPROPLEN]; + nvlist_t *userprops = zfs_get_user_props(zhp); + nvlist_t *propval; + char *propstr; + boolean_t right_justify; + int width; + + for (; pl != NULL; pl = pl->pl_next) { + if (!first) { + if (scripted) + (void) printf("\t"); + else + (void) printf(" "); + } else { + first = B_FALSE; + } + + right_justify = B_FALSE; + if (pl->pl_prop != ZPROP_INVAL) { + if (zfs_prop_get(zhp, pl->pl_prop, property, + sizeof (property), NULL, NULL, 0, B_FALSE) != 0) + propstr = "-"; + else + propstr = property; + + right_justify = zfs_prop_align_right(pl->pl_prop); + } else { + if (nvlist_lookup_nvlist(userprops, + pl->pl_user_prop, &propval) != 0) + propstr = "-"; + else + verify(nvlist_lookup_string(propval, + ZPROP_VALUE, &propstr) == 0); + } + + width = pl->pl_width; + + /* + * If this is being called in scripted mode, or if this is the + * last column and it is left-justified, don't include a width + * format specifier. + */ + if (scripted || (pl->pl_next == NULL && !right_justify)) + (void) printf("%s", propstr); + else if (right_justify) + (void) printf("%*s", width, propstr); + else + (void) printf("%-*s", width, propstr); + } + + (void) printf("\n"); +} + +/* + * Generic callback function to list a dataset or snapshot. + */ +static int +list_callback(zfs_handle_t *zhp, void *data) +{ + list_cbdata_t *cbp = data; + + if (cbp->cb_first) { + if (!cbp->cb_scripted) + print_header(cbp->cb_proplist); + cbp->cb_first = B_FALSE; + } + + print_dataset(zhp, cbp->cb_proplist, cbp->cb_scripted); + + return (0); +} + +static int +zfs_do_list(int argc, char **argv) +{ + int c; + boolean_t scripted = B_FALSE; + static char default_fields[] = + "name,used,available,referenced,mountpoint"; + int types = ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME; + boolean_t types_specified = B_FALSE; + char *fields = NULL; + list_cbdata_t cb = { 0 }; + char *value; + int ret; + zfs_sort_column_t *sortcol = NULL; + int flags = ZFS_ITER_PROP_LISTSNAPS | ZFS_ITER_ARGS_CAN_BE_PATHS; + + /* check options */ + while ((c = getopt(argc, argv, ":o:rt:Hs:S:")) != -1) { + switch (c) { + case 'o': + fields = optarg; + break; + case 'r': + flags |= ZFS_ITER_RECURSE; + break; + case 'H': + scripted = B_TRUE; + break; + case 's': + if (zfs_add_sort_column(&sortcol, optarg, + B_FALSE) != 0) { + (void) fprintf(stderr, + gettext("invalid property '%s'\n"), optarg); + usage(B_FALSE); + } + break; + case 'S': + if (zfs_add_sort_column(&sortcol, optarg, + B_TRUE) != 0) { + (void) fprintf(stderr, + gettext("invalid property '%s'\n"), optarg); + usage(B_FALSE); + } + break; + case 't': + types = 0; + types_specified = B_TRUE; + flags &= ~ZFS_ITER_PROP_LISTSNAPS; + while (*optarg != '\0') { + static char *type_subopts[] = { "filesystem", + "volume", "snapshot", "all", NULL }; + + switch (getsubopt(&optarg, type_subopts, + &value)) { + case 0: + types |= ZFS_TYPE_FILESYSTEM; + break; + case 1: + types |= ZFS_TYPE_VOLUME; + break; + case 2: + types |= ZFS_TYPE_SNAPSHOT; + break; + case 3: + types = ZFS_TYPE_DATASET; + break; + + default: + (void) fprintf(stderr, + gettext("invalid type '%s'\n"), + value); + usage(B_FALSE); + } + } + break; + case ':': + (void) fprintf(stderr, gettext("missing argument for " + "'%c' option\n"), optopt); + usage(B_FALSE); + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + if (fields == NULL) + fields = default_fields; + + /* + * If "-o space" and no types were specified, don't display snapshots. + */ + if (strcmp(fields, "space") == 0 && types_specified == B_FALSE) + types &= ~ZFS_TYPE_SNAPSHOT; + + /* + * If the user specifies '-o all', the zprop_get_list() doesn't + * normally include the name of the dataset. For 'zfs list', we always + * want this property to be first. + */ + if (zprop_get_list(g_zfs, fields, &cb.cb_proplist, ZFS_TYPE_DATASET) + != 0) + usage(B_FALSE); + + cb.cb_scripted = scripted; + cb.cb_first = B_TRUE; + + ret = zfs_for_each(argc, argv, flags, types, sortcol, &cb.cb_proplist, + list_callback, &cb); + + zprop_free_list(cb.cb_proplist); + zfs_free_sort_columns(sortcol); + + if (ret == 0 && cb.cb_first && !cb.cb_scripted) + (void) printf(gettext("no datasets available\n")); + + return (ret); +} + +/* + * zfs rename <fs | snap | vol> <fs | snap | vol> + * zfs rename -p <fs | vol> <fs | vol> + * zfs rename -r <snap> <snap> + * + * Renames the given dataset to another of the same type. + * + * The '-p' flag creates all the non-existing ancestors of the target first. + */ +/* ARGSUSED */ +static int +zfs_do_rename(int argc, char **argv) +{ + zfs_handle_t *zhp; + int c; + int ret; + boolean_t recurse = B_FALSE; + boolean_t parents = B_FALSE; + + /* check options */ + while ((c = getopt(argc, argv, "pr")) != -1) { + switch (c) { + case 'p': + parents = B_TRUE; + break; + case 'r': + recurse = B_TRUE; + break; + case '?': + default: + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing source dataset " + "argument\n")); + usage(B_FALSE); + } + if (argc < 2) { + (void) fprintf(stderr, gettext("missing target dataset " + "argument\n")); + usage(B_FALSE); + } + if (argc > 2) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + if (recurse && parents) { + (void) fprintf(stderr, gettext("-p and -r options are mutually " + "exclusive\n")); + usage(B_FALSE); + } + + if (recurse && strchr(argv[0], '@') == 0) { + (void) fprintf(stderr, gettext("source dataset for recursive " + "rename must be a snapshot\n")); + usage(B_FALSE); + } + + if ((zhp = zfs_open(g_zfs, argv[0], parents ? ZFS_TYPE_FILESYSTEM | + ZFS_TYPE_VOLUME : ZFS_TYPE_DATASET)) == NULL) + return (1); + + /* If we were asked and the name looks good, try to create ancestors. */ + if (parents && zfs_name_valid(argv[1], zfs_get_type(zhp)) && + zfs_create_ancestors(g_zfs, argv[1]) != 0) { + zfs_close(zhp); + return (1); + } + + ret = (zfs_rename(zhp, argv[1], recurse) != 0); + + zfs_close(zhp); + return (ret); +} + +/* + * zfs promote <fs> + * + * Promotes the given clone fs to be the parent + */ +/* ARGSUSED */ +static int +zfs_do_promote(int argc, char **argv) +{ + zfs_handle_t *zhp; + int ret; + + /* check options */ + if (argc > 1 && argv[1][0] == '-') { + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + argv[1][1]); + usage(B_FALSE); + } + + /* check number of arguments */ + if (argc < 2) { + (void) fprintf(stderr, gettext("missing clone filesystem" + " argument\n")); + usage(B_FALSE); + } + if (argc > 2) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + zhp = zfs_open(g_zfs, argv[1], ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME); + if (zhp == NULL) + return (1); + + ret = (zfs_promote(zhp) != 0); + + + zfs_close(zhp); + return (ret); +} + +/* + * zfs rollback [-rRf] <snapshot> + * + * -r Delete any intervening snapshots before doing rollback + * -R Delete any snapshots and their clones + * -f ignored for backwards compatability + * + * Given a filesystem, rollback to a specific snapshot, discarding any changes + * since then and making it the active dataset. If more recent snapshots exist, + * the command will complain unless the '-r' flag is given. + */ +typedef struct rollback_cbdata { + uint64_t cb_create; + boolean_t cb_first; + int cb_doclones; + char *cb_target; + int cb_error; + boolean_t cb_recurse; + boolean_t cb_dependent; +} rollback_cbdata_t; + +/* + * Report any snapshots more recent than the one specified. Used when '-r' is + * not specified. We reuse this same callback for the snapshot dependents - if + * 'cb_dependent' is set, then this is a dependent and we should report it + * without checking the transaction group. + */ +static int +rollback_check(zfs_handle_t *zhp, void *data) +{ + rollback_cbdata_t *cbp = data; + + if (cbp->cb_doclones) { + zfs_close(zhp); + return (0); + } + + if (!cbp->cb_dependent) { + if (strcmp(zfs_get_name(zhp), cbp->cb_target) != 0 && + zfs_get_type(zhp) == ZFS_TYPE_SNAPSHOT && + zfs_prop_get_int(zhp, ZFS_PROP_CREATETXG) > + cbp->cb_create) { + + if (cbp->cb_first && !cbp->cb_recurse) { + (void) fprintf(stderr, gettext("cannot " + "rollback to '%s': more recent snapshots " + "exist\n"), + cbp->cb_target); + (void) fprintf(stderr, gettext("use '-r' to " + "force deletion of the following " + "snapshots:\n")); + cbp->cb_first = 0; + cbp->cb_error = 1; + } + + if (cbp->cb_recurse) { + cbp->cb_dependent = B_TRUE; + if (zfs_iter_dependents(zhp, B_TRUE, + rollback_check, cbp) != 0) { + zfs_close(zhp); + return (-1); + } + cbp->cb_dependent = B_FALSE; + } else { + (void) fprintf(stderr, "%s\n", + zfs_get_name(zhp)); + } + } + } else { + if (cbp->cb_first && cbp->cb_recurse) { + (void) fprintf(stderr, gettext("cannot rollback to " + "'%s': clones of previous snapshots exist\n"), + cbp->cb_target); + (void) fprintf(stderr, gettext("use '-R' to " + "force deletion of the following clones and " + "dependents:\n")); + cbp->cb_first = 0; + cbp->cb_error = 1; + } + + (void) fprintf(stderr, "%s\n", zfs_get_name(zhp)); + } + + zfs_close(zhp); + return (0); +} + +static int +zfs_do_rollback(int argc, char **argv) +{ + int ret; + int c; + boolean_t force = B_FALSE; + rollback_cbdata_t cb = { 0 }; + zfs_handle_t *zhp, *snap; + char parentname[ZFS_MAXNAMELEN]; + char *delim; + + /* check options */ + while ((c = getopt(argc, argv, "rRf")) != -1) { + switch (c) { + case 'r': + cb.cb_recurse = 1; + break; + case 'R': + cb.cb_recurse = 1; + cb.cb_doclones = 1; + break; + case 'f': + force = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing dataset argument\n")); + usage(B_FALSE); + } + if (argc > 1) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + /* open the snapshot */ + if ((snap = zfs_open(g_zfs, argv[0], ZFS_TYPE_SNAPSHOT)) == NULL) + return (1); + + /* open the parent dataset */ + (void) strlcpy(parentname, argv[0], sizeof (parentname)); + verify((delim = strrchr(parentname, '@')) != NULL); + *delim = '\0'; + if ((zhp = zfs_open(g_zfs, parentname, ZFS_TYPE_DATASET)) == NULL) { + zfs_close(snap); + return (1); + } + + /* + * Check for more recent snapshots and/or clones based on the presence + * of '-r' and '-R'. + */ + cb.cb_target = argv[0]; + cb.cb_create = zfs_prop_get_int(snap, ZFS_PROP_CREATETXG); + cb.cb_first = B_TRUE; + cb.cb_error = 0; + if ((ret = zfs_iter_children(zhp, rollback_check, &cb)) != 0) + goto out; + + if ((ret = cb.cb_error) != 0) + goto out; + + /* + * Rollback parent to the given snapshot. + */ + ret = zfs_rollback(zhp, snap, force); + +out: + zfs_close(snap); + zfs_close(zhp); + + if (ret == 0) + return (0); + else + return (1); +} + +/* + * zfs set property=value { fs | snap | vol } ... + * + * Sets the given property for all datasets specified on the command line. + */ +typedef struct set_cbdata { + char *cb_propname; + char *cb_value; +} set_cbdata_t; + +static int +set_callback(zfs_handle_t *zhp, void *data) +{ + set_cbdata_t *cbp = data; + + if (zfs_prop_set(zhp, cbp->cb_propname, cbp->cb_value) != 0) { + switch (libzfs_errno(g_zfs)) { + case EZFS_MOUNTFAILED: + (void) fprintf(stderr, gettext("property may be set " + "but unable to remount filesystem\n")); + break; + case EZFS_SHARENFSFAILED: + (void) fprintf(stderr, gettext("property may be set " + "but unable to reshare filesystem\n")); + break; + } + return (1); + } + return (0); +} + +static int +zfs_do_set(int argc, char **argv) +{ + set_cbdata_t cb; + int ret; + + /* check for options */ + if (argc > 1 && argv[1][0] == '-') { + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + argv[1][1]); + usage(B_FALSE); + } + + /* check number of arguments */ + if (argc < 2) { + (void) fprintf(stderr, gettext("missing property=value " + "argument\n")); + usage(B_FALSE); + } + if (argc < 3) { + (void) fprintf(stderr, gettext("missing dataset name\n")); + usage(B_FALSE); + } + + /* validate property=value argument */ + cb.cb_propname = argv[1]; + if (((cb.cb_value = strchr(cb.cb_propname, '=')) == NULL) || + (cb.cb_value[1] == '\0')) { + (void) fprintf(stderr, gettext("missing value in " + "property=value argument\n")); + usage(B_FALSE); + } + + *cb.cb_value = '\0'; + cb.cb_value++; + + if (*cb.cb_propname == '\0') { + (void) fprintf(stderr, + gettext("missing property in property=value argument\n")); + usage(B_FALSE); + } + + ret = zfs_for_each(argc - 2, argv + 2, NULL, + ZFS_TYPE_DATASET, NULL, NULL, set_callback, &cb); + + return (ret); +} + +/* + * zfs snapshot [-r] [-o prop=value] ... <fs@snap> + * + * Creates a snapshot with the given name. While functionally equivalent to + * 'zfs create', it is a separate command to differentiate intent. + */ +static int +zfs_do_snapshot(int argc, char **argv) +{ + boolean_t recursive = B_FALSE; + int ret; + char c; + nvlist_t *props; + + if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) { + (void) fprintf(stderr, gettext("internal error: " + "out of memory\n")); + return (1); + } + + /* check options */ + while ((c = getopt(argc, argv, "ro:")) != -1) { + switch (c) { + case 'o': + if (parseprop(props)) + return (1); + break; + case 'r': + recursive = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + goto usage; + } + } + + argc -= optind; + argv += optind; + + /* check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing snapshot argument\n")); + goto usage; + } + if (argc > 1) { + (void) fprintf(stderr, gettext("too many arguments\n")); + goto usage; + } + + ret = zfs_snapshot(g_zfs, argv[0], recursive, props); + nvlist_free(props); + if (ret && recursive) + (void) fprintf(stderr, gettext("no snapshots were created\n")); + return (ret != 0); + +usage: + nvlist_free(props); + usage(B_FALSE); + return (-1); +} + +/* + * zfs send [-v] -R [-i|-I <@snap>] <fs@snap> + * zfs send [-v] [-i|-I <@snap>] <fs@snap> + * + * Send a backup stream to stdout. + */ +static int +zfs_do_send(int argc, char **argv) +{ + char *fromname = NULL; + char *toname = NULL; + char *cp; + zfs_handle_t *zhp; + boolean_t doall = B_FALSE; + boolean_t replicate = B_FALSE; + boolean_t fromorigin = B_FALSE; + boolean_t verbose = B_FALSE; + int c, err; + + /* check options */ + while ((c = getopt(argc, argv, ":i:I:Rv")) != -1) { + switch (c) { + case 'i': + if (fromname) + usage(B_FALSE); + fromname = optarg; + break; + case 'I': + if (fromname) + usage(B_FALSE); + fromname = optarg; + doall = B_TRUE; + break; + case 'R': + replicate = B_TRUE; + break; + case 'v': + verbose = B_TRUE; + break; + case ':': + (void) fprintf(stderr, gettext("missing argument for " + "'%c' option\n"), optopt); + usage(B_FALSE); + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing snapshot argument\n")); + usage(B_FALSE); + } + if (argc > 1) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + if (isatty(STDOUT_FILENO)) { + (void) fprintf(stderr, + gettext("Error: Stream can not be written to a terminal.\n" + "You must redirect standard output.\n")); + return (1); + } + + cp = strchr(argv[0], '@'); + if (cp == NULL) { + (void) fprintf(stderr, + gettext("argument must be a snapshot\n")); + usage(B_FALSE); + } + *cp = '\0'; + toname = cp + 1; + zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME); + if (zhp == NULL) + return (1); + + /* + * If they specified the full path to the snapshot, chop off + * everything except the short name of the snapshot, but special + * case if they specify the origin. + */ + if (fromname && (cp = strchr(fromname, '@')) != NULL) { + char origin[ZFS_MAXNAMELEN]; + zprop_source_t src; + + (void) zfs_prop_get(zhp, ZFS_PROP_ORIGIN, + origin, sizeof (origin), &src, NULL, 0, B_FALSE); + + if (strcmp(origin, fromname) == 0) { + fromname = NULL; + fromorigin = B_TRUE; + } else { + *cp = '\0'; + if (cp != fromname && strcmp(argv[0], fromname)) { + (void) fprintf(stderr, + gettext("incremental source must be " + "in same filesystem\n")); + usage(B_FALSE); + } + fromname = cp + 1; + if (strchr(fromname, '@') || strchr(fromname, '/')) { + (void) fprintf(stderr, + gettext("invalid incremental source\n")); + usage(B_FALSE); + } + } + } + + if (replicate && fromname == NULL) + doall = B_TRUE; + + err = zfs_send(zhp, fromname, toname, replicate, doall, fromorigin, + verbose, STDOUT_FILENO); + zfs_close(zhp); + + return (err != 0); +} + +/* + * zfs receive [-dnvF] <fs@snap> + * + * Restore a backup stream from stdin. + */ +static int +zfs_do_receive(int argc, char **argv) +{ + int c, err; + recvflags_t flags; + + bzero(&flags, sizeof (recvflags_t)); + /* check options */ + while ((c = getopt(argc, argv, ":dnvF")) != -1) { + switch (c) { + case 'd': + flags.isprefix = B_TRUE; + break; + case 'n': + flags.dryrun = B_TRUE; + break; + case 'v': + flags.verbose = B_TRUE; + break; + case 'F': + flags.force = B_TRUE; + break; + case ':': + (void) fprintf(stderr, gettext("missing argument for " + "'%c' option\n"), optopt); + usage(B_FALSE); + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing snapshot argument\n")); + usage(B_FALSE); + } + if (argc > 1) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + if (isatty(STDIN_FILENO)) { + (void) fprintf(stderr, + gettext("Error: Backup stream can not be read " + "from a terminal.\n" + "You must redirect standard input.\n")); + return (1); + } + + err = zfs_receive(g_zfs, argv[0], flags, STDIN_FILENO, NULL); + + return (err != 0); +} + +typedef struct allow_cb { + int a_permcnt; + size_t a_treeoffset; +} allow_cb_t; + +static void +zfs_print_perms(avl_tree_t *tree) +{ + zfs_perm_node_t *permnode; + + permnode = avl_first(tree); + while (permnode != NULL) { + (void) printf("%s", permnode->z_pname); + permnode = AVL_NEXT(tree, permnode); + if (permnode) + (void) printf(","); + else + (void) printf("\n"); + } +} + +/* + * Iterate over user/groups/everyone/... and the call perm_iter + * function to print actual permission when tree has >0 nodes. + */ +static void +zfs_iter_perms(avl_tree_t *tree, const char *banner, allow_cb_t *cb) +{ + zfs_allow_node_t *item; + avl_tree_t *ptree; + + item = avl_first(tree); + while (item) { + ptree = (void *)((char *)item + cb->a_treeoffset); + if (avl_numnodes(ptree)) { + if (cb->a_permcnt++ == 0) + (void) printf("%s\n", banner); + (void) printf("\t%s", item->z_key); + /* + * Avoid an extra space being printed + * for "everyone" which is keyed with a null + * string + */ + if (item->z_key[0] != '\0') + (void) printf(" "); + zfs_print_perms(ptree); + } + item = AVL_NEXT(tree, item); + } +} + +#define LINES "-------------------------------------------------------------\n" +static int +zfs_print_allows(char *ds) +{ + zfs_allow_t *curperms, *perms; + zfs_handle_t *zhp; + allow_cb_t allowcb = { 0 }; + char banner[MAXPATHLEN]; + + if (ds[0] == '-') + usage(B_FALSE); + + if (strrchr(ds, '@')) { + (void) fprintf(stderr, gettext("Snapshots don't have 'allow'" + " permissions\n")); + return (1); + } + if ((zhp = zfs_open(g_zfs, ds, ZFS_TYPE_DATASET)) == NULL) + return (1); + + if (zfs_perm_get(zhp, &perms)) { + (void) fprintf(stderr, + gettext("Failed to retrieve 'allows' on %s\n"), ds); + zfs_close(zhp); + return (1); + } + + zfs_close(zhp); + + if (perms != NULL) + (void) printf("%s", LINES); + for (curperms = perms; curperms; curperms = curperms->z_next) { + + (void) snprintf(banner, sizeof (banner), + "Permission sets on (%s)", curperms->z_setpoint); + allowcb.a_treeoffset = + offsetof(zfs_allow_node_t, z_localdescend); + allowcb.a_permcnt = 0; + zfs_iter_perms(&curperms->z_sets, banner, &allowcb); + + (void) snprintf(banner, sizeof (banner), + "Create time permissions on (%s)", curperms->z_setpoint); + allowcb.a_treeoffset = + offsetof(zfs_allow_node_t, z_localdescend); + allowcb.a_permcnt = 0; + zfs_iter_perms(&curperms->z_crperms, banner, &allowcb); + + + (void) snprintf(banner, sizeof (banner), + "Local permissions on (%s)", curperms->z_setpoint); + allowcb.a_treeoffset = offsetof(zfs_allow_node_t, z_local); + allowcb.a_permcnt = 0; + zfs_iter_perms(&curperms->z_user, banner, &allowcb); + zfs_iter_perms(&curperms->z_group, banner, &allowcb); + zfs_iter_perms(&curperms->z_everyone, banner, &allowcb); + + (void) snprintf(banner, sizeof (banner), + "Descendent permissions on (%s)", curperms->z_setpoint); + allowcb.a_treeoffset = offsetof(zfs_allow_node_t, z_descend); + allowcb.a_permcnt = 0; + zfs_iter_perms(&curperms->z_user, banner, &allowcb); + zfs_iter_perms(&curperms->z_group, banner, &allowcb); + zfs_iter_perms(&curperms->z_everyone, banner, &allowcb); + + (void) snprintf(banner, sizeof (banner), + "Local+Descendent permissions on (%s)", + curperms->z_setpoint); + allowcb.a_treeoffset = + offsetof(zfs_allow_node_t, z_localdescend); + allowcb.a_permcnt = 0; + zfs_iter_perms(&curperms->z_user, banner, &allowcb); + zfs_iter_perms(&curperms->z_group, banner, &allowcb); + zfs_iter_perms(&curperms->z_everyone, banner, &allowcb); + + (void) printf("%s", LINES); + } + zfs_free_allows(perms); + return (0); +} + +#define ALLOWOPTIONS "ldcsu:g:e" +#define UNALLOWOPTIONS "ldcsu:g:er" + +/* + * Validate options, and build necessary datastructure to display/remove/add + * permissions. + * Returns 0 - If permissions should be added/removed + * Returns 1 - If permissions should be displayed. + * Returns -1 - on failure + */ +int +parse_allow_args(int *argc, char **argv[], boolean_t unallow, + char **ds, int *recurse, nvlist_t **zperms) +{ + int c; + char *options = unallow ? UNALLOWOPTIONS : ALLOWOPTIONS; + zfs_deleg_inherit_t deleg_type = ZFS_DELEG_NONE; + zfs_deleg_who_type_t who_type = ZFS_DELEG_WHO_UNKNOWN; + char *who = NULL; + char *perms = NULL; + zfs_handle_t *zhp; + + while ((c = getopt(*argc, *argv, options)) != -1) { + switch (c) { + case 'l': + if (who_type == ZFS_DELEG_CREATE || + who_type == ZFS_DELEG_NAMED_SET) + usage(B_FALSE); + + deleg_type |= ZFS_DELEG_PERM_LOCAL; + break; + case 'd': + if (who_type == ZFS_DELEG_CREATE || + who_type == ZFS_DELEG_NAMED_SET) + usage(B_FALSE); + + deleg_type |= ZFS_DELEG_PERM_DESCENDENT; + break; + case 'r': + *recurse = B_TRUE; + break; + case 'c': + if (who_type != ZFS_DELEG_WHO_UNKNOWN) + usage(B_FALSE); + if (deleg_type) + usage(B_FALSE); + who_type = ZFS_DELEG_CREATE; + break; + case 's': + if (who_type != ZFS_DELEG_WHO_UNKNOWN) + usage(B_FALSE); + if (deleg_type) + usage(B_FALSE); + who_type = ZFS_DELEG_NAMED_SET; + break; + case 'u': + if (who_type != ZFS_DELEG_WHO_UNKNOWN) + usage(B_FALSE); + who_type = ZFS_DELEG_USER; + who = optarg; + break; + case 'g': + if (who_type != ZFS_DELEG_WHO_UNKNOWN) + usage(B_FALSE); + who_type = ZFS_DELEG_GROUP; + who = optarg; + break; + case 'e': + if (who_type != ZFS_DELEG_WHO_UNKNOWN) + usage(B_FALSE); + who_type = ZFS_DELEG_EVERYONE; + break; + default: + usage(B_FALSE); + break; + } + } + + if (deleg_type == 0) + deleg_type = ZFS_DELEG_PERM_LOCALDESCENDENT; + + *argc -= optind; + *argv += optind; + + if (unallow == B_FALSE && *argc == 1) { + /* + * Only print permissions if no options were processed + */ + if (optind == 1) + return (1); + else + usage(B_FALSE); + } + + /* + * initialize variables for zfs_build_perms based on number + * of arguments. + * 3 arguments ==> zfs [un]allow joe perm,perm,perm <dataset> or + * zfs [un]allow -s @set1 perm,perm <dataset> + * 2 arguments ==> zfs [un]allow -c perm,perm <dataset> or + * zfs [un]allow -u|-g <name> perm <dataset> or + * zfs [un]allow -e perm,perm <dataset> + * zfs unallow joe <dataset> + * zfs unallow -s @set1 <dataset> + * 1 argument ==> zfs [un]allow -e <dataset> or + * zfs [un]allow -c <dataset> + */ + + switch (*argc) { + case 3: + perms = (*argv)[1]; + who = (*argv)[0]; + *ds = (*argv)[2]; + + /* + * advance argc/argv for do_allow cases. + * for do_allow case make sure who have a know who type + * and its not a permission set. + */ + if (unallow == B_TRUE) { + *argc -= 2; + *argv += 2; + } else if (who_type != ZFS_DELEG_WHO_UNKNOWN && + who_type != ZFS_DELEG_NAMED_SET) + usage(B_FALSE); + break; + + case 2: + if (unallow == B_TRUE && (who_type == ZFS_DELEG_EVERYONE || + who_type == ZFS_DELEG_CREATE || who != NULL)) { + perms = (*argv)[0]; + *ds = (*argv)[1]; + } else { + if (unallow == B_FALSE && + (who_type == ZFS_DELEG_WHO_UNKNOWN || + who_type == ZFS_DELEG_NAMED_SET)) + usage(B_FALSE); + else if (who_type == ZFS_DELEG_WHO_UNKNOWN || + who_type == ZFS_DELEG_NAMED_SET) + who = (*argv)[0]; + else if (who_type != ZFS_DELEG_NAMED_SET) + perms = (*argv)[0]; + *ds = (*argv)[1]; + } + if (unallow == B_TRUE) { + (*argc)--; + (*argv)++; + } + break; + + case 1: + if (unallow == B_FALSE) + usage(B_FALSE); + if (who == NULL && who_type != ZFS_DELEG_CREATE && + who_type != ZFS_DELEG_EVERYONE) + usage(B_FALSE); + *ds = (*argv)[0]; + break; + + default: + usage(B_FALSE); + } + + if (strrchr(*ds, '@')) { + (void) fprintf(stderr, + gettext("Can't set or remove 'allow' permissions " + "on snapshots.\n")); + return (-1); + } + + if ((zhp = zfs_open(g_zfs, *ds, ZFS_TYPE_DATASET)) == NULL) + return (-1); + + if ((zfs_build_perms(zhp, who, perms, + who_type, deleg_type, zperms)) != 0) { + zfs_close(zhp); + return (-1); + } + zfs_close(zhp); + return (0); +} + +static int +zfs_do_allow(int argc, char **argv) +{ + char *ds; + nvlist_t *zperms = NULL; + zfs_handle_t *zhp; + int unused; + int ret; + + if ((ret = parse_allow_args(&argc, &argv, B_FALSE, &ds, + &unused, &zperms)) == -1) + return (1); + + if (ret == 1) + return (zfs_print_allows(argv[0])); + + if ((zhp = zfs_open(g_zfs, ds, ZFS_TYPE_DATASET)) == NULL) + return (1); + + if (zfs_perm_set(zhp, zperms)) { + zfs_close(zhp); + nvlist_free(zperms); + return (1); + } + nvlist_free(zperms); + zfs_close(zhp); + + return (0); +} + +static int +unallow_callback(zfs_handle_t *zhp, void *data) +{ + nvlist_t *nvp = (nvlist_t *)data; + int error; + + error = zfs_perm_remove(zhp, nvp); + if (error) { + (void) fprintf(stderr, gettext("Failed to remove permissions " + "on %s\n"), zfs_get_name(zhp)); + } + return (error); +} + +static int +zfs_do_unallow(int argc, char **argv) +{ + int recurse = B_FALSE; + char *ds; + int error; + nvlist_t *zperms = NULL; + int flags = 0; + + if (parse_allow_args(&argc, &argv, B_TRUE, + &ds, &recurse, &zperms) == -1) + return (1); + + if (recurse) + flags |= ZFS_ITER_RECURSE; + error = zfs_for_each(argc, argv, flags, + ZFS_TYPE_FILESYSTEM|ZFS_TYPE_VOLUME, NULL, + NULL, unallow_callback, (void *)zperms); + + if (zperms) + nvlist_free(zperms); + + return (error); +} + +typedef struct get_all_cbdata { + zfs_handle_t **cb_handles; + size_t cb_alloc; + size_t cb_used; + uint_t cb_types; + boolean_t cb_verbose; +} get_all_cbdata_t; + +#define CHECK_SPINNER 30 +#define SPINNER_TIME 3 /* seconds */ +#define MOUNT_TIME 5 /* seconds */ + +static int +get_one_dataset(zfs_handle_t *zhp, void *data) +{ + static char spin[] = { '-', '\\', '|', '/' }; + static int spinval = 0; + static int spincheck = 0; + static time_t last_spin_time = (time_t)0; + get_all_cbdata_t *cbp = data; + zfs_type_t type = zfs_get_type(zhp); + + if (cbp->cb_verbose) { + if (--spincheck < 0) { + time_t now = time(NULL); + if (last_spin_time + SPINNER_TIME < now) { + (void) printf("\b%c", spin[spinval++ % 4]); + (void) fflush(stdout); + last_spin_time = now; + } + spincheck = CHECK_SPINNER; + } + } + + /* + * Interate over any nested datasets. + */ + if (type == ZFS_TYPE_FILESYSTEM && + zfs_iter_filesystems(zhp, get_one_dataset, data) != 0) { + zfs_close(zhp); + return (1); + } + + /* + * Skip any datasets whose type does not match. + */ + if ((type & cbp->cb_types) == 0) { + zfs_close(zhp); + return (0); + } + + if (cbp->cb_alloc == cbp->cb_used) { + zfs_handle_t **handles; + + if (cbp->cb_alloc == 0) + cbp->cb_alloc = 64; + else + cbp->cb_alloc *= 2; + + handles = safe_malloc(cbp->cb_alloc * sizeof (void *)); + + if (cbp->cb_handles) { + bcopy(cbp->cb_handles, handles, + cbp->cb_used * sizeof (void *)); + free(cbp->cb_handles); + } + + cbp->cb_handles = handles; + } + + cbp->cb_handles[cbp->cb_used++] = zhp; + + return (0); +} + +static void +get_all_datasets(uint_t types, zfs_handle_t ***dslist, size_t *count, + boolean_t verbose) +{ + get_all_cbdata_t cb = { 0 }; + cb.cb_types = types; + cb.cb_verbose = verbose; + + if (verbose) { + (void) printf("%s: *", gettext("Reading ZFS config")); + (void) fflush(stdout); + } + + (void) zfs_iter_root(g_zfs, get_one_dataset, &cb); + + *dslist = cb.cb_handles; + *count = cb.cb_used; + + if (verbose) { + (void) printf("\b%s\n", gettext("done.")); + } +} + +static int +dataset_cmp(const void *a, const void *b) +{ + zfs_handle_t **za = (zfs_handle_t **)a; + zfs_handle_t **zb = (zfs_handle_t **)b; + char mounta[MAXPATHLEN]; + char mountb[MAXPATHLEN]; + boolean_t gota, gotb; + + if ((gota = (zfs_get_type(*za) == ZFS_TYPE_FILESYSTEM)) != 0) + verify(zfs_prop_get(*za, ZFS_PROP_MOUNTPOINT, mounta, + sizeof (mounta), NULL, NULL, 0, B_FALSE) == 0); + if ((gotb = (zfs_get_type(*zb) == ZFS_TYPE_FILESYSTEM)) != 0) + verify(zfs_prop_get(*zb, ZFS_PROP_MOUNTPOINT, mountb, + sizeof (mountb), NULL, NULL, 0, B_FALSE) == 0); + + if (gota && gotb) + return (strcmp(mounta, mountb)); + + if (gota) + return (-1); + if (gotb) + return (1); + + return (strcmp(zfs_get_name(a), zfs_get_name(b))); +} + +/* + * Generic callback for sharing or mounting filesystems. Because the code is so + * similar, we have a common function with an extra parameter to determine which + * mode we are using. + */ +#define OP_SHARE 0x1 +#define OP_MOUNT 0x2 + +/* + * Share or mount a dataset. + */ +static int +share_mount_one(zfs_handle_t *zhp, int op, int flags, char *protocol, + boolean_t explicit, const char *options) +{ + char mountpoint[ZFS_MAXPROPLEN]; + char shareopts[ZFS_MAXPROPLEN]; + char smbshareopts[ZFS_MAXPROPLEN]; + const char *cmdname = op == OP_SHARE ? "share" : "mount"; + struct mnttab mnt; + uint64_t zoned, canmount; + zfs_type_t type = zfs_get_type(zhp); + boolean_t shared_nfs, shared_smb; + + assert(type & (ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME)); + + if (type == ZFS_TYPE_FILESYSTEM) { + /* + * Check to make sure we can mount/share this dataset. If we + * are in the global zone and the filesystem is exported to a + * local zone, or if we are in a local zone and the + * filesystem is not exported, then it is an error. + */ + zoned = zfs_prop_get_int(zhp, ZFS_PROP_ZONED); + + if (zoned && getzoneid() == GLOBAL_ZONEID) { + if (!explicit) + return (0); + + (void) fprintf(stderr, gettext("cannot %s '%s': " + "dataset is exported to a local zone\n"), cmdname, + zfs_get_name(zhp)); + return (1); + + } else if (!zoned && getzoneid() != GLOBAL_ZONEID) { + if (!explicit) + return (0); + + (void) fprintf(stderr, gettext("cannot %s '%s': " + "permission denied\n"), cmdname, + zfs_get_name(zhp)); + return (1); + } + + /* + * Ignore any filesystems which don't apply to us. This + * includes those with a legacy mountpoint, or those with + * legacy share options. + */ + verify(zfs_prop_get(zhp, ZFS_PROP_MOUNTPOINT, mountpoint, + sizeof (mountpoint), NULL, NULL, 0, B_FALSE) == 0); + verify(zfs_prop_get(zhp, ZFS_PROP_SHARENFS, shareopts, + sizeof (shareopts), NULL, NULL, 0, B_FALSE) == 0); + verify(zfs_prop_get(zhp, ZFS_PROP_SHARESMB, smbshareopts, + sizeof (smbshareopts), NULL, NULL, 0, B_FALSE) == 0); + canmount = zfs_prop_get_int(zhp, ZFS_PROP_CANMOUNT); + + if (op == OP_SHARE && strcmp(shareopts, "off") == 0 && + strcmp(smbshareopts, "off") == 0) { + if (!explicit) + return (0); + + (void) fprintf(stderr, gettext("cannot share '%s': " + "legacy share\n"), zfs_get_name(zhp)); + (void) fprintf(stderr, gettext("use share(1M) to " + "share this filesystem\n")); + return (1); + } + + /* + * We cannot share or mount legacy filesystems. If the + * shareopts is non-legacy but the mountpoint is legacy, we + * treat it as a legacy share. + */ + if (strcmp(mountpoint, "legacy") == 0) { + if (!explicit) + return (0); + + (void) fprintf(stderr, gettext("cannot %s '%s': " + "legacy mountpoint\n"), cmdname, zfs_get_name(zhp)); + (void) fprintf(stderr, gettext("use %s(1M) to " + "%s this filesystem\n"), cmdname, cmdname); + return (1); + } + + if (strcmp(mountpoint, "none") == 0) { + if (!explicit) + return (0); + + (void) fprintf(stderr, gettext("cannot %s '%s': no " + "mountpoint set\n"), cmdname, zfs_get_name(zhp)); + return (1); + } + + /* + * canmount explicit outcome + * on no pass through + * on yes pass through + * off no return 0 + * off yes display error, return 1 + * noauto no return 0 + * noauto yes pass through + */ + if (canmount == ZFS_CANMOUNT_OFF) { + if (!explicit) + return (0); + + (void) fprintf(stderr, gettext("cannot %s '%s': " + "'canmount' property is set to 'off'\n"), cmdname, + zfs_get_name(zhp)); + return (1); + } else if (canmount == ZFS_CANMOUNT_NOAUTO && !explicit) { + return (0); + } + + /* + * At this point, we have verified that the mountpoint and/or + * shareopts are appropriate for auto management. If the + * filesystem is already mounted or shared, return (failing + * for explicit requests); otherwise mount or share the + * filesystem. + */ + switch (op) { + case OP_SHARE: + + shared_nfs = zfs_is_shared_nfs(zhp, NULL); + shared_smb = zfs_is_shared_smb(zhp, NULL); + + if (shared_nfs && shared_smb || + (shared_nfs && strcmp(shareopts, "on") == 0 && + strcmp(smbshareopts, "off") == 0) || + (shared_smb && strcmp(smbshareopts, "on") == 0 && + strcmp(shareopts, "off") == 0)) { + if (!explicit) + return (0); + + (void) fprintf(stderr, gettext("cannot share " + "'%s': filesystem already shared\n"), + zfs_get_name(zhp)); + return (1); + } + + if (!zfs_is_mounted(zhp, NULL) && + zfs_mount(zhp, NULL, 0) != 0) + return (1); + + if (protocol == NULL) { + if (zfs_shareall(zhp) != 0) + return (1); + } else if (strcmp(protocol, "nfs") == 0) { + if (zfs_share_nfs(zhp)) + return (1); + } else if (strcmp(protocol, "smb") == 0) { + if (zfs_share_smb(zhp)) + return (1); + } else { + (void) fprintf(stderr, gettext("cannot share " + "'%s': invalid share type '%s' " + "specified\n"), + zfs_get_name(zhp), protocol); + return (1); + } + + break; + + case OP_MOUNT: + if (options == NULL) + mnt.mnt_mntopts = ""; + else + mnt.mnt_mntopts = (char *)options; + + if (!hasmntopt(&mnt, MNTOPT_REMOUNT) && + zfs_is_mounted(zhp, NULL)) { + if (!explicit) + return (0); + + (void) fprintf(stderr, gettext("cannot mount " + "'%s': filesystem already mounted\n"), + zfs_get_name(zhp)); + return (1); + } + + if (zfs_mount(zhp, options, flags) != 0) + return (1); + break; + } + } else { + assert(op == OP_SHARE); + + /* + * Ignore any volumes that aren't shared. + */ + verify(zfs_prop_get(zhp, ZFS_PROP_SHAREISCSI, shareopts, + sizeof (shareopts), NULL, NULL, 0, B_FALSE) == 0); + + if (strcmp(shareopts, "off") == 0) { + if (!explicit) + return (0); + + (void) fprintf(stderr, gettext("cannot share '%s': " + "'shareiscsi' property not set\n"), + zfs_get_name(zhp)); + (void) fprintf(stderr, gettext("set 'shareiscsi' " + "property or use iscsitadm(1M) to share this " + "volume\n")); + return (1); + } + + if (zfs_is_shared_iscsi(zhp)) { + if (!explicit) + return (0); + + (void) fprintf(stderr, gettext("cannot share " + "'%s': volume already shared\n"), + zfs_get_name(zhp)); + return (1); + } + + if (zfs_share_iscsi(zhp) != 0) + return (1); + } + + return (0); +} + +/* + * Reports progress in the form "(current/total)". Not thread-safe. + */ +static void +report_mount_progress(int current, int total) +{ + static int len; + static char *reverse = "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b" + "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b"; + static time_t last_progress_time; + time_t now = time(NULL); + + /* report 1..n instead of 0..n-1 */ + ++current; + + /* display header if we're here for the first time */ + if (current == 1) { + (void) printf(gettext("Mounting ZFS filesystems: ")); + len = 0; + } else if (current != total && last_progress_time + MOUNT_TIME >= now) { + /* too soon to report again */ + return; + } + + last_progress_time = now; + + /* back up to prepare for overwriting */ + if (len) + (void) printf("%*.*s", len, len, reverse); + + /* We put a newline at the end if this is the last one. */ + len = printf("(%d/%d)%s", current, total, current == total ? "\n" : ""); + (void) fflush(stdout); +} + +static void +append_options(char *mntopts, char *newopts) +{ + int len = strlen(mntopts); + + /* original length plus new string to append plus 1 for the comma */ + if (len + 1 + strlen(newopts) >= MNT_LINE_MAX) { + (void) fprintf(stderr, gettext("the opts argument for " + "'%c' option is too long (more than %d chars)\n"), + "-o", MNT_LINE_MAX); + usage(B_FALSE); + } + + if (*mntopts) + mntopts[len++] = ','; + + (void) strcpy(&mntopts[len], newopts); +} + +static int +share_mount(int op, int argc, char **argv) +{ + int do_all = 0; + boolean_t verbose = B_FALSE; + int c, ret = 0; + char *options = NULL; + int types, flags = 0; + + /* check options */ + while ((c = getopt(argc, argv, op == OP_MOUNT ? ":avo:O" : "a")) + != -1) { + switch (c) { + case 'a': + do_all = 1; + break; + case 'v': + verbose = B_TRUE; + break; + case 'o': + if (*optarg == '\0') { + (void) fprintf(stderr, gettext("empty mount " + "options (-o) specified\n")); + usage(B_FALSE); + } + + if (options == NULL) + options = safe_malloc(MNT_LINE_MAX + 1); + + /* option validation is done later */ + append_options(options, optarg); + break; + + case 'O': + flags |= MS_OVERLAY; + break; + case ':': + (void) fprintf(stderr, gettext("missing argument for " + "'%c' option\n"), optopt); + usage(B_FALSE); + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* check number of arguments */ + if (do_all) { + zfs_handle_t **dslist = NULL; + size_t i, count = 0; + char *protocol = NULL; + + if (op == OP_MOUNT) { + types = ZFS_TYPE_FILESYSTEM; + } else if (argc > 0) { + if (strcmp(argv[0], "nfs") == 0 || + strcmp(argv[0], "smb") == 0) { + types = ZFS_TYPE_FILESYSTEM; + } else if (strcmp(argv[0], "iscsi") == 0) { + types = ZFS_TYPE_VOLUME; + } else { + (void) fprintf(stderr, gettext("share type " + "must be 'nfs', 'smb' or 'iscsi'\n")); + usage(B_FALSE); + } + protocol = argv[0]; + argc--; + argv++; + } else { + types = ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME; + } + + if (argc != 0) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + get_all_datasets(types, &dslist, &count, verbose); + + if (count == 0) + return (0); + + qsort(dslist, count, sizeof (void *), dataset_cmp); + + for (i = 0; i < count; i++) { + if (verbose) + report_mount_progress(i, count); + + if (share_mount_one(dslist[i], op, flags, protocol, + B_FALSE, options) != 0) + ret = 1; + zfs_close(dslist[i]); + } + + free(dslist); + } else if (argc == 0) { + struct mnttab entry; + + if ((op == OP_SHARE) || (options != NULL)) { + (void) fprintf(stderr, gettext("missing filesystem " + "argument (specify -a for all)\n")); + usage(B_FALSE); + } + + /* + * When mount is given no arguments, go through /etc/mnttab and + * display any active ZFS mounts. We hide any snapshots, since + * they are controlled automatically. + */ + rewind(mnttab_file); + while (getmntent(mnttab_file, &entry) == 0) { + if (strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0 || + strchr(entry.mnt_special, '@') != NULL) + continue; + + (void) printf("%-30s %s\n", entry.mnt_special, + entry.mnt_mountp); + } + + } else { + zfs_handle_t *zhp; + + types = ZFS_TYPE_FILESYSTEM; + if (op == OP_SHARE) + types |= ZFS_TYPE_VOLUME; + + if (argc > 1) { + (void) fprintf(stderr, + gettext("too many arguments\n")); + usage(B_FALSE); + } + + if ((zhp = zfs_open(g_zfs, argv[0], types)) == NULL) { + ret = 1; + } else { + ret = share_mount_one(zhp, op, flags, NULL, B_TRUE, + options); + zfs_close(zhp); + } + } + + return (ret); +} + +/* + * zfs mount -a [nfs | iscsi] + * zfs mount filesystem + * + * Mount all filesystems, or mount the given filesystem. + */ +static int +zfs_do_mount(int argc, char **argv) +{ + return (share_mount(OP_MOUNT, argc, argv)); +} + +/* + * zfs share -a [nfs | iscsi | smb] + * zfs share filesystem + * + * Share all filesystems, or share the given filesystem. + */ +static int +zfs_do_share(int argc, char **argv) +{ + return (share_mount(OP_SHARE, argc, argv)); +} + +typedef struct unshare_unmount_node { + zfs_handle_t *un_zhp; + char *un_mountp; + uu_avl_node_t un_avlnode; +} unshare_unmount_node_t; + +/* ARGSUSED */ +static int +unshare_unmount_compare(const void *larg, const void *rarg, void *unused) +{ + const unshare_unmount_node_t *l = larg; + const unshare_unmount_node_t *r = rarg; + + return (strcmp(l->un_mountp, r->un_mountp)); +} + +/* + * Convenience routine used by zfs_do_umount() and manual_unmount(). Given an + * absolute path, find the entry /etc/mnttab, verify that its a ZFS filesystem, + * and unmount it appropriately. + */ +static int +unshare_unmount_path(int op, char *path, int flags, boolean_t is_manual) +{ + zfs_handle_t *zhp; + int ret; + struct stat64 statbuf; + struct extmnttab entry; + const char *cmdname = (op == OP_SHARE) ? "unshare" : "unmount"; + ino_t path_inode; + + /* + * Search for the path in /etc/mnttab. Rather than looking for the + * specific path, which can be fooled by non-standard paths (i.e. ".." + * or "//"), we stat() the path and search for the corresponding + * (major,minor) device pair. + */ + if (stat64(path, &statbuf) != 0) { + (void) fprintf(stderr, gettext("cannot %s '%s': %s\n"), + cmdname, path, strerror(errno)); + return (1); + } + path_inode = statbuf.st_ino; + + /* + * Search for the given (major,minor) pair in the mount table. + */ + rewind(mnttab_file); + while ((ret = getextmntent(mnttab_file, &entry, 0)) == 0) { + if (entry.mnt_major == major(statbuf.st_dev) && + entry.mnt_minor == minor(statbuf.st_dev)) + break; + } + if (ret != 0) { + if (op == OP_SHARE) { + (void) fprintf(stderr, gettext("cannot %s '%s': not " + "currently mounted\n"), cmdname, path); + return (1); + } + (void) fprintf(stderr, gettext("warning: %s not in mnttab\n"), + path); + if ((ret = umount2(path, flags)) != 0) + (void) fprintf(stderr, gettext("%s: %s\n"), path, + strerror(errno)); + return (ret != 0); + } + + if (strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0) { + (void) fprintf(stderr, gettext("cannot %s '%s': not a ZFS " + "filesystem\n"), cmdname, path); + return (1); + } + + if ((zhp = zfs_open(g_zfs, entry.mnt_special, + ZFS_TYPE_FILESYSTEM)) == NULL) + return (1); + + ret = 1; + if (stat64(entry.mnt_mountp, &statbuf) != 0) { + (void) fprintf(stderr, gettext("cannot %s '%s': %s\n"), + cmdname, path, strerror(errno)); + goto out; + } else if (statbuf.st_ino != path_inode) { + (void) fprintf(stderr, gettext("cannot " + "%s '%s': not a mountpoint\n"), cmdname, path); + goto out; + } + + if (op == OP_SHARE) { + char nfs_mnt_prop[ZFS_MAXPROPLEN]; + char smbshare_prop[ZFS_MAXPROPLEN]; + + verify(zfs_prop_get(zhp, ZFS_PROP_SHARENFS, nfs_mnt_prop, + sizeof (nfs_mnt_prop), NULL, NULL, 0, B_FALSE) == 0); + verify(zfs_prop_get(zhp, ZFS_PROP_SHARESMB, smbshare_prop, + sizeof (smbshare_prop), NULL, NULL, 0, B_FALSE) == 0); + + if (strcmp(nfs_mnt_prop, "off") == 0 && + strcmp(smbshare_prop, "off") == 0) { + (void) fprintf(stderr, gettext("cannot unshare " + "'%s': legacy share\n"), path); + (void) fprintf(stderr, gettext("use " + "unshare(1M) to unshare this filesystem\n")); + } else if (!zfs_is_shared(zhp)) { + (void) fprintf(stderr, gettext("cannot unshare '%s': " + "not currently shared\n"), path); + } else { + ret = zfs_unshareall_bypath(zhp, path); + } + } else { + char mtpt_prop[ZFS_MAXPROPLEN]; + + verify(zfs_prop_get(zhp, ZFS_PROP_MOUNTPOINT, mtpt_prop, + sizeof (mtpt_prop), NULL, NULL, 0, B_FALSE) == 0); + + if (is_manual) { + ret = zfs_unmount(zhp, NULL, flags); + } else if (strcmp(mtpt_prop, "legacy") == 0) { + (void) fprintf(stderr, gettext("cannot unmount " + "'%s': legacy mountpoint\n"), + zfs_get_name(zhp)); + (void) fprintf(stderr, gettext("use umount(1M) " + "to unmount this filesystem\n")); + } else { + ret = zfs_unmountall(zhp, flags); + } + } + +out: + zfs_close(zhp); + + return (ret != 0); +} + +/* + * Generic callback for unsharing or unmounting a filesystem. + */ +static int +unshare_unmount(int op, int argc, char **argv) +{ + int do_all = 0; + int flags = 0; + int ret = 0; + int types, c; + zfs_handle_t *zhp; + char nfsiscsi_mnt_prop[ZFS_MAXPROPLEN]; + char sharesmb[ZFS_MAXPROPLEN]; + + /* check options */ + while ((c = getopt(argc, argv, op == OP_SHARE ? "a" : "af")) != -1) { + switch (c) { + case 'a': + do_all = 1; + break; + case 'f': + flags = MS_FORCE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + if (do_all) { + /* + * We could make use of zfs_for_each() to walk all datasets in + * the system, but this would be very inefficient, especially + * since we would have to linearly search /etc/mnttab for each + * one. Instead, do one pass through /etc/mnttab looking for + * zfs entries and call zfs_unmount() for each one. + * + * Things get a little tricky if the administrator has created + * mountpoints beneath other ZFS filesystems. In this case, we + * have to unmount the deepest filesystems first. To accomplish + * this, we place all the mountpoints in an AVL tree sorted by + * the special type (dataset name), and walk the result in + * reverse to make sure to get any snapshots first. + */ + struct mnttab entry; + uu_avl_pool_t *pool; + uu_avl_t *tree; + unshare_unmount_node_t *node; + uu_avl_index_t idx; + uu_avl_walk_t *walk; + + if (argc != 0) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + if ((pool = uu_avl_pool_create("unmount_pool", + sizeof (unshare_unmount_node_t), + offsetof(unshare_unmount_node_t, un_avlnode), + unshare_unmount_compare, + UU_DEFAULT)) == NULL) { + (void) fprintf(stderr, gettext("internal error: " + "out of memory\n")); + exit(1); + } + + if ((tree = uu_avl_create(pool, NULL, UU_DEFAULT)) == NULL) { + (void) fprintf(stderr, gettext("internal error: " + "out of memory\n")); + exit(1); + } + + rewind(mnttab_file); + while (getmntent(mnttab_file, &entry) == 0) { + + /* ignore non-ZFS entries */ + if (strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0) + continue; + + /* ignore snapshots */ + if (strchr(entry.mnt_special, '@') != NULL) + continue; + + if ((zhp = zfs_open(g_zfs, entry.mnt_special, + ZFS_TYPE_FILESYSTEM)) == NULL) { + ret = 1; + continue; + } + + switch (op) { + case OP_SHARE: + verify(zfs_prop_get(zhp, ZFS_PROP_SHARENFS, + nfsiscsi_mnt_prop, + sizeof (nfsiscsi_mnt_prop), + NULL, NULL, 0, B_FALSE) == 0); + if (strcmp(nfsiscsi_mnt_prop, "off") != 0) + break; + verify(zfs_prop_get(zhp, ZFS_PROP_SHARESMB, + nfsiscsi_mnt_prop, + sizeof (nfsiscsi_mnt_prop), + NULL, NULL, 0, B_FALSE) == 0); + if (strcmp(nfsiscsi_mnt_prop, "off") == 0) + continue; + break; + case OP_MOUNT: + /* Ignore legacy mounts */ + verify(zfs_prop_get(zhp, ZFS_PROP_MOUNTPOINT, + nfsiscsi_mnt_prop, + sizeof (nfsiscsi_mnt_prop), + NULL, NULL, 0, B_FALSE) == 0); + if (strcmp(nfsiscsi_mnt_prop, "legacy") == 0) + continue; + /* Ignore canmount=noauto mounts */ + if (zfs_prop_get_int(zhp, ZFS_PROP_CANMOUNT) == + ZFS_CANMOUNT_NOAUTO) + continue; + default: + break; + } + + node = safe_malloc(sizeof (unshare_unmount_node_t)); + node->un_zhp = zhp; + + if ((node->un_mountp = strdup(entry.mnt_mountp)) == + NULL) { + (void) fprintf(stderr, gettext("internal error:" + " out of memory\n")); + exit(1); + } + + uu_avl_node_init(node, &node->un_avlnode, pool); + + if (uu_avl_find(tree, node, NULL, &idx) == NULL) { + uu_avl_insert(tree, node, idx); + } else { + zfs_close(node->un_zhp); + free(node->un_mountp); + free(node); + } + } + + /* + * Walk the AVL tree in reverse, unmounting each filesystem and + * removing it from the AVL tree in the process. + */ + if ((walk = uu_avl_walk_start(tree, + UU_WALK_REVERSE | UU_WALK_ROBUST)) == NULL) { + (void) fprintf(stderr, + gettext("internal error: out of memory")); + exit(1); + } + + while ((node = uu_avl_walk_next(walk)) != NULL) { + uu_avl_remove(tree, node); + + switch (op) { + case OP_SHARE: + if (zfs_unshareall_bypath(node->un_zhp, + node->un_mountp) != 0) + ret = 1; + break; + + case OP_MOUNT: + if (zfs_unmount(node->un_zhp, + node->un_mountp, flags) != 0) + ret = 1; + break; + } + + zfs_close(node->un_zhp); + free(node->un_mountp); + free(node); + } + + uu_avl_walk_end(walk); + uu_avl_destroy(tree); + uu_avl_pool_destroy(pool); + + if (op == OP_SHARE) { + /* + * Finally, unshare any volumes shared via iSCSI. + */ + zfs_handle_t **dslist = NULL; + size_t i, count = 0; + + get_all_datasets(ZFS_TYPE_VOLUME, &dslist, &count, + B_FALSE); + + if (count != 0) { + qsort(dslist, count, sizeof (void *), + dataset_cmp); + + for (i = 0; i < count; i++) { + if (zfs_unshare_iscsi(dslist[i]) != 0) + ret = 1; + zfs_close(dslist[i]); + } + + free(dslist); + } + } + } else { + if (argc != 1) { + if (argc == 0) + (void) fprintf(stderr, + gettext("missing filesystem argument\n")); + else + (void) fprintf(stderr, + gettext("too many arguments\n")); + usage(B_FALSE); + } + + /* + * We have an argument, but it may be a full path or a ZFS + * filesystem. Pass full paths off to unmount_path() (shared by + * manual_unmount), otherwise open the filesystem and pass to + * zfs_unmount(). + */ + if (argv[0][0] == '/') + return (unshare_unmount_path(op, argv[0], + flags, B_FALSE)); + + types = ZFS_TYPE_FILESYSTEM; + if (op == OP_SHARE) + types |= ZFS_TYPE_VOLUME; + + if ((zhp = zfs_open(g_zfs, argv[0], types)) == NULL) + return (1); + + if (zfs_get_type(zhp) == ZFS_TYPE_FILESYSTEM) { + verify(zfs_prop_get(zhp, op == OP_SHARE ? + ZFS_PROP_SHARENFS : ZFS_PROP_MOUNTPOINT, + nfsiscsi_mnt_prop, sizeof (nfsiscsi_mnt_prop), NULL, + NULL, 0, B_FALSE) == 0); + + switch (op) { + case OP_SHARE: + verify(zfs_prop_get(zhp, ZFS_PROP_SHARENFS, + nfsiscsi_mnt_prop, + sizeof (nfsiscsi_mnt_prop), + NULL, NULL, 0, B_FALSE) == 0); + verify(zfs_prop_get(zhp, ZFS_PROP_SHARESMB, + sharesmb, sizeof (sharesmb), NULL, NULL, + 0, B_FALSE) == 0); + + if (strcmp(nfsiscsi_mnt_prop, "off") == 0 && + strcmp(sharesmb, "off") == 0) { + (void) fprintf(stderr, gettext("cannot " + "unshare '%s': legacy share\n"), + zfs_get_name(zhp)); + (void) fprintf(stderr, gettext("use " + "unshare(1M) to unshare this " + "filesystem\n")); + ret = 1; + } else if (!zfs_is_shared(zhp)) { + (void) fprintf(stderr, gettext("cannot " + "unshare '%s': not currently " + "shared\n"), zfs_get_name(zhp)); + ret = 1; + } else if (zfs_unshareall(zhp) != 0) { + ret = 1; + } + break; + + case OP_MOUNT: + if (strcmp(nfsiscsi_mnt_prop, "legacy") == 0) { + (void) fprintf(stderr, gettext("cannot " + "unmount '%s': legacy " + "mountpoint\n"), zfs_get_name(zhp)); + (void) fprintf(stderr, gettext("use " + "umount(1M) to unmount this " + "filesystem\n")); + ret = 1; + } else if (!zfs_is_mounted(zhp, NULL)) { + (void) fprintf(stderr, gettext("cannot " + "unmount '%s': not currently " + "mounted\n"), + zfs_get_name(zhp)); + ret = 1; + } else if (zfs_unmountall(zhp, flags) != 0) { + ret = 1; + } + break; + } + } else { + assert(op == OP_SHARE); + + verify(zfs_prop_get(zhp, ZFS_PROP_SHAREISCSI, + nfsiscsi_mnt_prop, sizeof (nfsiscsi_mnt_prop), + NULL, NULL, 0, B_FALSE) == 0); + + if (strcmp(nfsiscsi_mnt_prop, "off") == 0) { + (void) fprintf(stderr, gettext("cannot unshare " + "'%s': 'shareiscsi' property not set\n"), + zfs_get_name(zhp)); + (void) fprintf(stderr, gettext("set " + "'shareiscsi' property or use " + "iscsitadm(1M) to share this volume\n")); + ret = 1; + } else if (!zfs_is_shared_iscsi(zhp)) { + (void) fprintf(stderr, gettext("cannot " + "unshare '%s': not currently shared\n"), + zfs_get_name(zhp)); + ret = 1; + } else if (zfs_unshare_iscsi(zhp) != 0) { + ret = 1; + } + } + + zfs_close(zhp); + } + + return (ret); +} + +/* + * zfs unmount -a + * zfs unmount filesystem + * + * Unmount all filesystems, or a specific ZFS filesystem. + */ +static int +zfs_do_unmount(int argc, char **argv) +{ + return (unshare_unmount(OP_MOUNT, argc, argv)); +} + +/* + * zfs unshare -a + * zfs unshare filesystem + * + * Unshare all filesystems, or a specific ZFS filesystem. + */ +static int +zfs_do_unshare(int argc, char **argv) +{ + return (unshare_unmount(OP_SHARE, argc, argv)); +} + +/* + * Called when invoked as /etc/fs/zfs/mount. Do the mount if the mountpoint is + * 'legacy'. Otherwise, complain that use should be using 'zfs mount'. + */ +static int +manual_mount(int argc, char **argv) +{ + zfs_handle_t *zhp; + char mountpoint[ZFS_MAXPROPLEN]; + char mntopts[MNT_LINE_MAX] = { '\0' }; + int ret; + int c; + int flags = 0; + char *dataset, *path; + + /* check options */ + while ((c = getopt(argc, argv, ":mo:O")) != -1) { + switch (c) { + case 'o': + (void) strlcpy(mntopts, optarg, sizeof (mntopts)); + break; + case 'O': + flags |= MS_OVERLAY; + break; + case 'm': + flags |= MS_NOMNTTAB; + break; + case ':': + (void) fprintf(stderr, gettext("missing argument for " + "'%c' option\n"), optopt); + usage(B_FALSE); + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + (void) fprintf(stderr, gettext("usage: mount [-o opts] " + "<path>\n")); + return (2); + } + } + + argc -= optind; + argv += optind; + + /* check that we only have two arguments */ + if (argc != 2) { + if (argc == 0) + (void) fprintf(stderr, gettext("missing dataset " + "argument\n")); + else if (argc == 1) + (void) fprintf(stderr, + gettext("missing mountpoint argument\n")); + else + (void) fprintf(stderr, gettext("too many arguments\n")); + (void) fprintf(stderr, "usage: mount <dataset> <mountpoint>\n"); + return (2); + } + + dataset = argv[0]; + path = argv[1]; + + /* try to open the dataset */ + if ((zhp = zfs_open(g_zfs, dataset, ZFS_TYPE_FILESYSTEM)) == NULL) + return (1); + + (void) zfs_prop_get(zhp, ZFS_PROP_MOUNTPOINT, mountpoint, + sizeof (mountpoint), NULL, NULL, 0, B_FALSE); + + /* check for legacy mountpoint and complain appropriately */ + ret = 0; + if (strcmp(mountpoint, ZFS_MOUNTPOINT_LEGACY) == 0) { + if (mount(dataset, path, MS_OPTIONSTR | flags, MNTTYPE_ZFS, + NULL, 0, mntopts, sizeof (mntopts)) != 0) { + (void) fprintf(stderr, gettext("mount failed: %s\n"), + strerror(errno)); + ret = 1; + } + } else { + (void) fprintf(stderr, gettext("filesystem '%s' cannot be " + "mounted using 'mount -F zfs'\n"), dataset); + (void) fprintf(stderr, gettext("Use 'zfs set mountpoint=%s' " + "instead.\n"), path); + (void) fprintf(stderr, gettext("If you must use 'mount -F zfs' " + "or /etc/vfstab, use 'zfs set mountpoint=legacy'.\n")); + (void) fprintf(stderr, gettext("See zfs(1M) for more " + "information.\n")); + ret = 1; + } + + return (ret); +} + +/* + * Called when invoked as /etc/fs/zfs/umount. Unlike a manual mount, we allow + * unmounts of non-legacy filesystems, as this is the dominant administrative + * interface. + */ +static int +manual_unmount(int argc, char **argv) +{ + int flags = 0; + int c; + + /* check options */ + while ((c = getopt(argc, argv, "f")) != -1) { + switch (c) { + case 'f': + flags = MS_FORCE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + (void) fprintf(stderr, gettext("usage: unmount [-f] " + "<path>\n")); + return (2); + } + } + + argc -= optind; + argv += optind; + + /* check arguments */ + if (argc != 1) { + if (argc == 0) + (void) fprintf(stderr, gettext("missing path " + "argument\n")); + else + (void) fprintf(stderr, gettext("too many arguments\n")); + (void) fprintf(stderr, gettext("usage: unmount [-f] <path>\n")); + return (2); + } + + return (unshare_unmount_path(OP_MOUNT, argv[0], flags, B_TRUE)); +} + +static int +volcheck(zpool_handle_t *zhp, void *data) +{ + boolean_t isinit = *((boolean_t *)data); + + if (isinit) + return (zpool_create_zvol_links(zhp)); + else + return (zpool_remove_zvol_links(zhp)); +} + +/* + * Iterate over all pools in the system and either create or destroy /dev/zvol + * links, depending on the value of 'isinit'. + */ +static int +do_volcheck(boolean_t isinit) +{ + return (zpool_iter(g_zfs, volcheck, &isinit) ? 1 : 0); +} + +static int +find_command_idx(char *command, int *idx) +{ + int i; + + for (i = 0; i < NCOMMAND; i++) { + if (command_table[i].name == NULL) + continue; + + if (strcmp(command, command_table[i].name) == 0) { + *idx = i; + return (0); + } + } + return (1); +} + +int +main(int argc, char **argv) +{ + int ret; + int i; + char *progname; + char *cmdname; + + (void) setlocale(LC_ALL, ""); + (void) textdomain(TEXT_DOMAIN); + + opterr = 0; + + if ((g_zfs = libzfs_init()) == NULL) { + (void) fprintf(stderr, gettext("internal error: failed to " + "initialize ZFS library\n")); + return (1); + } + + zpool_set_history_str("zfs", argc, argv, history_str); + verify(zpool_stage_history(g_zfs, history_str) == 0); + + libzfs_print_on_error(g_zfs, B_TRUE); + + if ((mnttab_file = fopen(MNTTAB, "r")) == NULL) { + (void) fprintf(stderr, gettext("internal error: unable to " + "open %s\n"), MNTTAB); + return (1); + } + + /* + * This command also doubles as the /etc/fs mount and unmount program. + * Determine if we should take this behavior based on argv[0]. + */ + progname = basename(argv[0]); + if (strcmp(progname, "mount") == 0) { + ret = manual_mount(argc, argv); + } else if (strcmp(progname, "umount") == 0) { + ret = manual_unmount(argc, argv); + } else { + /* + * Make sure the user has specified some command. + */ + if (argc < 2) { + (void) fprintf(stderr, gettext("missing command\n")); + usage(B_FALSE); + } + + cmdname = argv[1]; + + /* + * The 'umount' command is an alias for 'unmount' + */ + if (strcmp(cmdname, "umount") == 0) + cmdname = "unmount"; + + /* + * The 'recv' command is an alias for 'receive' + */ + if (strcmp(cmdname, "recv") == 0) + cmdname = "receive"; + + /* + * Special case '-?' + */ + if (strcmp(cmdname, "-?") == 0) + usage(B_TRUE); + + /* + * 'volinit' and 'volfini' do not appear in the usage message, + * so we have to special case them here. + */ + if (strcmp(cmdname, "volinit") == 0) + return (do_volcheck(B_TRUE)); + else if (strcmp(cmdname, "volfini") == 0) + return (do_volcheck(B_FALSE)); + + /* + * Run the appropriate command. + */ + if (find_command_idx(cmdname, &i) == 0) { + current_command = &command_table[i]; + ret = command_table[i].func(argc - 1, argv + 1); + } else if (strchr(cmdname, '=') != NULL) { + verify(find_command_idx("set", &i) == 0); + current_command = &command_table[i]; + ret = command_table[i].func(argc, argv); + } else { + (void) fprintf(stderr, gettext("unrecognized " + "command '%s'\n"), cmdname); + usage(B_FALSE); + } + } + + (void) fclose(mnttab_file); + + libzfs_fini(g_zfs); + + /* + * The 'ZFS_ABORT' environment variable causes us to dump core on exit + * for the purposes of running ::findleaks. + */ + if (getenv("ZFS_ABORT") != NULL) { + (void) printf("dumping core by request\n"); + abort(); + } + + return (ret); +} diff --git a/cmd/zfs/zfs_util.h b/cmd/zfs/zfs_util.h new file mode 100644 index 000000000..c7f2f1618 --- /dev/null +++ b/cmd/zfs/zfs_util.h @@ -0,0 +1,44 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ +/* + * Copyright 2006 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#ifndef _ZFS_UTIL_H +#define _ZFS_UTIL_H + +#pragma ident "%Z%%M% %I% %E% SMI" + +#include <libzfs.h> + +#ifdef __cplusplus +extern "C" { +#endif + +void * safe_malloc(size_t size); +libzfs_handle_t *g_zfs; + +#ifdef __cplusplus +} +#endif + +#endif /* _ZFS_UTIL_H */ diff --git a/cmd/zinject/translate.c b/cmd/zinject/translate.c new file mode 100644 index 000000000..c85e024b6 --- /dev/null +++ b/cmd/zinject/translate.c @@ -0,0 +1,474 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ +/* + * Copyright 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#include <libzfs.h> + +#undef verify /* both libzfs.h and zfs_context.h want to define this */ + +#include <sys/zfs_context.h> + +#include <errno.h> +#include <fcntl.h> +#include <stdarg.h> +#include <stddef.h> +#include <stdio.h> +#include <stdlib.h> +#include <strings.h> +#include <sys/file.h> +#include <sys/mntent.h> +#include <sys/mnttab.h> +#include <sys/param.h> +#include <sys/stat.h> + +#include <sys/dmu.h> +#include <sys/dmu_objset.h> +#include <sys/dnode.h> +#include <sys/vdev_impl.h> + +#include <sys/mkdev.h> + +#include "zinject.h" + +extern void kernel_init(int); +extern void kernel_fini(void); + +static int debug; + +static void +ziprintf(const char *fmt, ...) +{ + va_list ap; + + if (!debug) + return; + + va_start(ap, fmt); + (void) vprintf(fmt, ap); + va_end(ap); +} + +/* + * Given a full path to a file, translate into a dataset name and a relative + * path within the dataset. 'dataset' must be at least MAXNAMELEN characters, + * and 'relpath' must be at least MAXPATHLEN characters. We also pass a stat64 + * buffer, which we need later to get the object ID. + */ +static int +parse_pathname(const char *fullpath, char *dataset, char *relpath, + struct stat64 *statbuf) +{ + struct extmnttab mp; + FILE *fp; + int match; + const char *rel; + + if (fullpath[0] != '/') { + (void) fprintf(stderr, "invalid object '%s': must be full " + "path\n", fullpath); + usage(); + return (-1); + } + + if (strlen(fullpath) >= MAXPATHLEN) { + (void) fprintf(stderr, "invalid object; pathname too long\n"); + return (-1); + } + + if (stat64(fullpath, statbuf) != 0) { + (void) fprintf(stderr, "cannot open '%s': %s\n", + fullpath, strerror(errno)); + return (-1); + } + + if ((fp = fopen(MNTTAB, "r")) == NULL) { + (void) fprintf(stderr, "cannot open /etc/mnttab\n"); + return (-1); + } + + match = 0; + while (getextmntent(fp, &mp, sizeof (mp)) == 0) { + if (makedev(mp.mnt_major, mp.mnt_minor) == statbuf->st_dev) { + match = 1; + break; + } + } + + if (!match) { + (void) fprintf(stderr, "cannot find mountpoint for '%s'\n", + fullpath); + return (-1); + } + + if (strcmp(mp.mnt_fstype, MNTTYPE_ZFS) != 0) { + (void) fprintf(stderr, "invalid path '%s': not a ZFS " + "filesystem\n", fullpath); + return (-1); + } + + if (strncmp(fullpath, mp.mnt_mountp, strlen(mp.mnt_mountp)) != 0) { + (void) fprintf(stderr, "invalid path '%s': mountpoint " + "doesn't match path\n", fullpath); + return (-1); + } + + (void) strcpy(dataset, mp.mnt_special); + + rel = fullpath + strlen(mp.mnt_mountp); + if (rel[0] == '/') + rel++; + (void) strcpy(relpath, rel); + + return (0); +} + +/* + * Convert from a (dataset, path) pair into a (objset, object) pair. Note that + * we grab the object number from the inode number, since looking this up via + * libzpool is a real pain. + */ +/* ARGSUSED */ +static int +object_from_path(const char *dataset, const char *path, struct stat64 *statbuf, + zinject_record_t *record) +{ + objset_t *os; + int err; + + /* + * Before doing any libzpool operations, call sync() to ensure that the + * on-disk state is consistent with the in-core state. + */ + sync(); + + if ((err = dmu_objset_open(dataset, DMU_OST_ZFS, + DS_MODE_USER | DS_MODE_READONLY, &os)) != 0) { + (void) fprintf(stderr, "cannot open dataset '%s': %s\n", + dataset, strerror(err)); + return (-1); + } + + record->zi_objset = dmu_objset_id(os); + record->zi_object = statbuf->st_ino; + + dmu_objset_close(os); + + return (0); +} + +/* + * Calculate the real range based on the type, level, and range given. + */ +static int +calculate_range(const char *dataset, err_type_t type, int level, char *range, + zinject_record_t *record) +{ + objset_t *os = NULL; + dnode_t *dn = NULL; + int err; + int ret = -1; + + /* + * Determine the numeric range from the string. + */ + if (range == NULL) { + /* + * If range is unspecified, set the range to [0,-1], which + * indicates that the whole object should be treated as an + * error. + */ + record->zi_start = 0; + record->zi_end = -1ULL; + } else { + char *end; + + /* XXX add support for suffixes */ + record->zi_start = strtoull(range, &end, 10); + + + if (*end == '\0') + record->zi_end = record->zi_start + 1; + else if (*end == ',') + record->zi_end = strtoull(end + 1, &end, 10); + + if (*end != '\0') { + (void) fprintf(stderr, "invalid range '%s': must be " + "a numeric range of the form 'start[,end]'\n", + range); + goto out; + } + } + + switch (type) { + case TYPE_DATA: + break; + + case TYPE_DNODE: + /* + * If this is a request to inject faults into the dnode, then we + * must translate the current (objset,object) pair into an + * offset within the metadnode for the objset. Specifying any + * kind of range with type 'dnode' is illegal. + */ + if (range != NULL) { + (void) fprintf(stderr, "range cannot be specified when " + "type is 'dnode'\n"); + goto out; + } + + record->zi_start = record->zi_object * sizeof (dnode_phys_t); + record->zi_end = record->zi_start + sizeof (dnode_phys_t); + record->zi_object = 0; + break; + } + + /* + * Get the dnode associated with object, so we can calculate the block + * size. + */ + if ((err = dmu_objset_open(dataset, DMU_OST_ANY, + DS_MODE_USER | DS_MODE_READONLY, &os)) != 0) { + (void) fprintf(stderr, "cannot open dataset '%s': %s\n", + dataset, strerror(err)); + goto out; + } + + if (record->zi_object == 0) { + dn = os->os->os_meta_dnode; + } else { + err = dnode_hold(os->os, record->zi_object, FTAG, &dn); + if (err != 0) { + (void) fprintf(stderr, "failed to hold dnode " + "for object %llu\n", + (u_longlong_t)record->zi_object); + goto out; + } + } + + + ziprintf("data shift: %d\n", (int)dn->dn_datablkshift); + ziprintf(" ind shift: %d\n", (int)dn->dn_indblkshift); + + /* + * Translate range into block IDs. + */ + if (record->zi_start != 0 || record->zi_end != -1ULL) { + record->zi_start >>= dn->dn_datablkshift; + record->zi_end >>= dn->dn_datablkshift; + } + + /* + * Check level, and then translate level 0 blkids into ranges + * appropriate for level of indirection. + */ + record->zi_level = level; + if (level > 0) { + ziprintf("level 0 blkid range: [%llu, %llu]\n", + record->zi_start, record->zi_end); + + if (level >= dn->dn_nlevels) { + (void) fprintf(stderr, "level %d exceeds max level " + "of object (%d)\n", level, dn->dn_nlevels - 1); + goto out; + } + + if (record->zi_start != 0 || record->zi_end != 0) { + int shift = dn->dn_indblkshift - SPA_BLKPTRSHIFT; + + for (; level > 0; level--) { + record->zi_start >>= shift; + record->zi_end >>= shift; + } + } + } + + ret = 0; +out: + if (dn) { + if (dn != os->os->os_meta_dnode) + dnode_rele(dn, FTAG); + } + if (os) + dmu_objset_close(os); + + return (ret); +} + +int +translate_record(err_type_t type, const char *object, const char *range, + int level, zinject_record_t *record, char *poolname, char *dataset) +{ + char path[MAXPATHLEN]; + char *slash; + struct stat64 statbuf; + int ret = -1; + + kernel_init(FREAD); + + debug = (getenv("ZINJECT_DEBUG") != NULL); + + ziprintf("translating: %s\n", object); + + if (MOS_TYPE(type)) { + /* + * MOS objects are treated specially. + */ + switch (type) { + case TYPE_MOS: + record->zi_type = 0; + break; + case TYPE_MOSDIR: + record->zi_type = DMU_OT_OBJECT_DIRECTORY; + break; + case TYPE_METASLAB: + record->zi_type = DMU_OT_OBJECT_ARRAY; + break; + case TYPE_CONFIG: + record->zi_type = DMU_OT_PACKED_NVLIST; + break; + case TYPE_BPLIST: + record->zi_type = DMU_OT_BPLIST; + break; + case TYPE_SPACEMAP: + record->zi_type = DMU_OT_SPACE_MAP; + break; + case TYPE_ERRLOG: + record->zi_type = DMU_OT_ERROR_LOG; + break; + } + + dataset[0] = '\0'; + (void) strcpy(poolname, object); + return (0); + } + + /* + * Convert a full path into a (dataset, file) pair. + */ + if (parse_pathname(object, dataset, path, &statbuf) != 0) + goto err; + + ziprintf(" dataset: %s\n", dataset); + ziprintf(" path: %s\n", path); + + /* + * Convert (dataset, file) into (objset, object) + */ + if (object_from_path(dataset, path, &statbuf, record) != 0) + goto err; + + ziprintf("raw objset: %llu\n", record->zi_objset); + ziprintf("raw object: %llu\n", record->zi_object); + + /* + * For the given object, calculate the real (type, level, range) + */ + if (calculate_range(dataset, type, level, (char *)range, record) != 0) + goto err; + + ziprintf(" objset: %llu\n", record->zi_objset); + ziprintf(" object: %llu\n", record->zi_object); + if (record->zi_start == 0 && + record->zi_end == -1ULL) + ziprintf(" range: all\n"); + else + ziprintf(" range: [%llu, %llu]\n", record->zi_start, + record->zi_end); + + /* + * Copy the pool name + */ + (void) strcpy(poolname, dataset); + if ((slash = strchr(poolname, '/')) != NULL) + *slash = '\0'; + + ret = 0; + +err: + kernel_fini(); + return (ret); +} + +int +translate_raw(const char *str, zinject_record_t *record) +{ + /* + * A raw bookmark of the form objset:object:level:blkid, where each + * number is a hexidecimal value. + */ + if (sscanf(str, "%llx:%llx:%x:%llx", (u_longlong_t *)&record->zi_objset, + (u_longlong_t *)&record->zi_object, &record->zi_level, + (u_longlong_t *)&record->zi_start) != 4) { + (void) fprintf(stderr, "bad raw spec '%s': must be of the form " + "'objset:object:level:blkid'\n", str); + return (-1); + } + + record->zi_end = record->zi_start; + + return (0); +} + +int +translate_device(const char *pool, const char *device, err_type_t label_type, + zinject_record_t *record) +{ + char *end; + zpool_handle_t *zhp; + nvlist_t *tgt; + boolean_t isspare, iscache; + + /* + * Given a device name or GUID, create an appropriate injection record + * with zi_guid set. + */ + if ((zhp = zpool_open(g_zfs, pool)) == NULL) + return (-1); + + record->zi_guid = strtoull(device, &end, 16); + if (record->zi_guid == 0 || *end != '\0') { + tgt = zpool_find_vdev(zhp, device, &isspare, &iscache, NULL); + + if (tgt == NULL) { + (void) fprintf(stderr, "cannot find device '%s' in " + "pool '%s'\n", device, pool); + return (-1); + } + + verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID, + &record->zi_guid) == 0); + } + + switch (label_type) { + case TYPE_LABEL_UBERBLOCK: + record->zi_start = offsetof(vdev_label_t, vl_uberblock[0]); + record->zi_end = record->zi_start + VDEV_UBERBLOCK_RING - 1; + break; + case TYPE_LABEL_NVLIST: + record->zi_start = offsetof(vdev_label_t, vl_vdev_phys); + record->zi_end = record->zi_start + VDEV_PHYS_SIZE - 1; + break; + } + return (0); +} diff --git a/cmd/zinject/zinject.c b/cmd/zinject/zinject.c new file mode 100644 index 000000000..02dc18b9c --- /dev/null +++ b/cmd/zinject/zinject.c @@ -0,0 +1,768 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ +/* + * Copyright 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#pragma ident "%Z%%M% %I% %E% SMI" + +/* + * ZFS Fault Injector + * + * This userland component takes a set of options and uses libzpool to translate + * from a user-visible object type and name to an internal representation. + * There are two basic types of faults: device faults and data faults. + * + * + * DEVICE FAULTS + * + * Errors can be injected into a particular vdev using the '-d' option. This + * option takes a path or vdev GUID to uniquely identify the device within a + * pool. There are two types of errors that can be injected, EIO and ENXIO, + * that can be controlled through the '-e' option. The default is ENXIO. For + * EIO failures, any attempt to read data from the device will return EIO, but + * subsequent attempt to reopen the device will succeed. For ENXIO failures, + * any attempt to read from the device will return EIO, but any attempt to + * reopen the device will also return ENXIO. + * For label faults, the -L option must be specified. This allows faults + * to be injected into either the nvlist or uberblock region of all the labels + * for the specified device. + * + * This form of the command looks like: + * + * zinject -d device [-e errno] [-L <uber | nvlist>] pool + * + * + * DATA FAULTS + * + * We begin with a tuple of the form: + * + * <type,level,range,object> + * + * type A string describing the type of data to target. Each type + * implicitly describes how to interpret 'object'. Currently, + * the following values are supported: + * + * data User data for a file + * dnode Dnode for a file or directory + * + * The following MOS objects are special. Instead of injecting + * errors on a particular object or blkid, we inject errors across + * all objects of the given type. + * + * mos Any data in the MOS + * mosdir object directory + * config pool configuration + * bplist blkptr list + * spacemap spacemap + * metaslab metaslab + * errlog persistent error log + * + * level Object level. Defaults to '0', not applicable to all types. If + * a range is given, this corresponds to the indirect block + * corresponding to the specific range. + * + * range A numerical range [start,end) within the object. Defaults to + * the full size of the file. + * + * object A string describing the logical location of the object. For + * files and directories (currently the only supported types), + * this is the path of the object on disk. + * + * This is translated, via libzpool, into the following internal representation: + * + * <type,objset,object,level,range> + * + * These types should be self-explanatory. This tuple is then passed to the + * kernel via a special ioctl() to initiate fault injection for the given + * object. Note that 'type' is not strictly necessary for fault injection, but + * is used when translating existing faults into a human-readable string. + * + * + * The command itself takes one of the forms: + * + * zinject + * zinject <-a | -u pool> + * zinject -c <id|all> + * zinject [-q] <-t type> [-f freq] [-u] [-a] [-m] [-e errno] [-l level] + * [-r range] <object> + * zinject [-f freq] [-a] [-m] [-u] -b objset:object:level:start:end pool + * + * With no arguments, the command prints all currently registered injection + * handlers, with their numeric identifiers. + * + * The '-c' option will clear the given handler, or all handlers if 'all' is + * specified. + * + * The '-e' option takes a string describing the errno to simulate. This must + * be either 'io' or 'checksum'. In most cases this will result in the same + * behavior, but RAID-Z will produce a different set of ereports for this + * situation. + * + * The '-a', '-u', and '-m' flags toggle internal flush behavior. If '-a' is + * specified, then the ARC cache is flushed appropriately. If '-u' is + * specified, then the underlying SPA is unloaded. Either of these flags can be + * specified independently of any other handlers. The '-m' flag automatically + * does an unmount and remount of the underlying dataset to aid in flushing the + * cache. + * + * The '-f' flag controls the frequency of errors injected, expressed as a + * integer percentage between 1 and 100. The default is 100. + * + * The this form is responsible for actually injecting the handler into the + * framework. It takes the arguments described above, translates them to the + * internal tuple using libzpool, and then issues an ioctl() to register the + * handler. + * + * The final form can target a specific bookmark, regardless of whether a + * human-readable interface has been designed. It allows developers to specify + * a particular block by number. + */ + +#include <errno.h> +#include <fcntl.h> +#include <stdio.h> +#include <stdlib.h> +#include <strings.h> +#include <unistd.h> + +#include <sys/fs/zfs.h> +#include <sys/mount.h> + +#include <libzfs.h> + +#undef verify /* both libzfs.h and zfs_context.h want to define this */ + +#include "zinject.h" + +libzfs_handle_t *g_zfs; +int zfs_fd; + +#define ECKSUM EBADE + +static const char *errtable[TYPE_INVAL] = { + "data", + "dnode", + "mos", + "mosdir", + "metaslab", + "config", + "bplist", + "spacemap", + "errlog", + "uber", + "nvlist" +}; + +static err_type_t +name_to_type(const char *arg) +{ + int i; + for (i = 0; i < TYPE_INVAL; i++) + if (strcmp(errtable[i], arg) == 0) + return (i); + + return (TYPE_INVAL); +} + +static const char * +type_to_name(uint64_t type) +{ + switch (type) { + case DMU_OT_OBJECT_DIRECTORY: + return ("mosdir"); + case DMU_OT_OBJECT_ARRAY: + return ("metaslab"); + case DMU_OT_PACKED_NVLIST: + return ("config"); + case DMU_OT_BPLIST: + return ("bplist"); + case DMU_OT_SPACE_MAP: + return ("spacemap"); + case DMU_OT_ERROR_LOG: + return ("errlog"); + default: + return ("-"); + } +} + + +/* + * Print usage message. + */ +void +usage(void) +{ + (void) printf( + "usage:\n" + "\n" + "\tzinject\n" + "\n" + "\t\tList all active injection records.\n" + "\n" + "\tzinject -c <id|all>\n" + "\n" + "\t\tClear the particular record (if given a numeric ID), or\n" + "\t\tall records if 'all' is specificed.\n" + "\n" + "\tzinject -d device [-e errno] [-L <nvlist|uber>] pool\n" + "\t\tInject a fault into a particular device or the device's\n" + "\t\tlabel. Label injection can either be 'nvlist' or 'uber'.\n" + "\t\t'errno' can either be 'nxio' (the default) or 'io'.\n" + "\n" + "\tzinject -b objset:object:level:blkid pool\n" + "\n" + "\t\tInject an error into pool 'pool' with the numeric bookmark\n" + "\t\tspecified by the remaining tuple. Each number is in\n" + "\t\thexidecimal, and only one block can be specified.\n" + "\n" + "\tzinject [-q] <-t type> [-e errno] [-l level] [-r range]\n" + "\t [-a] [-m] [-u] [-f freq] <object>\n" + "\n" + "\t\tInject an error into the object specified by the '-t' option\n" + "\t\tand the object descriptor. The 'object' parameter is\n" + "\t\tinterperted depending on the '-t' option.\n" + "\n" + "\t\t-q\tQuiet mode. Only print out the handler number added.\n" + "\t\t-e\tInject a specific error. Must be either 'io' or\n" + "\t\t\t'checksum'. Default is 'io'.\n" + "\t\t-l\tInject error at a particular block level. Default is " + "0.\n" + "\t\t-m\tAutomatically remount underlying filesystem.\n" + "\t\t-r\tInject error over a particular logical range of an\n" + "\t\t\tobject. Will be translated to the appropriate blkid\n" + "\t\t\trange according to the object's properties.\n" + "\t\t-a\tFlush the ARC cache. Can be specified without any\n" + "\t\t\tassociated object.\n" + "\t\t-u\tUnload the associated pool. Can be specified with only\n" + "\t\t\ta pool object.\n" + "\t\t-f\tOnly inject errors a fraction of the time. Expressed as\n" + "\t\t\ta percentage between 1 and 100.\n" + "\n" + "\t-t data\t\tInject an error into the plain file contents of a\n" + "\t\t\tfile. The object must be specified as a complete path\n" + "\t\t\tto a file on a ZFS filesystem.\n" + "\n" + "\t-t dnode\tInject an error into the metadnode in the block\n" + "\t\t\tcorresponding to the dnode for a file or directory. The\n" + "\t\t\t'-r' option is incompatible with this mode. The object\n" + "\t\t\tis specified as a complete path to a file or directory\n" + "\t\t\ton a ZFS filesystem.\n" + "\n" + "\t-t <mos>\tInject errors into the MOS for objects of the given\n" + "\t\t\ttype. Valid types are: mos, mosdir, config, bplist,\n" + "\t\t\tspacemap, metaslab, errlog. The only valid <object> is\n" + "\t\t\tthe poolname.\n"); +} + +static int +iter_handlers(int (*func)(int, const char *, zinject_record_t *, void *), + void *data) +{ + zfs_cmd_t zc; + int ret; + + zc.zc_guid = 0; + + while (ioctl(zfs_fd, ZFS_IOC_INJECT_LIST_NEXT, &zc) == 0) + if ((ret = func((int)zc.zc_guid, zc.zc_name, + &zc.zc_inject_record, data)) != 0) + return (ret); + + return (0); +} + +static int +print_data_handler(int id, const char *pool, zinject_record_t *record, + void *data) +{ + int *count = data; + + if (record->zi_guid != 0) + return (0); + + if (*count == 0) { + (void) printf("%3s %-15s %-6s %-6s %-8s %3s %-15s\n", + "ID", "POOL", "OBJSET", "OBJECT", "TYPE", "LVL", "RANGE"); + (void) printf("--- --------------- ------ " + "------ -------- --- ---------------\n"); + } + + *count += 1; + + (void) printf("%3d %-15s %-6llu %-6llu %-8s %3d ", id, pool, + (u_longlong_t)record->zi_objset, (u_longlong_t)record->zi_object, + type_to_name(record->zi_type), record->zi_level); + + if (record->zi_start == 0 && + record->zi_end == -1ULL) + (void) printf("all\n"); + else + (void) printf("[%llu, %llu]\n", (u_longlong_t)record->zi_start, + (u_longlong_t)record->zi_end); + + return (0); +} + +static int +print_device_handler(int id, const char *pool, zinject_record_t *record, + void *data) +{ + int *count = data; + + if (record->zi_guid == 0) + return (0); + + if (*count == 0) { + (void) printf("%3s %-15s %s\n", "ID", "POOL", "GUID"); + (void) printf("--- --------------- ----------------\n"); + } + + *count += 1; + + (void) printf("%3d %-15s %llx\n", id, pool, + (u_longlong_t)record->zi_guid); + + return (0); +} + +/* + * Print all registered error handlers. Returns the number of handlers + * registered. + */ +static int +print_all_handlers(void) +{ + int count = 0; + + (void) iter_handlers(print_device_handler, &count); + (void) printf("\n"); + count = 0; + (void) iter_handlers(print_data_handler, &count); + + return (count); +} + +/* ARGSUSED */ +static int +cancel_one_handler(int id, const char *pool, zinject_record_t *record, + void *data) +{ + zfs_cmd_t zc; + + zc.zc_guid = (uint64_t)id; + + if (ioctl(zfs_fd, ZFS_IOC_CLEAR_FAULT, &zc) != 0) { + (void) fprintf(stderr, "failed to remove handler %d: %s\n", + id, strerror(errno)); + return (1); + } + + return (0); +} + +/* + * Remove all fault injection handlers. + */ +static int +cancel_all_handlers(void) +{ + int ret = iter_handlers(cancel_one_handler, NULL); + + (void) printf("removed all registered handlers\n"); + + return (ret); +} + +/* + * Remove a specific fault injection handler. + */ +static int +cancel_handler(int id) +{ + zfs_cmd_t zc; + + zc.zc_guid = (uint64_t)id; + + if (ioctl(zfs_fd, ZFS_IOC_CLEAR_FAULT, &zc) != 0) { + (void) fprintf(stderr, "failed to remove handler %d: %s\n", + id, strerror(errno)); + return (1); + } + + (void) printf("removed handler %d\n", id); + + return (0); +} + +/* + * Register a new fault injection handler. + */ +static int +register_handler(const char *pool, int flags, zinject_record_t *record, + int quiet) +{ + zfs_cmd_t zc; + + (void) strcpy(zc.zc_name, pool); + zc.zc_inject_record = *record; + zc.zc_guid = flags; + + if (ioctl(zfs_fd, ZFS_IOC_INJECT_FAULT, &zc) != 0) { + (void) fprintf(stderr, "failed to add handler: %s\n", + strerror(errno)); + return (1); + } + + if (flags & ZINJECT_NULL) + return (0); + + if (quiet) { + (void) printf("%llu\n", (u_longlong_t)zc.zc_guid); + } else { + (void) printf("Added handler %llu with the following " + "properties:\n", (u_longlong_t)zc.zc_guid); + (void) printf(" pool: %s\n", pool); + if (record->zi_guid) { + (void) printf(" vdev: %llx\n", + (u_longlong_t)record->zi_guid); + } else { + (void) printf("objset: %llu\n", + (u_longlong_t)record->zi_objset); + (void) printf("object: %llu\n", + (u_longlong_t)record->zi_object); + (void) printf(" type: %llu\n", + (u_longlong_t)record->zi_type); + (void) printf(" level: %d\n", record->zi_level); + if (record->zi_start == 0 && + record->zi_end == -1ULL) + (void) printf(" range: all\n"); + else + (void) printf(" range: [%llu, %llu)\n", + (u_longlong_t)record->zi_start, + (u_longlong_t)record->zi_end); + } + } + + return (0); +} + +int +main(int argc, char **argv) +{ + int c; + char *range = NULL; + char *cancel = NULL; + char *end; + char *raw = NULL; + char *device = NULL; + int level = 0; + int quiet = 0; + int error = 0; + int domount = 0; + err_type_t type = TYPE_INVAL; + err_type_t label = TYPE_INVAL; + zinject_record_t record = { 0 }; + char pool[MAXNAMELEN]; + char dataset[MAXNAMELEN]; + zfs_handle_t *zhp; + int ret; + int flags = 0; + + if ((g_zfs = libzfs_init()) == NULL) { + (void) fprintf(stderr, "internal error: failed to " + "initialize ZFS library\n"); + return (1); + } + + libzfs_print_on_error(g_zfs, B_TRUE); + + if ((zfs_fd = open(ZFS_DEV, O_RDWR)) < 0) { + (void) fprintf(stderr, "failed to open ZFS device\n"); + return (1); + } + + if (argc == 1) { + /* + * No arguments. Print the available handlers. If there are no + * available handlers, direct the user to '-h' for help + * information. + */ + if (print_all_handlers() == 0) { + (void) printf("No handlers registered.\n"); + (void) printf("Run 'zinject -h' for usage " + "information.\n"); + } + + return (0); + } + + while ((c = getopt(argc, argv, ":ab:d:f:qhc:t:l:mr:e:uL:")) != -1) { + switch (c) { + case 'a': + flags |= ZINJECT_FLUSH_ARC; + break; + case 'b': + raw = optarg; + break; + case 'c': + cancel = optarg; + break; + case 'd': + device = optarg; + break; + case 'e': + if (strcasecmp(optarg, "io") == 0) { + error = EIO; + } else if (strcasecmp(optarg, "checksum") == 0) { + error = ECKSUM; + } else if (strcasecmp(optarg, "nxio") == 0) { + error = ENXIO; + } else { + (void) fprintf(stderr, "invalid error type " + "'%s': must be 'io', 'checksum' or " + "'nxio'\n", optarg); + usage(); + return (1); + } + break; + case 'f': + record.zi_freq = atoi(optarg); + if (record.zi_freq < 1 || record.zi_freq > 100) { + (void) fprintf(stderr, "frequency range must " + "be in the range (0, 100]\n"); + return (1); + } + break; + case 'h': + usage(); + return (0); + case 'l': + level = (int)strtol(optarg, &end, 10); + if (*end != '\0') { + (void) fprintf(stderr, "invalid level '%s': " + "must be an integer\n", optarg); + usage(); + return (1); + } + break; + case 'm': + domount = 1; + break; + case 'q': + quiet = 1; + break; + case 'r': + range = optarg; + break; + case 't': + if ((type = name_to_type(optarg)) == TYPE_INVAL && + !MOS_TYPE(type)) { + (void) fprintf(stderr, "invalid type '%s'\n", + optarg); + usage(); + return (1); + } + break; + case 'u': + flags |= ZINJECT_UNLOAD_SPA; + break; + case 'L': + if ((label = name_to_type(optarg)) == TYPE_INVAL && + !LABEL_TYPE(type)) { + (void) fprintf(stderr, "invalid label type " + "'%s'\n", optarg); + usage(); + return (1); + } + break; + case ':': + (void) fprintf(stderr, "option -%c requires an " + "operand\n", optopt); + usage(); + return (1); + case '?': + (void) fprintf(stderr, "invalid option '%c'\n", + optopt); + usage(); + return (2); + } + } + + argc -= optind; + argv += optind; + + if (cancel != NULL) { + /* + * '-c' is invalid with any other options. + */ + if (raw != NULL || range != NULL || type != TYPE_INVAL || + level != 0) { + (void) fprintf(stderr, "cancel (-c) incompatible with " + "any other options\n"); + usage(); + return (2); + } + if (argc != 0) { + (void) fprintf(stderr, "extraneous argument to '-c'\n"); + usage(); + return (2); + } + + if (strcmp(cancel, "all") == 0) { + return (cancel_all_handlers()); + } else { + int id = (int)strtol(cancel, &end, 10); + if (*end != '\0') { + (void) fprintf(stderr, "invalid handle id '%s':" + " must be an integer or 'all'\n", cancel); + usage(); + return (1); + } + return (cancel_handler(id)); + } + } + + if (device != NULL) { + /* + * Device (-d) injection uses a completely different mechanism + * for doing injection, so handle it separately here. + */ + if (raw != NULL || range != NULL || type != TYPE_INVAL || + level != 0) { + (void) fprintf(stderr, "device (-d) incompatible with " + "data error injection\n"); + usage(); + return (2); + } + + if (argc != 1) { + (void) fprintf(stderr, "device (-d) injection requires " + "a single pool name\n"); + usage(); + return (2); + } + + (void) strcpy(pool, argv[0]); + dataset[0] = '\0'; + + if (error == ECKSUM) { + (void) fprintf(stderr, "device error type must be " + "'io' or 'nxio'\n"); + return (1); + } + + if (translate_device(pool, device, label, &record) != 0) + return (1); + if (!error) + error = ENXIO; + } else if (raw != NULL) { + if (range != NULL || type != TYPE_INVAL || level != 0) { + (void) fprintf(stderr, "raw (-b) format with " + "any other options\n"); + usage(); + return (2); + } + + if (argc != 1) { + (void) fprintf(stderr, "raw (-b) format expects a " + "single pool name\n"); + usage(); + return (2); + } + + (void) strcpy(pool, argv[0]); + dataset[0] = '\0'; + + if (error == ENXIO) { + (void) fprintf(stderr, "data error type must be " + "'checksum' or 'io'\n"); + return (1); + } + + if (translate_raw(raw, &record) != 0) + return (1); + if (!error) + error = EIO; + } else if (type == TYPE_INVAL) { + if (flags == 0) { + (void) fprintf(stderr, "at least one of '-b', '-d', " + "'-t', '-a', or '-u' must be specified\n"); + usage(); + return (2); + } + + if (argc == 1 && (flags & ZINJECT_UNLOAD_SPA)) { + (void) strcpy(pool, argv[0]); + dataset[0] = '\0'; + } else if (argc != 0) { + (void) fprintf(stderr, "extraneous argument for " + "'-f'\n"); + usage(); + return (2); + } + + flags |= ZINJECT_NULL; + } else { + if (argc != 1) { + (void) fprintf(stderr, "missing object\n"); + usage(); + return (2); + } + + if (error == ENXIO) { + (void) fprintf(stderr, "data error type must be " + "'checksum' or 'io'\n"); + return (1); + } + + if (translate_record(type, argv[0], range, level, &record, pool, + dataset) != 0) + return (1); + if (!error) + error = EIO; + } + + /* + * If this is pool-wide metadata, unmount everything. The ioctl() will + * unload the pool, so that we trigger spa-wide reopen of metadata next + * time we access the pool. + */ + if (dataset[0] != '\0' && domount) { + if ((zhp = zfs_open(g_zfs, dataset, ZFS_TYPE_DATASET)) == NULL) + return (1); + + if (zfs_unmount(zhp, NULL, 0) != 0) + return (1); + } + + record.zi_error = error; + + ret = register_handler(pool, flags, &record, quiet); + + if (dataset[0] != '\0' && domount) + ret = (zfs_mount(zhp, NULL, 0) != 0); + + libzfs_fini(g_zfs); + + return (ret); +} diff --git a/cmd/zinject/zinject.h b/cmd/zinject/zinject.h new file mode 100644 index 000000000..adc3efe80 --- /dev/null +++ b/cmd/zinject/zinject.h @@ -0,0 +1,71 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ +/* + * Copyright 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#ifndef _ZINJECT_H +#define _ZINJECT_H + +#pragma ident "%Z%%M% %I% %E% SMI" + +#include <sys/zfs_ioctl.h> + +#ifdef __cplusplus +extern "C" { +#endif + +typedef enum { + TYPE_DATA, /* plain file contents */ + TYPE_DNODE, /* metadnode contents */ + TYPE_MOS, /* all MOS data */ + TYPE_MOSDIR, /* MOS object directory */ + TYPE_METASLAB, /* metaslab objects */ + TYPE_CONFIG, /* MOS config */ + TYPE_BPLIST, /* block pointer list */ + TYPE_SPACEMAP, /* space map objects */ + TYPE_ERRLOG, /* persistent error log */ + TYPE_LABEL_UBERBLOCK, /* label specific uberblock */ + TYPE_LABEL_NVLIST, /* label specific nvlist */ + TYPE_INVAL +} err_type_t; + +#define MOS_TYPE(t) \ + ((t) >= TYPE_MOS && (t) < TYPE_LABEL_UBERBLOCK) + +#define LABEL_TYPE(t) \ + ((t) >= TYPE_LABEL_UBERBLOCK && (t) < TYPE_INVAL) + +int translate_record(err_type_t type, const char *object, const char *range, + int level, zinject_record_t *record, char *poolname, char *dataset); +int translate_raw(const char *raw, zinject_record_t *record); +int translate_device(const char *pool, const char *device, + err_type_t label_type, zinject_record_t *record); +void usage(void); + +extern libzfs_handle_t *g_zfs; + +#ifdef __cplusplus +} +#endif + +#endif /* _ZINJECT_H */ diff --git a/cmd/zpool/zpool_iter.c b/cmd/zpool/zpool_iter.c new file mode 100644 index 000000000..2f0daefd5 --- /dev/null +++ b/cmd/zpool/zpool_iter.c @@ -0,0 +1,252 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ +/* + * Copyright 2007 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#pragma ident "%Z%%M% %I% %E% SMI" + +#include <libintl.h> +#include <libuutil.h> +#include <stddef.h> +#include <stdio.h> +#include <stdlib.h> +#include <strings.h> + +#include <libzfs.h> + +#include "zpool_util.h" + +/* + * Private interface for iterating over pools specified on the command line. + * Most consumers will call for_each_pool, but in order to support iostat, we + * allow fined grained control through the zpool_list_t interface. + */ + +typedef struct zpool_node { + zpool_handle_t *zn_handle; + uu_avl_node_t zn_avlnode; + int zn_mark; +} zpool_node_t; + +struct zpool_list { + boolean_t zl_findall; + uu_avl_t *zl_avl; + uu_avl_pool_t *zl_pool; + zprop_list_t **zl_proplist; +}; + +/* ARGSUSED */ +static int +zpool_compare(const void *larg, const void *rarg, void *unused) +{ + zpool_handle_t *l = ((zpool_node_t *)larg)->zn_handle; + zpool_handle_t *r = ((zpool_node_t *)rarg)->zn_handle; + const char *lname = zpool_get_name(l); + const char *rname = zpool_get_name(r); + + return (strcmp(lname, rname)); +} + +/* + * Callback function for pool_list_get(). Adds the given pool to the AVL tree + * of known pools. + */ +static int +add_pool(zpool_handle_t *zhp, void *data) +{ + zpool_list_t *zlp = data; + zpool_node_t *node = safe_malloc(sizeof (zpool_node_t)); + uu_avl_index_t idx; + + node->zn_handle = zhp; + uu_avl_node_init(node, &node->zn_avlnode, zlp->zl_pool); + if (uu_avl_find(zlp->zl_avl, node, NULL, &idx) == NULL) { + if (zlp->zl_proplist && + zpool_expand_proplist(zhp, zlp->zl_proplist) != 0) { + zpool_close(zhp); + free(node); + return (-1); + } + uu_avl_insert(zlp->zl_avl, node, idx); + } else { + zpool_close(zhp); + free(node); + return (-1); + } + + return (0); +} + +/* + * Create a list of pools based on the given arguments. If we're given no + * arguments, then iterate over all pools in the system and add them to the AVL + * tree. Otherwise, add only those pool explicitly specified on the command + * line. + */ +zpool_list_t * +pool_list_get(int argc, char **argv, zprop_list_t **proplist, int *err) +{ + zpool_list_t *zlp; + + zlp = safe_malloc(sizeof (zpool_list_t)); + + zlp->zl_pool = uu_avl_pool_create("zfs_pool", sizeof (zpool_node_t), + offsetof(zpool_node_t, zn_avlnode), zpool_compare, UU_DEFAULT); + + if (zlp->zl_pool == NULL) + zpool_no_memory(); + + if ((zlp->zl_avl = uu_avl_create(zlp->zl_pool, NULL, + UU_DEFAULT)) == NULL) + zpool_no_memory(); + + zlp->zl_proplist = proplist; + + if (argc == 0) { + (void) zpool_iter(g_zfs, add_pool, zlp); + zlp->zl_findall = B_TRUE; + } else { + int i; + + for (i = 0; i < argc; i++) { + zpool_handle_t *zhp; + + if (zhp = zpool_open_canfail(g_zfs, argv[i])) { + if (add_pool(zhp, zlp) != 0) + *err = B_TRUE; + } else { + *err = B_TRUE; + } + } + } + + return (zlp); +} + +/* + * Search for any new pools, adding them to the list. We only add pools when no + * options were given on the command line. Otherwise, we keep the list fixed as + * those that were explicitly specified. + */ +void +pool_list_update(zpool_list_t *zlp) +{ + if (zlp->zl_findall) + (void) zpool_iter(g_zfs, add_pool, zlp); +} + +/* + * Iterate over all pools in the list, executing the callback for each + */ +int +pool_list_iter(zpool_list_t *zlp, int unavail, zpool_iter_f func, + void *data) +{ + zpool_node_t *node, *next_node; + int ret = 0; + + for (node = uu_avl_first(zlp->zl_avl); node != NULL; node = next_node) { + next_node = uu_avl_next(zlp->zl_avl, node); + if (zpool_get_state(node->zn_handle) != POOL_STATE_UNAVAIL || + unavail) + ret |= func(node->zn_handle, data); + } + + return (ret); +} + +/* + * Remove the given pool from the list. When running iostat, we want to remove + * those pools that no longer exist. + */ +void +pool_list_remove(zpool_list_t *zlp, zpool_handle_t *zhp) +{ + zpool_node_t search, *node; + + search.zn_handle = zhp; + if ((node = uu_avl_find(zlp->zl_avl, &search, NULL, NULL)) != NULL) { + uu_avl_remove(zlp->zl_avl, node); + zpool_close(node->zn_handle); + free(node); + } +} + +/* + * Free all the handles associated with this list. + */ +void +pool_list_free(zpool_list_t *zlp) +{ + uu_avl_walk_t *walk; + zpool_node_t *node; + + if ((walk = uu_avl_walk_start(zlp->zl_avl, UU_WALK_ROBUST)) == NULL) { + (void) fprintf(stderr, + gettext("internal error: out of memory")); + exit(1); + } + + while ((node = uu_avl_walk_next(walk)) != NULL) { + uu_avl_remove(zlp->zl_avl, node); + zpool_close(node->zn_handle); + free(node); + } + + uu_avl_walk_end(walk); + uu_avl_destroy(zlp->zl_avl); + uu_avl_pool_destroy(zlp->zl_pool); + + free(zlp); +} + +/* + * Returns the number of elements in the pool list. + */ +int +pool_list_count(zpool_list_t *zlp) +{ + return (uu_avl_numnodes(zlp->zl_avl)); +} + +/* + * High level function which iterates over all pools given on the command line, + * using the pool_list_* interfaces. + */ +int +for_each_pool(int argc, char **argv, boolean_t unavail, + zprop_list_t **proplist, zpool_iter_f func, void *data) +{ + zpool_list_t *list; + int ret = 0; + + if ((list = pool_list_get(argc, argv, proplist, &ret)) == NULL) + return (1); + + if (pool_list_iter(list, unavail, func, data) != 0) + ret = 1; + + pool_list_free(list); + + return (ret); +} diff --git a/cmd/zpool/zpool_main.c b/cmd/zpool/zpool_main.c new file mode 100644 index 000000000..54bba8645 --- /dev/null +++ b/cmd/zpool/zpool_main.c @@ -0,0 +1,3950 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ + +/* + * Copyright 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#include <assert.h> +#include <ctype.h> +#include <dirent.h> +#include <errno.h> +#include <fcntl.h> +#include <libgen.h> +#include <libintl.h> +#include <libuutil.h> +#include <locale.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <strings.h> +#include <unistd.h> +#include <priv.h> +#include <pwd.h> +#include <zone.h> +#include <sys/fs/zfs.h> + +#include <sys/stat.h> + +#include <libzfs.h> + +#include "zpool_util.h" +#include "zfs_comutil.h" + +static int zpool_do_create(int, char **); +static int zpool_do_destroy(int, char **); + +static int zpool_do_add(int, char **); +static int zpool_do_remove(int, char **); + +static int zpool_do_list(int, char **); +static int zpool_do_iostat(int, char **); +static int zpool_do_status(int, char **); + +static int zpool_do_online(int, char **); +static int zpool_do_offline(int, char **); +static int zpool_do_clear(int, char **); + +static int zpool_do_attach(int, char **); +static int zpool_do_detach(int, char **); +static int zpool_do_replace(int, char **); + +static int zpool_do_scrub(int, char **); + +static int zpool_do_import(int, char **); +static int zpool_do_export(int, char **); + +static int zpool_do_upgrade(int, char **); + +static int zpool_do_history(int, char **); + +static int zpool_do_get(int, char **); +static int zpool_do_set(int, char **); + +/* + * These libumem hooks provide a reasonable set of defaults for the allocator's + * debugging facilities. + */ + +#ifdef DEBUG +const char * +_umem_debug_init(void) +{ + return ("default,verbose"); /* $UMEM_DEBUG setting */ +} + +const char * +_umem_logging_init(void) +{ + return ("fail,contents"); /* $UMEM_LOGGING setting */ +} +#endif + +typedef enum { + HELP_ADD, + HELP_ATTACH, + HELP_CLEAR, + HELP_CREATE, + HELP_DESTROY, + HELP_DETACH, + HELP_EXPORT, + HELP_HISTORY, + HELP_IMPORT, + HELP_IOSTAT, + HELP_LIST, + HELP_OFFLINE, + HELP_ONLINE, + HELP_REPLACE, + HELP_REMOVE, + HELP_SCRUB, + HELP_STATUS, + HELP_UPGRADE, + HELP_GET, + HELP_SET +} zpool_help_t; + + +typedef struct zpool_command { + const char *name; + int (*func)(int, char **); + zpool_help_t usage; +} zpool_command_t; + +/* + * Master command table. Each ZFS command has a name, associated function, and + * usage message. The usage messages need to be internationalized, so we have + * to have a function to return the usage message based on a command index. + * + * These commands are organized according to how they are displayed in the usage + * message. An empty command (one with a NULL name) indicates an empty line in + * the generic usage message. + */ +static zpool_command_t command_table[] = { + { "create", zpool_do_create, HELP_CREATE }, + { "destroy", zpool_do_destroy, HELP_DESTROY }, + { NULL }, + { "add", zpool_do_add, HELP_ADD }, + { "remove", zpool_do_remove, HELP_REMOVE }, + { NULL }, + { "list", zpool_do_list, HELP_LIST }, + { "iostat", zpool_do_iostat, HELP_IOSTAT }, + { "status", zpool_do_status, HELP_STATUS }, + { NULL }, + { "online", zpool_do_online, HELP_ONLINE }, + { "offline", zpool_do_offline, HELP_OFFLINE }, + { "clear", zpool_do_clear, HELP_CLEAR }, + { NULL }, + { "attach", zpool_do_attach, HELP_ATTACH }, + { "detach", zpool_do_detach, HELP_DETACH }, + { "replace", zpool_do_replace, HELP_REPLACE }, + { NULL }, + { "scrub", zpool_do_scrub, HELP_SCRUB }, + { NULL }, + { "import", zpool_do_import, HELP_IMPORT }, + { "export", zpool_do_export, HELP_EXPORT }, + { "upgrade", zpool_do_upgrade, HELP_UPGRADE }, + { NULL }, + { "history", zpool_do_history, HELP_HISTORY }, + { "get", zpool_do_get, HELP_GET }, + { "set", zpool_do_set, HELP_SET }, +}; + +#define NCOMMAND (sizeof (command_table) / sizeof (command_table[0])) + +zpool_command_t *current_command; +static char history_str[HIS_MAX_RECORD_LEN]; + +static const char * +get_usage(zpool_help_t idx) { + switch (idx) { + case HELP_ADD: + return (gettext("\tadd [-fn] <pool> <vdev> ...\n")); + case HELP_ATTACH: + return (gettext("\tattach [-f] <pool> <device> " + "<new-device>\n")); + case HELP_CLEAR: + return (gettext("\tclear <pool> [device]\n")); + case HELP_CREATE: + return (gettext("\tcreate [-fn] [-o property=value] ... \n" + "\t [-O file-system-property=value] ... \n" + "\t [-m mountpoint] [-R root] <pool> <vdev> ...\n")); + case HELP_DESTROY: + return (gettext("\tdestroy [-f] <pool>\n")); + case HELP_DETACH: + return (gettext("\tdetach <pool> <device>\n")); + case HELP_EXPORT: + return (gettext("\texport [-f] <pool> ...\n")); + case HELP_HISTORY: + return (gettext("\thistory [-il] [<pool>] ...\n")); + case HELP_IMPORT: + return (gettext("\timport [-d dir] [-D]\n" + "\timport [-o mntopts] [-o property=value] ... \n" + "\t [-d dir | -c cachefile] [-D] [-f] [-R root] -a\n" + "\timport [-o mntopts] [-o property=value] ... \n" + "\t [-d dir | -c cachefile] [-D] [-f] [-R root] " + "<pool | id> [newpool]\n")); + case HELP_IOSTAT: + return (gettext("\tiostat [-v] [pool] ... [interval " + "[count]]\n")); + case HELP_LIST: + return (gettext("\tlist [-H] [-o property[,...]] " + "[pool] ...\n")); + case HELP_OFFLINE: + return (gettext("\toffline [-t] <pool> <device> ...\n")); + case HELP_ONLINE: + return (gettext("\tonline <pool> <device> ...\n")); + case HELP_REPLACE: + return (gettext("\treplace [-f] <pool> <device> " + "[new-device]\n")); + case HELP_REMOVE: + return (gettext("\tremove <pool> <device> ...\n")); + case HELP_SCRUB: + return (gettext("\tscrub [-s] <pool> ...\n")); + case HELP_STATUS: + return (gettext("\tstatus [-vx] [pool] ...\n")); + case HELP_UPGRADE: + return (gettext("\tupgrade\n" + "\tupgrade -v\n" + "\tupgrade [-V version] <-a | pool ...>\n")); + case HELP_GET: + return (gettext("\tget <\"all\" | property[,...]> " + "<pool> ...\n")); + case HELP_SET: + return (gettext("\tset <property=value> <pool> \n")); + } + + abort(); + /* NOTREACHED */ +} + + +/* + * Callback routine that will print out a pool property value. + */ +static int +print_prop_cb(int prop, void *cb) +{ + FILE *fp = cb; + + (void) fprintf(fp, "\t%-13s ", zpool_prop_to_name(prop)); + + if (zpool_prop_readonly(prop)) + (void) fprintf(fp, " NO "); + else + (void) fprintf(fp, " YES "); + + if (zpool_prop_values(prop) == NULL) + (void) fprintf(fp, "-\n"); + else + (void) fprintf(fp, "%s\n", zpool_prop_values(prop)); + + return (ZPROP_CONT); +} + +/* + * Display usage message. If we're inside a command, display only the usage for + * that command. Otherwise, iterate over the entire command table and display + * a complete usage message. + */ +void +usage(boolean_t requested) +{ + FILE *fp = requested ? stdout : stderr; + + if (current_command == NULL) { + int i; + + (void) fprintf(fp, gettext("usage: zpool command args ...\n")); + (void) fprintf(fp, + gettext("where 'command' is one of the following:\n\n")); + + for (i = 0; i < NCOMMAND; i++) { + if (command_table[i].name == NULL) + (void) fprintf(fp, "\n"); + else + (void) fprintf(fp, "%s", + get_usage(command_table[i].usage)); + } + } else { + (void) fprintf(fp, gettext("usage:\n")); + (void) fprintf(fp, "%s", get_usage(current_command->usage)); + } + + if (current_command != NULL && + ((strcmp(current_command->name, "set") == 0) || + (strcmp(current_command->name, "get") == 0) || + (strcmp(current_command->name, "list") == 0))) { + + (void) fprintf(fp, + gettext("\nthe following properties are supported:\n")); + + (void) fprintf(fp, "\n\t%-13s %s %s\n\n", + "PROPERTY", "EDIT", "VALUES"); + + /* Iterate over all properties */ + (void) zprop_iter(print_prop_cb, fp, B_FALSE, B_TRUE, + ZFS_TYPE_POOL); + } + + /* + * See comments at end of main(). + */ + if (getenv("ZFS_ABORT") != NULL) { + (void) printf("dumping core by request\n"); + abort(); + } + + exit(requested ? 0 : 2); +} + +void +print_vdev_tree(zpool_handle_t *zhp, const char *name, nvlist_t *nv, int indent, + boolean_t print_logs) +{ + nvlist_t **child; + uint_t c, children; + char *vname; + + if (name != NULL) + (void) printf("\t%*s%s\n", indent, "", name); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) + return; + + for (c = 0; c < children; c++) { + uint64_t is_log = B_FALSE; + + (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG, + &is_log); + if ((is_log && !print_logs) || (!is_log && print_logs)) + continue; + + vname = zpool_vdev_name(g_zfs, zhp, child[c]); + print_vdev_tree(zhp, vname, child[c], indent + 2, + B_FALSE); + free(vname); + } +} + +/* + * Add a property pair (name, string-value) into a property nvlist. + */ +static int +add_prop_list(const char *propname, char *propval, nvlist_t **props, + boolean_t poolprop) +{ + zpool_prop_t prop = ZPROP_INVAL; + zfs_prop_t fprop; + nvlist_t *proplist; + const char *normnm; + char *strval; + + if (*props == NULL && + nvlist_alloc(props, NV_UNIQUE_NAME, 0) != 0) { + (void) fprintf(stderr, + gettext("internal error: out of memory\n")); + return (1); + } + + proplist = *props; + + if (poolprop) { + if ((prop = zpool_name_to_prop(propname)) == ZPROP_INVAL) { + (void) fprintf(stderr, gettext("property '%s' is " + "not a valid pool property\n"), propname); + return (2); + } + normnm = zpool_prop_to_name(prop); + } else { + if ((fprop = zfs_name_to_prop(propname)) == ZPROP_INVAL) { + (void) fprintf(stderr, gettext("property '%s' is " + "not a valid file system property\n"), propname); + return (2); + } + normnm = zfs_prop_to_name(fprop); + } + + if (nvlist_lookup_string(proplist, normnm, &strval) == 0 && + prop != ZPOOL_PROP_CACHEFILE) { + (void) fprintf(stderr, gettext("property '%s' " + "specified multiple times\n"), propname); + return (2); + } + + if (nvlist_add_string(proplist, normnm, propval) != 0) { + (void) fprintf(stderr, gettext("internal " + "error: out of memory\n")); + return (1); + } + + return (0); +} + +/* + * zpool add [-fn] <pool> <vdev> ... + * + * -f Force addition of devices, even if they appear in use + * -n Do not add the devices, but display the resulting layout if + * they were to be added. + * + * Adds the given vdevs to 'pool'. As with create, the bulk of this work is + * handled by get_vdev_spec(), which constructs the nvlist needed to pass to + * libzfs. + */ +int +zpool_do_add(int argc, char **argv) +{ + boolean_t force = B_FALSE; + boolean_t dryrun = B_FALSE; + int c; + nvlist_t *nvroot; + char *poolname; + int ret; + zpool_handle_t *zhp; + nvlist_t *config; + + /* check options */ + while ((c = getopt(argc, argv, "fn")) != -1) { + switch (c) { + case 'f': + force = B_TRUE; + break; + case 'n': + dryrun = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* get pool name and check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name argument\n")); + usage(B_FALSE); + } + if (argc < 2) { + (void) fprintf(stderr, gettext("missing vdev specification\n")); + usage(B_FALSE); + } + + poolname = argv[0]; + + argc--; + argv++; + + if ((zhp = zpool_open(g_zfs, poolname)) == NULL) + return (1); + + if ((config = zpool_get_config(zhp, NULL)) == NULL) { + (void) fprintf(stderr, gettext("pool '%s' is unavailable\n"), + poolname); + zpool_close(zhp); + return (1); + } + + /* pass off to get_vdev_spec for processing */ + nvroot = make_root_vdev(zhp, force, !force, B_FALSE, dryrun, + argc, argv); + if (nvroot == NULL) { + zpool_close(zhp); + return (1); + } + + if (dryrun) { + nvlist_t *poolnvroot; + + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &poolnvroot) == 0); + + (void) printf(gettext("would update '%s' to the following " + "configuration:\n"), zpool_get_name(zhp)); + + /* print original main pool and new tree */ + print_vdev_tree(zhp, poolname, poolnvroot, 0, B_FALSE); + print_vdev_tree(zhp, NULL, nvroot, 0, B_FALSE); + + /* Do the same for the logs */ + if (num_logs(poolnvroot) > 0) { + print_vdev_tree(zhp, "logs", poolnvroot, 0, B_TRUE); + print_vdev_tree(zhp, NULL, nvroot, 0, B_TRUE); + } else if (num_logs(nvroot) > 0) { + print_vdev_tree(zhp, "logs", nvroot, 0, B_TRUE); + } + + ret = 0; + } else { + ret = (zpool_add(zhp, nvroot) != 0); + } + + nvlist_free(nvroot); + zpool_close(zhp); + + return (ret); +} + +/* + * zpool remove <pool> <vdev> ... + * + * Removes the given vdev from the pool. Currently, this only supports removing + * spares and cache devices from the pool. Eventually, we'll want to support + * removing leaf vdevs (as an alias for 'detach') as well as toplevel vdevs. + */ +int +zpool_do_remove(int argc, char **argv) +{ + char *poolname; + int i, ret = 0; + zpool_handle_t *zhp; + + argc--; + argv++; + + /* get pool name and check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name argument\n")); + usage(B_FALSE); + } + if (argc < 2) { + (void) fprintf(stderr, gettext("missing device\n")); + usage(B_FALSE); + } + + poolname = argv[0]; + + if ((zhp = zpool_open(g_zfs, poolname)) == NULL) + return (1); + + for (i = 1; i < argc; i++) { + if (zpool_vdev_remove(zhp, argv[i]) != 0) + ret = 1; + } + + return (ret); +} + +/* + * zpool create [-fn] [-o property=value] ... + * [-O file-system-property=value] ... + * [-R root] [-m mountpoint] <pool> <dev> ... + * + * -f Force creation, even if devices appear in use + * -n Do not create the pool, but display the resulting layout if it + * were to be created. + * -R Create a pool under an alternate root + * -m Set default mountpoint for the root dataset. By default it's + * '/<pool>' + * -o Set property=value. + * -O Set fsproperty=value in the pool's root file system + * + * Creates the named pool according to the given vdev specification. The + * bulk of the vdev processing is done in get_vdev_spec() in zpool_vdev.c. Once + * we get the nvlist back from get_vdev_spec(), we either print out the contents + * (if '-n' was specified), or pass it to libzfs to do the creation. + */ +int +zpool_do_create(int argc, char **argv) +{ + boolean_t force = B_FALSE; + boolean_t dryrun = B_FALSE; + int c; + nvlist_t *nvroot = NULL; + char *poolname; + int ret = 1; + char *altroot = NULL; + char *mountpoint = NULL; + nvlist_t *fsprops = NULL; + nvlist_t *props = NULL; + char *propval; + + /* check options */ + while ((c = getopt(argc, argv, ":fnR:m:o:O:")) != -1) { + switch (c) { + case 'f': + force = B_TRUE; + break; + case 'n': + dryrun = B_TRUE; + break; + case 'R': + altroot = optarg; + if (add_prop_list(zpool_prop_to_name( + ZPOOL_PROP_ALTROOT), optarg, &props, B_TRUE)) + goto errout; + if (nvlist_lookup_string(props, + zpool_prop_to_name(ZPOOL_PROP_CACHEFILE), + &propval) == 0) + break; + if (add_prop_list(zpool_prop_to_name( + ZPOOL_PROP_CACHEFILE), "none", &props, B_TRUE)) + goto errout; + break; + case 'm': + mountpoint = optarg; + break; + case 'o': + if ((propval = strchr(optarg, '=')) == NULL) { + (void) fprintf(stderr, gettext("missing " + "'=' for -o option\n")); + goto errout; + } + *propval = '\0'; + propval++; + + if (add_prop_list(optarg, propval, &props, B_TRUE)) + goto errout; + break; + case 'O': + if ((propval = strchr(optarg, '=')) == NULL) { + (void) fprintf(stderr, gettext("missing " + "'=' for -O option\n")); + goto errout; + } + *propval = '\0'; + propval++; + + if (add_prop_list(optarg, propval, &fsprops, B_FALSE)) + goto errout; + break; + case ':': + (void) fprintf(stderr, gettext("missing argument for " + "'%c' option\n"), optopt); + goto badusage; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + goto badusage; + } + } + + argc -= optind; + argv += optind; + + /* get pool name and check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name argument\n")); + goto badusage; + } + if (argc < 2) { + (void) fprintf(stderr, gettext("missing vdev specification\n")); + goto badusage; + } + + poolname = argv[0]; + + /* + * As a special case, check for use of '/' in the name, and direct the + * user to use 'zfs create' instead. + */ + if (strchr(poolname, '/') != NULL) { + (void) fprintf(stderr, gettext("cannot create '%s': invalid " + "character '/' in pool name\n"), poolname); + (void) fprintf(stderr, gettext("use 'zfs create' to " + "create a dataset\n")); + goto errout; + } + + /* pass off to get_vdev_spec for bulk processing */ + nvroot = make_root_vdev(NULL, force, !force, B_FALSE, dryrun, + argc - 1, argv + 1); + if (nvroot == NULL) + goto errout; + + /* make_root_vdev() allows 0 toplevel children if there are spares */ + if (!zfs_allocatable_devs(nvroot)) { + (void) fprintf(stderr, gettext("invalid vdev " + "specification: at least one toplevel vdev must be " + "specified\n")); + goto errout; + } + + + if (altroot != NULL && altroot[0] != '/') { + (void) fprintf(stderr, gettext("invalid alternate root '%s': " + "must be an absolute path\n"), altroot); + goto errout; + } + + /* + * Check the validity of the mountpoint and direct the user to use the + * '-m' mountpoint option if it looks like its in use. + */ + if (mountpoint == NULL || + (strcmp(mountpoint, ZFS_MOUNTPOINT_LEGACY) != 0 && + strcmp(mountpoint, ZFS_MOUNTPOINT_NONE) != 0)) { + char buf[MAXPATHLEN]; + DIR *dirp; + + if (mountpoint && mountpoint[0] != '/') { + (void) fprintf(stderr, gettext("invalid mountpoint " + "'%s': must be an absolute path, 'legacy', or " + "'none'\n"), mountpoint); + goto errout; + } + + if (mountpoint == NULL) { + if (altroot != NULL) + (void) snprintf(buf, sizeof (buf), "%s/%s", + altroot, poolname); + else + (void) snprintf(buf, sizeof (buf), "/%s", + poolname); + } else { + if (altroot != NULL) + (void) snprintf(buf, sizeof (buf), "%s%s", + altroot, mountpoint); + else + (void) snprintf(buf, sizeof (buf), "%s", + mountpoint); + } + + if ((dirp = opendir(buf)) == NULL && errno != ENOENT) { + (void) fprintf(stderr, gettext("mountpoint '%s' : " + "%s\n"), buf, strerror(errno)); + (void) fprintf(stderr, gettext("use '-m' " + "option to provide a different default\n")); + goto errout; + } else if (dirp) { + int count = 0; + + while (count < 3 && readdir(dirp) != NULL) + count++; + (void) closedir(dirp); + + if (count > 2) { + (void) fprintf(stderr, gettext("mountpoint " + "'%s' exists and is not empty\n"), buf); + (void) fprintf(stderr, gettext("use '-m' " + "option to provide a " + "different default\n")); + goto errout; + } + } + } + + if (dryrun) { + /* + * For a dry run invocation, print out a basic message and run + * through all the vdevs in the list and print out in an + * appropriate hierarchy. + */ + (void) printf(gettext("would create '%s' with the " + "following layout:\n\n"), poolname); + + print_vdev_tree(NULL, poolname, nvroot, 0, B_FALSE); + if (num_logs(nvroot) > 0) + print_vdev_tree(NULL, "logs", nvroot, 0, B_TRUE); + + ret = 0; + } else { + /* + * Hand off to libzfs. + */ + if (zpool_create(g_zfs, poolname, + nvroot, props, fsprops) == 0) { + zfs_handle_t *pool = zfs_open(g_zfs, poolname, + ZFS_TYPE_FILESYSTEM); + if (pool != NULL) { + if (mountpoint != NULL) + verify(zfs_prop_set(pool, + zfs_prop_to_name( + ZFS_PROP_MOUNTPOINT), + mountpoint) == 0); + if (zfs_mount(pool, NULL, 0) == 0) + ret = zfs_shareall(pool); + zfs_close(pool); + } + } else if (libzfs_errno(g_zfs) == EZFS_INVALIDNAME) { + (void) fprintf(stderr, gettext("pool name may have " + "been omitted\n")); + } + } + +errout: + nvlist_free(nvroot); + nvlist_free(fsprops); + nvlist_free(props); + return (ret); +badusage: + nvlist_free(fsprops); + nvlist_free(props); + usage(B_FALSE); + return (2); +} + +/* + * zpool destroy <pool> + * + * -f Forcefully unmount any datasets + * + * Destroy the given pool. Automatically unmounts any datasets in the pool. + */ +int +zpool_do_destroy(int argc, char **argv) +{ + boolean_t force = B_FALSE; + int c; + char *pool; + zpool_handle_t *zhp; + int ret; + + /* check options */ + while ((c = getopt(argc, argv, "f")) != -1) { + switch (c) { + case 'f': + force = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* check arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool argument\n")); + usage(B_FALSE); + } + if (argc > 1) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + pool = argv[0]; + + if ((zhp = zpool_open_canfail(g_zfs, pool)) == NULL) { + /* + * As a special case, check for use of '/' in the name, and + * direct the user to use 'zfs destroy' instead. + */ + if (strchr(pool, '/') != NULL) + (void) fprintf(stderr, gettext("use 'zfs destroy' to " + "destroy a dataset\n")); + return (1); + } + + if (zpool_disable_datasets(zhp, force) != 0) { + (void) fprintf(stderr, gettext("could not destroy '%s': " + "could not unmount datasets\n"), zpool_get_name(zhp)); + return (1); + } + + ret = (zpool_destroy(zhp) != 0); + + zpool_close(zhp); + + return (ret); +} + +/* + * zpool export [-f] <pool> ... + * + * -f Forcefully unmount datasets + * + * Export the given pools. By default, the command will attempt to cleanly + * unmount any active datasets within the pool. If the '-f' flag is specified, + * then the datasets will be forcefully unmounted. + */ +int +zpool_do_export(int argc, char **argv) +{ + boolean_t force = B_FALSE; + int c; + zpool_handle_t *zhp; + int ret; + int i; + + /* check options */ + while ((c = getopt(argc, argv, "f")) != -1) { + switch (c) { + case 'f': + force = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* check arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool argument\n")); + usage(B_FALSE); + } + + ret = 0; + for (i = 0; i < argc; i++) { + if ((zhp = zpool_open_canfail(g_zfs, argv[i])) == NULL) { + ret = 1; + continue; + } + + if (zpool_disable_datasets(zhp, force) != 0) { + ret = 1; + zpool_close(zhp); + continue; + } + + if (zpool_export(zhp, force) != 0) + ret = 1; + + zpool_close(zhp); + } + + return (ret); +} + +/* + * Given a vdev configuration, determine the maximum width needed for the device + * name column. + */ +static int +max_width(zpool_handle_t *zhp, nvlist_t *nv, int depth, int max) +{ + char *name = zpool_vdev_name(g_zfs, zhp, nv); + nvlist_t **child; + uint_t c, children; + int ret; + + if (strlen(name) + depth > max) + max = strlen(name) + depth; + + free(name); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES, + &child, &children) == 0) { + for (c = 0; c < children; c++) + if ((ret = max_width(zhp, child[c], depth + 2, + max)) > max) + max = ret; + } + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE, + &child, &children) == 0) { + for (c = 0; c < children; c++) + if ((ret = max_width(zhp, child[c], depth + 2, + max)) > max) + max = ret; + } + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) == 0) { + for (c = 0; c < children; c++) + if ((ret = max_width(zhp, child[c], depth + 2, + max)) > max) + max = ret; + } + + + return (max); +} + + +/* + * Print the configuration of an exported pool. Iterate over all vdevs in the + * pool, printing out the name and status for each one. + */ +void +print_import_config(const char *name, nvlist_t *nv, int namewidth, int depth, + boolean_t print_logs) +{ + nvlist_t **child; + uint_t c, children; + vdev_stat_t *vs; + char *type, *vname; + + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) == 0); + if (strcmp(type, VDEV_TYPE_MISSING) == 0) + return; + + verify(nvlist_lookup_uint64_array(nv, ZPOOL_CONFIG_STATS, + (uint64_t **)&vs, &c) == 0); + + (void) printf("\t%*s%-*s", depth, "", namewidth - depth, name); + (void) printf(" %s", zpool_state_to_name(vs->vs_state, vs->vs_aux)); + + if (vs->vs_aux != 0) { + (void) printf(" "); + + switch (vs->vs_aux) { + case VDEV_AUX_OPEN_FAILED: + (void) printf(gettext("cannot open")); + break; + + case VDEV_AUX_BAD_GUID_SUM: + (void) printf(gettext("missing device")); + break; + + case VDEV_AUX_NO_REPLICAS: + (void) printf(gettext("insufficient replicas")); + break; + + case VDEV_AUX_VERSION_NEWER: + (void) printf(gettext("newer version")); + break; + + case VDEV_AUX_ERR_EXCEEDED: + (void) printf(gettext("too many errors")); + break; + + default: + (void) printf(gettext("corrupted data")); + break; + } + } + (void) printf("\n"); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) + return; + + for (c = 0; c < children; c++) { + uint64_t is_log = B_FALSE; + + (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG, + &is_log); + if ((is_log && !print_logs) || (!is_log && print_logs)) + continue; + + vname = zpool_vdev_name(g_zfs, NULL, child[c]); + print_import_config(vname, child[c], + namewidth, depth + 2, B_FALSE); + free(vname); + } + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE, + &child, &children) == 0) { + (void) printf(gettext("\tcache\n")); + for (c = 0; c < children; c++) { + vname = zpool_vdev_name(g_zfs, NULL, child[c]); + (void) printf("\t %s\n", vname); + free(vname); + } + } + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES, + &child, &children) == 0) { + (void) printf(gettext("\tspares\n")); + for (c = 0; c < children; c++) { + vname = zpool_vdev_name(g_zfs, NULL, child[c]); + (void) printf("\t %s\n", vname); + free(vname); + } + } +} + +/* + * Display the status for the given pool. + */ +static void +show_import(nvlist_t *config) +{ + uint64_t pool_state; + vdev_stat_t *vs; + char *name; + uint64_t guid; + char *msgid; + nvlist_t *nvroot; + int reason; + const char *health; + uint_t vsc; + int namewidth; + + verify(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME, + &name) == 0); + verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, + &guid) == 0); + verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE, + &pool_state) == 0); + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &nvroot) == 0); + + verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_STATS, + (uint64_t **)&vs, &vsc) == 0); + health = zpool_state_to_name(vs->vs_state, vs->vs_aux); + + reason = zpool_import_status(config, &msgid); + + (void) printf(gettext(" pool: %s\n"), name); + (void) printf(gettext(" id: %llu\n"), (u_longlong_t)guid); + (void) printf(gettext(" state: %s"), health); + if (pool_state == POOL_STATE_DESTROYED) + (void) printf(gettext(" (DESTROYED)")); + (void) printf("\n"); + + switch (reason) { + case ZPOOL_STATUS_MISSING_DEV_R: + case ZPOOL_STATUS_MISSING_DEV_NR: + case ZPOOL_STATUS_BAD_GUID_SUM: + (void) printf(gettext("status: One or more devices are missing " + "from the system.\n")); + break; + + case ZPOOL_STATUS_CORRUPT_LABEL_R: + case ZPOOL_STATUS_CORRUPT_LABEL_NR: + (void) printf(gettext("status: One or more devices contains " + "corrupted data.\n")); + break; + + case ZPOOL_STATUS_CORRUPT_DATA: + (void) printf(gettext("status: The pool data is corrupted.\n")); + break; + + case ZPOOL_STATUS_OFFLINE_DEV: + (void) printf(gettext("status: One or more devices " + "are offlined.\n")); + break; + + case ZPOOL_STATUS_CORRUPT_POOL: + (void) printf(gettext("status: The pool metadata is " + "corrupted.\n")); + break; + + case ZPOOL_STATUS_VERSION_OLDER: + (void) printf(gettext("status: The pool is formatted using an " + "older on-disk version.\n")); + break; + + case ZPOOL_STATUS_VERSION_NEWER: + (void) printf(gettext("status: The pool is formatted using an " + "incompatible version.\n")); + break; + + case ZPOOL_STATUS_HOSTID_MISMATCH: + (void) printf(gettext("status: The pool was last accessed by " + "another system.\n")); + break; + + case ZPOOL_STATUS_FAULTED_DEV_R: + case ZPOOL_STATUS_FAULTED_DEV_NR: + (void) printf(gettext("status: One or more devices are " + "faulted.\n")); + break; + + case ZPOOL_STATUS_BAD_LOG: + (void) printf(gettext("status: An intent log record cannot be " + "read.\n")); + break; + + default: + /* + * No other status can be seen when importing pools. + */ + assert(reason == ZPOOL_STATUS_OK); + } + + /* + * Print out an action according to the overall state of the pool. + */ + if (vs->vs_state == VDEV_STATE_HEALTHY) { + if (reason == ZPOOL_STATUS_VERSION_OLDER) + (void) printf(gettext("action: The pool can be " + "imported using its name or numeric identifier, " + "though\n\tsome features will not be available " + "without an explicit 'zpool upgrade'.\n")); + else if (reason == ZPOOL_STATUS_HOSTID_MISMATCH) + (void) printf(gettext("action: The pool can be " + "imported using its name or numeric " + "identifier and\n\tthe '-f' flag.\n")); + else + (void) printf(gettext("action: The pool can be " + "imported using its name or numeric " + "identifier.\n")); + } else if (vs->vs_state == VDEV_STATE_DEGRADED) { + (void) printf(gettext("action: The pool can be imported " + "despite missing or damaged devices. The\n\tfault " + "tolerance of the pool may be compromised if imported.\n")); + } else { + switch (reason) { + case ZPOOL_STATUS_VERSION_NEWER: + (void) printf(gettext("action: The pool cannot be " + "imported. Access the pool on a system running " + "newer\n\tsoftware, or recreate the pool from " + "backup.\n")); + break; + case ZPOOL_STATUS_MISSING_DEV_R: + case ZPOOL_STATUS_MISSING_DEV_NR: + case ZPOOL_STATUS_BAD_GUID_SUM: + (void) printf(gettext("action: The pool cannot be " + "imported. Attach the missing\n\tdevices and try " + "again.\n")); + break; + default: + (void) printf(gettext("action: The pool cannot be " + "imported due to damaged devices or data.\n")); + } + } + + /* + * If the state is "closed" or "can't open", and the aux state + * is "corrupt data": + */ + if (((vs->vs_state == VDEV_STATE_CLOSED) || + (vs->vs_state == VDEV_STATE_CANT_OPEN)) && + (vs->vs_aux == VDEV_AUX_CORRUPT_DATA)) { + if (pool_state == POOL_STATE_DESTROYED) + (void) printf(gettext("\tThe pool was destroyed, " + "but can be imported using the '-Df' flags.\n")); + else if (pool_state != POOL_STATE_EXPORTED) + (void) printf(gettext("\tThe pool may be active on " + "another system, but can be imported using\n\t" + "the '-f' flag.\n")); + } + + if (msgid != NULL) + (void) printf(gettext(" see: http://www.sun.com/msg/%s\n"), + msgid); + + (void) printf(gettext("config:\n\n")); + + namewidth = max_width(NULL, nvroot, 0, 0); + if (namewidth < 10) + namewidth = 10; + + print_import_config(name, nvroot, namewidth, 0, B_FALSE); + if (num_logs(nvroot) > 0) { + (void) printf(gettext("\tlogs\n")); + print_import_config(name, nvroot, namewidth, 0, B_TRUE); + } + + if (reason == ZPOOL_STATUS_BAD_GUID_SUM) { + (void) printf(gettext("\n\tAdditional devices are known to " + "be part of this pool, though their\n\texact " + "configuration cannot be determined.\n")); + } +} + +/* + * Perform the import for the given configuration. This passes the heavy + * lifting off to zpool_import_props(), and then mounts the datasets contained + * within the pool. + */ +static int +do_import(nvlist_t *config, const char *newname, const char *mntopts, + int force, nvlist_t *props, boolean_t allowfaulted) +{ + zpool_handle_t *zhp; + char *name; + uint64_t state; + uint64_t version; + int error = 0; + + verify(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME, + &name) == 0); + + verify(nvlist_lookup_uint64(config, + ZPOOL_CONFIG_POOL_STATE, &state) == 0); + verify(nvlist_lookup_uint64(config, + ZPOOL_CONFIG_VERSION, &version) == 0); + if (version > SPA_VERSION) { + (void) fprintf(stderr, gettext("cannot import '%s': pool " + "is formatted using a newer ZFS version\n"), name); + return (1); + } else if (state != POOL_STATE_EXPORTED && !force) { + uint64_t hostid; + + if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_HOSTID, + &hostid) == 0) { + if ((unsigned long)hostid != gethostid()) { + char *hostname; + uint64_t timestamp; + time_t t; + + verify(nvlist_lookup_string(config, + ZPOOL_CONFIG_HOSTNAME, &hostname) == 0); + verify(nvlist_lookup_uint64(config, + ZPOOL_CONFIG_TIMESTAMP, ×tamp) == 0); + t = timestamp; + (void) fprintf(stderr, gettext("cannot import " + "'%s': pool may be in use from other " + "system, it was last accessed by %s " + "(hostid: 0x%lx) on %s"), name, hostname, + (unsigned long)hostid, + asctime(localtime(&t))); + (void) fprintf(stderr, gettext("use '-f' to " + "import anyway\n")); + return (1); + } + } else { + (void) fprintf(stderr, gettext("cannot import '%s': " + "pool may be in use from other system\n"), name); + (void) fprintf(stderr, gettext("use '-f' to import " + "anyway\n")); + return (1); + } + } + + if (zpool_import_props(g_zfs, config, newname, props, + allowfaulted) != 0) + return (1); + + if (newname != NULL) + name = (char *)newname; + + verify((zhp = zpool_open_canfail(g_zfs, name)) != NULL); + + if (zpool_enable_datasets(zhp, mntopts, 0) != 0) { + zpool_close(zhp); + return (1); + } + + zpool_close(zhp); + return (error); +} + +/* + * zpool import [-d dir] [-D] + * import [-o mntopts] [-o prop=value] ... [-R root] [-D] + * [-d dir | -c cachefile] [-f] -a + * import [-o mntopts] [-o prop=value] ... [-R root] [-D] + * [-d dir | -c cachefile] [-f] <pool | id> [newpool] + * + * -c Read pool information from a cachefile instead of searching + * devices. + * + * -d Scan in a specific directory, other than /dev/dsk. More than + * one directory can be specified using multiple '-d' options. + * + * -D Scan for previously destroyed pools or import all or only + * specified destroyed pools. + * + * -R Temporarily import the pool, with all mountpoints relative to + * the given root. The pool will remain exported when the machine + * is rebooted. + * + * -f Force import, even if it appears that the pool is active. + * + * -F Import even in the presence of faulted vdevs. This is an + * intentionally undocumented option for testing purposes, and + * treats the pool configuration as complete, leaving any bad + * vdevs in the FAULTED state. + * + * -a Import all pools found. + * + * -o Set property=value and/or temporary mount options (without '='). + * + * The import command scans for pools to import, and import pools based on pool + * name and GUID. The pool can also be renamed as part of the import process. + */ +int +zpool_do_import(int argc, char **argv) +{ + char **searchdirs = NULL; + int nsearch = 0; + int c; + int err; + nvlist_t *pools = NULL; + boolean_t do_all = B_FALSE; + boolean_t do_destroyed = B_FALSE; + char *mntopts = NULL; + boolean_t do_force = B_FALSE; + nvpair_t *elem; + nvlist_t *config; + uint64_t searchguid = 0; + char *searchname = NULL; + char *propval; + nvlist_t *found_config; + nvlist_t *props = NULL; + boolean_t first; + boolean_t allow_faulted = B_FALSE; + uint64_t pool_state; + char *cachefile = NULL; + + /* check options */ + while ((c = getopt(argc, argv, ":ac:d:DfFo:p:R:")) != -1) { + switch (c) { + case 'a': + do_all = B_TRUE; + break; + case 'c': + cachefile = optarg; + break; + case 'd': + if (searchdirs == NULL) { + searchdirs = safe_malloc(sizeof (char *)); + } else { + char **tmp = safe_malloc((nsearch + 1) * + sizeof (char *)); + bcopy(searchdirs, tmp, nsearch * + sizeof (char *)); + free(searchdirs); + searchdirs = tmp; + } + searchdirs[nsearch++] = optarg; + break; + case 'D': + do_destroyed = B_TRUE; + break; + case 'f': + do_force = B_TRUE; + break; + case 'F': + allow_faulted = B_TRUE; + break; + case 'o': + if ((propval = strchr(optarg, '=')) != NULL) { + *propval = '\0'; + propval++; + if (add_prop_list(optarg, propval, + &props, B_TRUE)) + goto error; + } else { + mntopts = optarg; + } + break; + case 'R': + if (add_prop_list(zpool_prop_to_name( + ZPOOL_PROP_ALTROOT), optarg, &props, B_TRUE)) + goto error; + if (nvlist_lookup_string(props, + zpool_prop_to_name(ZPOOL_PROP_CACHEFILE), + &propval) == 0) + break; + if (add_prop_list(zpool_prop_to_name( + ZPOOL_PROP_CACHEFILE), "none", &props, B_TRUE)) + goto error; + break; + case ':': + (void) fprintf(stderr, gettext("missing argument for " + "'%c' option\n"), optopt); + usage(B_FALSE); + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + if (cachefile && nsearch != 0) { + (void) fprintf(stderr, gettext("-c is incompatible with -d\n")); + usage(B_FALSE); + } + + if (searchdirs == NULL) { + searchdirs = safe_malloc(sizeof (char *)); + searchdirs[0] = "/dev/dsk"; + nsearch = 1; + } + + /* check argument count */ + if (do_all) { + if (argc != 0) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + } else { + if (argc > 2) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + /* + * Check for the SYS_CONFIG privilege. We do this explicitly + * here because otherwise any attempt to discover pools will + * silently fail. + */ + if (argc == 0 && !priv_ineffect(PRIV_SYS_CONFIG)) { + (void) fprintf(stderr, gettext("cannot " + "discover pools: permission denied\n")); + free(searchdirs); + return (1); + } + } + + /* + * Depending on the arguments given, we do one of the following: + * + * <none> Iterate through all pools and display information about + * each one. + * + * -a Iterate through all pools and try to import each one. + * + * <id> Find the pool that corresponds to the given GUID/pool + * name and import that one. + * + * -D Above options applies only to destroyed pools. + */ + if (argc != 0) { + char *endptr; + + errno = 0; + searchguid = strtoull(argv[0], &endptr, 10); + if (errno != 0 || *endptr != '\0') + searchname = argv[0]; + found_config = NULL; + } + + if (cachefile) { + pools = zpool_find_import_cached(g_zfs, cachefile, searchname, + searchguid); + } else if (searchname != NULL) { + pools = zpool_find_import_byname(g_zfs, nsearch, searchdirs, + searchname); + } else { + /* + * It's OK to search by guid even if searchguid is 0. + */ + pools = zpool_find_import_byguid(g_zfs, nsearch, searchdirs, + searchguid); + } + + if (pools == NULL) { + if (argc != 0) { + (void) fprintf(stderr, gettext("cannot import '%s': " + "no such pool available\n"), argv[0]); + } + free(searchdirs); + return (1); + } + + /* + * At this point we have a list of import candidate configs. Even if + * we were searching by pool name or guid, we still need to + * post-process the list to deal with pool state and possible + * duplicate names. + */ + err = 0; + elem = NULL; + first = B_TRUE; + while ((elem = nvlist_next_nvpair(pools, elem)) != NULL) { + + verify(nvpair_value_nvlist(elem, &config) == 0); + + verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE, + &pool_state) == 0); + if (!do_destroyed && pool_state == POOL_STATE_DESTROYED) + continue; + if (do_destroyed && pool_state != POOL_STATE_DESTROYED) + continue; + + if (argc == 0) { + if (first) + first = B_FALSE; + else if (!do_all) + (void) printf("\n"); + + if (do_all) + err |= do_import(config, NULL, mntopts, + do_force, props, allow_faulted); + else + show_import(config); + } else if (searchname != NULL) { + char *name; + + /* + * We are searching for a pool based on name. + */ + verify(nvlist_lookup_string(config, + ZPOOL_CONFIG_POOL_NAME, &name) == 0); + + if (strcmp(name, searchname) == 0) { + if (found_config != NULL) { + (void) fprintf(stderr, gettext( + "cannot import '%s': more than " + "one matching pool\n"), searchname); + (void) fprintf(stderr, gettext( + "import by numeric ID instead\n")); + err = B_TRUE; + } + found_config = config; + } + } else { + uint64_t guid; + + /* + * Search for a pool by guid. + */ + verify(nvlist_lookup_uint64(config, + ZPOOL_CONFIG_POOL_GUID, &guid) == 0); + + if (guid == searchguid) + found_config = config; + } + } + + /* + * If we were searching for a specific pool, verify that we found a + * pool, and then do the import. + */ + if (argc != 0 && err == 0) { + if (found_config == NULL) { + (void) fprintf(stderr, gettext("cannot import '%s': " + "no such pool available\n"), argv[0]); + err = B_TRUE; + } else { + err |= do_import(found_config, argc == 1 ? NULL : + argv[1], mntopts, do_force, props, allow_faulted); + } + } + + /* + * If we were just looking for pools, report an error if none were + * found. + */ + if (argc == 0 && first) + (void) fprintf(stderr, + gettext("no pools available to import\n")); + +error: + nvlist_free(props); + nvlist_free(pools); + free(searchdirs); + + return (err ? 1 : 0); +} + +typedef struct iostat_cbdata { + zpool_list_t *cb_list; + int cb_verbose; + int cb_iteration; + int cb_namewidth; +} iostat_cbdata_t; + +static void +print_iostat_separator(iostat_cbdata_t *cb) +{ + int i = 0; + + for (i = 0; i < cb->cb_namewidth; i++) + (void) printf("-"); + (void) printf(" ----- ----- ----- ----- ----- -----\n"); +} + +static void +print_iostat_header(iostat_cbdata_t *cb) +{ + (void) printf("%*s capacity operations bandwidth\n", + cb->cb_namewidth, ""); + (void) printf("%-*s used avail read write read write\n", + cb->cb_namewidth, "pool"); + print_iostat_separator(cb); +} + +/* + * Display a single statistic. + */ +static void +print_one_stat(uint64_t value) +{ + char buf[64]; + + zfs_nicenum(value, buf, sizeof (buf)); + (void) printf(" %5s", buf); +} + +/* + * Print out all the statistics for the given vdev. This can either be the + * toplevel configuration, or called recursively. If 'name' is NULL, then this + * is a verbose output, and we don't want to display the toplevel pool stats. + */ +void +print_vdev_stats(zpool_handle_t *zhp, const char *name, nvlist_t *oldnv, + nvlist_t *newnv, iostat_cbdata_t *cb, int depth) +{ + nvlist_t **oldchild, **newchild; + uint_t c, children; + vdev_stat_t *oldvs, *newvs; + vdev_stat_t zerovs = { 0 }; + uint64_t tdelta; + double scale; + char *vname; + + if (oldnv != NULL) { + verify(nvlist_lookup_uint64_array(oldnv, ZPOOL_CONFIG_STATS, + (uint64_t **)&oldvs, &c) == 0); + } else { + oldvs = &zerovs; + } + + verify(nvlist_lookup_uint64_array(newnv, ZPOOL_CONFIG_STATS, + (uint64_t **)&newvs, &c) == 0); + + if (strlen(name) + depth > cb->cb_namewidth) + (void) printf("%*s%s", depth, "", name); + else + (void) printf("%*s%s%*s", depth, "", name, + (int)(cb->cb_namewidth - strlen(name) - depth), ""); + + tdelta = newvs->vs_timestamp - oldvs->vs_timestamp; + + if (tdelta == 0) + scale = 1.0; + else + scale = (double)NANOSEC / tdelta; + + /* only toplevel vdevs have capacity stats */ + if (newvs->vs_space == 0) { + (void) printf(" - -"); + } else { + print_one_stat(newvs->vs_alloc); + print_one_stat(newvs->vs_space - newvs->vs_alloc); + } + + print_one_stat((uint64_t)(scale * (newvs->vs_ops[ZIO_TYPE_READ] - + oldvs->vs_ops[ZIO_TYPE_READ]))); + + print_one_stat((uint64_t)(scale * (newvs->vs_ops[ZIO_TYPE_WRITE] - + oldvs->vs_ops[ZIO_TYPE_WRITE]))); + + print_one_stat((uint64_t)(scale * (newvs->vs_bytes[ZIO_TYPE_READ] - + oldvs->vs_bytes[ZIO_TYPE_READ]))); + + print_one_stat((uint64_t)(scale * (newvs->vs_bytes[ZIO_TYPE_WRITE] - + oldvs->vs_bytes[ZIO_TYPE_WRITE]))); + + (void) printf("\n"); + + if (!cb->cb_verbose) + return; + + if (nvlist_lookup_nvlist_array(newnv, ZPOOL_CONFIG_CHILDREN, + &newchild, &children) != 0) + return; + + if (oldnv && nvlist_lookup_nvlist_array(oldnv, ZPOOL_CONFIG_CHILDREN, + &oldchild, &c) != 0) + return; + + for (c = 0; c < children; c++) { + vname = zpool_vdev_name(g_zfs, zhp, newchild[c]); + print_vdev_stats(zhp, vname, oldnv ? oldchild[c] : NULL, + newchild[c], cb, depth + 2); + free(vname); + } + + /* + * Include level 2 ARC devices in iostat output + */ + if (nvlist_lookup_nvlist_array(newnv, ZPOOL_CONFIG_L2CACHE, + &newchild, &children) != 0) + return; + + if (oldnv && nvlist_lookup_nvlist_array(oldnv, ZPOOL_CONFIG_L2CACHE, + &oldchild, &c) != 0) + return; + + if (children > 0) { + (void) printf("%-*s - - - - - " + "-\n", cb->cb_namewidth, "cache"); + for (c = 0; c < children; c++) { + vname = zpool_vdev_name(g_zfs, zhp, newchild[c]); + print_vdev_stats(zhp, vname, oldnv ? oldchild[c] : NULL, + newchild[c], cb, depth + 2); + free(vname); + } + } +} + +static int +refresh_iostat(zpool_handle_t *zhp, void *data) +{ + iostat_cbdata_t *cb = data; + boolean_t missing; + + /* + * If the pool has disappeared, remove it from the list and continue. + */ + if (zpool_refresh_stats(zhp, &missing) != 0) + return (-1); + + if (missing) + pool_list_remove(cb->cb_list, zhp); + + return (0); +} + +/* + * Callback to print out the iostats for the given pool. + */ +int +print_iostat(zpool_handle_t *zhp, void *data) +{ + iostat_cbdata_t *cb = data; + nvlist_t *oldconfig, *newconfig; + nvlist_t *oldnvroot, *newnvroot; + + newconfig = zpool_get_config(zhp, &oldconfig); + + if (cb->cb_iteration == 1) + oldconfig = NULL; + + verify(nvlist_lookup_nvlist(newconfig, ZPOOL_CONFIG_VDEV_TREE, + &newnvroot) == 0); + + if (oldconfig == NULL) + oldnvroot = NULL; + else + verify(nvlist_lookup_nvlist(oldconfig, ZPOOL_CONFIG_VDEV_TREE, + &oldnvroot) == 0); + + /* + * Print out the statistics for the pool. + */ + print_vdev_stats(zhp, zpool_get_name(zhp), oldnvroot, newnvroot, cb, 0); + + if (cb->cb_verbose) + print_iostat_separator(cb); + + return (0); +} + +int +get_namewidth(zpool_handle_t *zhp, void *data) +{ + iostat_cbdata_t *cb = data; + nvlist_t *config, *nvroot; + + if ((config = zpool_get_config(zhp, NULL)) != NULL) { + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &nvroot) == 0); + if (!cb->cb_verbose) + cb->cb_namewidth = strlen(zpool_get_name(zhp)); + else + cb->cb_namewidth = max_width(zhp, nvroot, 0, 0); + } + + /* + * The width must fall into the range [10,38]. The upper limit is the + * maximum we can have and still fit in 80 columns. + */ + if (cb->cb_namewidth < 10) + cb->cb_namewidth = 10; + if (cb->cb_namewidth > 38) + cb->cb_namewidth = 38; + + return (0); +} + +/* + * zpool iostat [-v] [pool] ... [interval [count]] + * + * -v Display statistics for individual vdevs + * + * This command can be tricky because we want to be able to deal with pool + * creation/destruction as well as vdev configuration changes. The bulk of this + * processing is handled by the pool_list_* routines in zpool_iter.c. We rely + * on pool_list_update() to detect the addition of new pools. Configuration + * changes are all handled within libzfs. + */ +int +zpool_do_iostat(int argc, char **argv) +{ + int c; + int ret; + int npools; + unsigned long interval = 0, count = 0; + zpool_list_t *list; + boolean_t verbose = B_FALSE; + iostat_cbdata_t cb; + + /* check options */ + while ((c = getopt(argc, argv, "v")) != -1) { + switch (c) { + case 'v': + verbose = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* + * Determine if the last argument is an integer or a pool name + */ + if (argc > 0 && isdigit(argv[argc - 1][0])) { + char *end; + + errno = 0; + interval = strtoul(argv[argc - 1], &end, 10); + + if (*end == '\0' && errno == 0) { + if (interval == 0) { + (void) fprintf(stderr, gettext("interval " + "cannot be zero\n")); + usage(B_FALSE); + } + + /* + * Ignore the last parameter + */ + argc--; + } else { + /* + * If this is not a valid number, just plow on. The + * user will get a more informative error message later + * on. + */ + interval = 0; + } + } + + /* + * If the last argument is also an integer, then we have both a count + * and an integer. + */ + if (argc > 0 && isdigit(argv[argc - 1][0])) { + char *end; + + errno = 0; + count = interval; + interval = strtoul(argv[argc - 1], &end, 10); + + if (*end == '\0' && errno == 0) { + if (interval == 0) { + (void) fprintf(stderr, gettext("interval " + "cannot be zero\n")); + usage(B_FALSE); + } + + /* + * Ignore the last parameter + */ + argc--; + } else { + interval = 0; + } + } + + /* + * Construct the list of all interesting pools. + */ + ret = 0; + if ((list = pool_list_get(argc, argv, NULL, &ret)) == NULL) + return (1); + + if (pool_list_count(list) == 0 && argc != 0) { + pool_list_free(list); + return (1); + } + + if (pool_list_count(list) == 0 && interval == 0) { + pool_list_free(list); + (void) fprintf(stderr, gettext("no pools available\n")); + return (1); + } + + /* + * Enter the main iostat loop. + */ + cb.cb_list = list; + cb.cb_verbose = verbose; + cb.cb_iteration = 0; + cb.cb_namewidth = 0; + + for (;;) { + pool_list_update(list); + + if ((npools = pool_list_count(list)) == 0) + break; + + /* + * Refresh all statistics. This is done as an explicit step + * before calculating the maximum name width, so that any + * configuration changes are properly accounted for. + */ + (void) pool_list_iter(list, B_FALSE, refresh_iostat, &cb); + + /* + * Iterate over all pools to determine the maximum width + * for the pool / device name column across all pools. + */ + cb.cb_namewidth = 0; + (void) pool_list_iter(list, B_FALSE, get_namewidth, &cb); + + /* + * If it's the first time, or verbose mode, print the header. + */ + if (++cb.cb_iteration == 1 || verbose) + print_iostat_header(&cb); + + (void) pool_list_iter(list, B_FALSE, print_iostat, &cb); + + /* + * If there's more than one pool, and we're not in verbose mode + * (which prints a separator for us), then print a separator. + */ + if (npools > 1 && !verbose) + print_iostat_separator(&cb); + + if (verbose) + (void) printf("\n"); + + /* + * Flush the output so that redirection to a file isn't buffered + * indefinitely. + */ + (void) fflush(stdout); + + if (interval == 0) + break; + + if (count != 0 && --count == 0) + break; + + (void) sleep(interval); + } + + pool_list_free(list); + + return (ret); +} + +typedef struct list_cbdata { + boolean_t cb_scripted; + boolean_t cb_first; + zprop_list_t *cb_proplist; +} list_cbdata_t; + +/* + * Given a list of columns to display, output appropriate headers for each one. + */ +static void +print_header(zprop_list_t *pl) +{ + const char *header; + boolean_t first = B_TRUE; + boolean_t right_justify; + + for (; pl != NULL; pl = pl->pl_next) { + if (pl->pl_prop == ZPROP_INVAL) + continue; + + if (!first) + (void) printf(" "); + else + first = B_FALSE; + + header = zpool_prop_column_name(pl->pl_prop); + right_justify = zpool_prop_align_right(pl->pl_prop); + + if (pl->pl_next == NULL && !right_justify) + (void) printf("%s", header); + else if (right_justify) + (void) printf("%*s", pl->pl_width, header); + else + (void) printf("%-*s", pl->pl_width, header); + } + + (void) printf("\n"); +} + +/* + * Given a pool and a list of properties, print out all the properties according + * to the described layout. + */ +static void +print_pool(zpool_handle_t *zhp, zprop_list_t *pl, int scripted) +{ + boolean_t first = B_TRUE; + char property[ZPOOL_MAXPROPLEN]; + char *propstr; + boolean_t right_justify; + int width; + + for (; pl != NULL; pl = pl->pl_next) { + if (!first) { + if (scripted) + (void) printf("\t"); + else + (void) printf(" "); + } else { + first = B_FALSE; + } + + right_justify = B_FALSE; + if (pl->pl_prop != ZPROP_INVAL) { + if (zpool_get_prop(zhp, pl->pl_prop, property, + sizeof (property), NULL) != 0) + propstr = "-"; + else + propstr = property; + + right_justify = zpool_prop_align_right(pl->pl_prop); + } else { + propstr = "-"; + } + + width = pl->pl_width; + + /* + * If this is being called in scripted mode, or if this is the + * last column and it is left-justified, don't include a width + * format specifier. + */ + if (scripted || (pl->pl_next == NULL && !right_justify)) + (void) printf("%s", propstr); + else if (right_justify) + (void) printf("%*s", width, propstr); + else + (void) printf("%-*s", width, propstr); + } + + (void) printf("\n"); +} + +/* + * Generic callback function to list a pool. + */ +int +list_callback(zpool_handle_t *zhp, void *data) +{ + list_cbdata_t *cbp = data; + + if (cbp->cb_first) { + if (!cbp->cb_scripted) + print_header(cbp->cb_proplist); + cbp->cb_first = B_FALSE; + } + + print_pool(zhp, cbp->cb_proplist, cbp->cb_scripted); + + return (0); +} + +/* + * zpool list [-H] [-o prop[,prop]*] [pool] ... + * + * -H Scripted mode. Don't display headers, and separate properties + * by a single tab. + * -o List of properties to display. Defaults to + * "name,size,used,available,capacity,health,altroot" + * + * List all pools in the system, whether or not they're healthy. Output space + * statistics for each one, as well as health status summary. + */ +int +zpool_do_list(int argc, char **argv) +{ + int c; + int ret; + list_cbdata_t cb = { 0 }; + static char default_props[] = + "name,size,used,available,capacity,health,altroot"; + char *props = default_props; + + /* check options */ + while ((c = getopt(argc, argv, ":Ho:")) != -1) { + switch (c) { + case 'H': + cb.cb_scripted = B_TRUE; + break; + case 'o': + props = optarg; + break; + case ':': + (void) fprintf(stderr, gettext("missing argument for " + "'%c' option\n"), optopt); + usage(B_FALSE); + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + if (zprop_get_list(g_zfs, props, &cb.cb_proplist, ZFS_TYPE_POOL) != 0) + usage(B_FALSE); + + cb.cb_first = B_TRUE; + + ret = for_each_pool(argc, argv, B_TRUE, &cb.cb_proplist, + list_callback, &cb); + + zprop_free_list(cb.cb_proplist); + + if (argc == 0 && cb.cb_first && !cb.cb_scripted) { + (void) printf(gettext("no pools available\n")); + return (0); + } + + return (ret); +} + +static nvlist_t * +zpool_get_vdev_by_name(nvlist_t *nv, char *name) +{ + nvlist_t **child; + uint_t c, children; + nvlist_t *match; + char *path; + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) { + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) == 0); + if (strncmp(name, "/dev/dsk/", 9) == 0) + name += 9; + if (strncmp(path, "/dev/dsk/", 9) == 0) + path += 9; + if (strcmp(name, path) == 0) + return (nv); + return (NULL); + } + + for (c = 0; c < children; c++) + if ((match = zpool_get_vdev_by_name(child[c], name)) != NULL) + return (match); + + return (NULL); +} + +static int +zpool_do_attach_or_replace(int argc, char **argv, int replacing) +{ + boolean_t force = B_FALSE; + int c; + nvlist_t *nvroot; + char *poolname, *old_disk, *new_disk; + zpool_handle_t *zhp; + int ret; + + /* check options */ + while ((c = getopt(argc, argv, "f")) != -1) { + switch (c) { + case 'f': + force = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* get pool name and check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name argument\n")); + usage(B_FALSE); + } + + poolname = argv[0]; + + if (argc < 2) { + (void) fprintf(stderr, + gettext("missing <device> specification\n")); + usage(B_FALSE); + } + + old_disk = argv[1]; + + if (argc < 3) { + if (!replacing) { + (void) fprintf(stderr, + gettext("missing <new_device> specification\n")); + usage(B_FALSE); + } + new_disk = old_disk; + argc -= 1; + argv += 1; + } else { + new_disk = argv[2]; + argc -= 2; + argv += 2; + } + + if (argc > 1) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + if ((zhp = zpool_open(g_zfs, poolname)) == NULL) + return (1); + + if (zpool_get_config(zhp, NULL) == NULL) { + (void) fprintf(stderr, gettext("pool '%s' is unavailable\n"), + poolname); + zpool_close(zhp); + return (1); + } + + nvroot = make_root_vdev(zhp, force, B_FALSE, replacing, B_FALSE, + argc, argv); + if (nvroot == NULL) { + zpool_close(zhp); + return (1); + } + + ret = zpool_vdev_attach(zhp, old_disk, new_disk, nvroot, replacing); + + nvlist_free(nvroot); + zpool_close(zhp); + + return (ret); +} + +/* + * zpool replace [-f] <pool> <device> <new_device> + * + * -f Force attach, even if <new_device> appears to be in use. + * + * Replace <device> with <new_device>. + */ +/* ARGSUSED */ +int +zpool_do_replace(int argc, char **argv) +{ + return (zpool_do_attach_or_replace(argc, argv, B_TRUE)); +} + +/* + * zpool attach [-f] <pool> <device> <new_device> + * + * -f Force attach, even if <new_device> appears to be in use. + * + * Attach <new_device> to the mirror containing <device>. If <device> is not + * part of a mirror, then <device> will be transformed into a mirror of + * <device> and <new_device>. In either case, <new_device> will begin life + * with a DTL of [0, now], and will immediately begin to resilver itself. + */ +int +zpool_do_attach(int argc, char **argv) +{ + return (zpool_do_attach_or_replace(argc, argv, B_FALSE)); +} + +/* + * zpool detach [-f] <pool> <device> + * + * -f Force detach of <device>, even if DTLs argue against it + * (not supported yet) + * + * Detach a device from a mirror. The operation will be refused if <device> + * is the last device in the mirror, or if the DTLs indicate that this device + * has the only valid copy of some data. + */ +/* ARGSUSED */ +int +zpool_do_detach(int argc, char **argv) +{ + int c; + char *poolname, *path; + zpool_handle_t *zhp; + int ret; + + /* check options */ + while ((c = getopt(argc, argv, "f")) != -1) { + switch (c) { + case 'f': + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* get pool name and check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name argument\n")); + usage(B_FALSE); + } + + if (argc < 2) { + (void) fprintf(stderr, + gettext("missing <device> specification\n")); + usage(B_FALSE); + } + + poolname = argv[0]; + path = argv[1]; + + if ((zhp = zpool_open(g_zfs, poolname)) == NULL) + return (1); + + ret = zpool_vdev_detach(zhp, path); + + zpool_close(zhp); + + return (ret); +} + +/* + * zpool online <pool> <device> ... + */ +int +zpool_do_online(int argc, char **argv) +{ + int c, i; + char *poolname; + zpool_handle_t *zhp; + int ret = 0; + vdev_state_t newstate; + + /* check options */ + while ((c = getopt(argc, argv, "t")) != -1) { + switch (c) { + case 't': + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* get pool name and check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name\n")); + usage(B_FALSE); + } + if (argc < 2) { + (void) fprintf(stderr, gettext("missing device name\n")); + usage(B_FALSE); + } + + poolname = argv[0]; + + if ((zhp = zpool_open(g_zfs, poolname)) == NULL) + return (1); + + for (i = 1; i < argc; i++) { + if (zpool_vdev_online(zhp, argv[i], 0, &newstate) == 0) { + if (newstate != VDEV_STATE_HEALTHY) { + (void) printf(gettext("warning: device '%s' " + "onlined, but remains in faulted state\n"), + argv[i]); + if (newstate == VDEV_STATE_FAULTED) + (void) printf(gettext("use 'zpool " + "clear' to restore a faulted " + "device\n")); + else + (void) printf(gettext("use 'zpool " + "replace' to replace devices " + "that are no longer present\n")); + } + } else { + ret = 1; + } + } + + zpool_close(zhp); + + return (ret); +} + +/* + * zpool offline [-ft] <pool> <device> ... + * + * -f Force the device into the offline state, even if doing + * so would appear to compromise pool availability. + * (not supported yet) + * + * -t Only take the device off-line temporarily. The offline + * state will not be persistent across reboots. + */ +/* ARGSUSED */ +int +zpool_do_offline(int argc, char **argv) +{ + int c, i; + char *poolname; + zpool_handle_t *zhp; + int ret = 0; + boolean_t istmp = B_FALSE; + + /* check options */ + while ((c = getopt(argc, argv, "ft")) != -1) { + switch (c) { + case 't': + istmp = B_TRUE; + break; + case 'f': + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* get pool name and check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name\n")); + usage(B_FALSE); + } + if (argc < 2) { + (void) fprintf(stderr, gettext("missing device name\n")); + usage(B_FALSE); + } + + poolname = argv[0]; + + if ((zhp = zpool_open(g_zfs, poolname)) == NULL) + return (1); + + for (i = 1; i < argc; i++) { + if (zpool_vdev_offline(zhp, argv[i], istmp) != 0) + ret = 1; + } + + zpool_close(zhp); + + return (ret); +} + +/* + * zpool clear <pool> [device] + * + * Clear all errors associated with a pool or a particular device. + */ +int +zpool_do_clear(int argc, char **argv) +{ + int ret = 0; + zpool_handle_t *zhp; + char *pool, *device; + + if (argc < 2) { + (void) fprintf(stderr, gettext("missing pool name\n")); + usage(B_FALSE); + } + + if (argc > 3) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + pool = argv[1]; + device = argc == 3 ? argv[2] : NULL; + + if ((zhp = zpool_open_canfail(g_zfs, pool)) == NULL) + return (1); + + if (zpool_clear(zhp, device) != 0) + ret = 1; + + zpool_close(zhp); + + return (ret); +} + +typedef struct scrub_cbdata { + int cb_type; + int cb_argc; + char **cb_argv; +} scrub_cbdata_t; + +int +scrub_callback(zpool_handle_t *zhp, void *data) +{ + scrub_cbdata_t *cb = data; + int err; + + /* + * Ignore faulted pools. + */ + if (zpool_get_state(zhp) == POOL_STATE_UNAVAIL) { + (void) fprintf(stderr, gettext("cannot scrub '%s': pool is " + "currently unavailable\n"), zpool_get_name(zhp)); + return (1); + } + + err = zpool_scrub(zhp, cb->cb_type); + + return (err != 0); +} + +/* + * zpool scrub [-s] <pool> ... + * + * -s Stop. Stops any in-progress scrub. + */ +int +zpool_do_scrub(int argc, char **argv) +{ + int c; + scrub_cbdata_t cb; + + cb.cb_type = POOL_SCRUB_EVERYTHING; + + /* check options */ + while ((c = getopt(argc, argv, "s")) != -1) { + switch (c) { + case 's': + cb.cb_type = POOL_SCRUB_NONE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + cb.cb_argc = argc; + cb.cb_argv = argv; + argc -= optind; + argv += optind; + + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name argument\n")); + usage(B_FALSE); + } + + return (for_each_pool(argc, argv, B_TRUE, NULL, scrub_callback, &cb)); +} + +typedef struct status_cbdata { + int cb_count; + boolean_t cb_allpools; + boolean_t cb_verbose; + boolean_t cb_explain; + boolean_t cb_first; +} status_cbdata_t; + +/* + * Print out detailed scrub status. + */ +void +print_scrub_status(nvlist_t *nvroot) +{ + vdev_stat_t *vs; + uint_t vsc; + time_t start, end, now; + double fraction_done; + uint64_t examined, total, minutes_left, minutes_taken; + char *scrub_type; + + verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_STATS, + (uint64_t **)&vs, &vsc) == 0); + + /* + * If there's never been a scrub, there's not much to say. + */ + if (vs->vs_scrub_end == 0 && vs->vs_scrub_type == POOL_SCRUB_NONE) { + (void) printf(gettext("none requested\n")); + return; + } + + scrub_type = (vs->vs_scrub_type == POOL_SCRUB_RESILVER) ? + "resilver" : "scrub"; + + start = vs->vs_scrub_start; + end = vs->vs_scrub_end; + now = time(NULL); + examined = vs->vs_scrub_examined; + total = vs->vs_alloc; + + if (end != 0) { + minutes_taken = (uint64_t)((end - start) / 60); + + (void) printf(gettext("%s %s after %lluh%um with %llu errors " + "on %s"), + scrub_type, vs->vs_scrub_complete ? "completed" : "stopped", + (u_longlong_t)(minutes_taken / 60), + (uint_t)(minutes_taken % 60), + (u_longlong_t)vs->vs_scrub_errors, ctime(&end)); + return; + } + + if (examined == 0) + examined = 1; + if (examined > total) + total = examined; + + fraction_done = (double)examined / total; + minutes_left = (uint64_t)((now - start) * + (1 - fraction_done) / fraction_done / 60); + minutes_taken = (uint64_t)((now - start) / 60); + + (void) printf(gettext("%s in progress for %lluh%um, %.2f%% done, " + "%lluh%um to go\n"), + scrub_type, (u_longlong_t)(minutes_taken / 60), + (uint_t)(minutes_taken % 60), 100 * fraction_done, + (u_longlong_t)(minutes_left / 60), (uint_t)(minutes_left % 60)); +} + +typedef struct spare_cbdata { + uint64_t cb_guid; + zpool_handle_t *cb_zhp; +} spare_cbdata_t; + +static boolean_t +find_vdev(nvlist_t *nv, uint64_t search) +{ + uint64_t guid; + nvlist_t **child; + uint_t c, children; + + if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) == 0 && + search == guid) + return (B_TRUE); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) == 0) { + for (c = 0; c < children; c++) + if (find_vdev(child[c], search)) + return (B_TRUE); + } + + return (B_FALSE); +} + +static int +find_spare(zpool_handle_t *zhp, void *data) +{ + spare_cbdata_t *cbp = data; + nvlist_t *config, *nvroot; + + config = zpool_get_config(zhp, NULL); + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &nvroot) == 0); + + if (find_vdev(nvroot, cbp->cb_guid)) { + cbp->cb_zhp = zhp; + return (1); + } + + zpool_close(zhp); + return (0); +} + +/* + * Print out configuration state as requested by status_callback. + */ +void +print_status_config(zpool_handle_t *zhp, const char *name, nvlist_t *nv, + int namewidth, int depth, boolean_t isspare, boolean_t print_logs) +{ + nvlist_t **child; + uint_t c, children; + vdev_stat_t *vs; + char rbuf[6], wbuf[6], cbuf[6], repaired[7]; + char *vname; + uint64_t notpresent; + spare_cbdata_t cb; + char *state; + + verify(nvlist_lookup_uint64_array(nv, ZPOOL_CONFIG_STATS, + (uint64_t **)&vs, &c) == 0); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) + children = 0; + + state = zpool_state_to_name(vs->vs_state, vs->vs_aux); + if (isspare) { + /* + * For hot spares, we use the terms 'INUSE' and 'AVAILABLE' for + * online drives. + */ + if (vs->vs_aux == VDEV_AUX_SPARED) + state = "INUSE"; + else if (vs->vs_state == VDEV_STATE_HEALTHY) + state = "AVAIL"; + } + + (void) printf("\t%*s%-*s %-8s", depth, "", namewidth - depth, + name, state); + + if (!isspare) { + zfs_nicenum(vs->vs_read_errors, rbuf, sizeof (rbuf)); + zfs_nicenum(vs->vs_write_errors, wbuf, sizeof (wbuf)); + zfs_nicenum(vs->vs_checksum_errors, cbuf, sizeof (cbuf)); + (void) printf(" %5s %5s %5s", rbuf, wbuf, cbuf); + } + + if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NOT_PRESENT, + ¬present) == 0) { + char *path; + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) == 0); + (void) printf(" was %s", path); + } else if (vs->vs_aux != 0) { + (void) printf(" "); + + switch (vs->vs_aux) { + case VDEV_AUX_OPEN_FAILED: + (void) printf(gettext("cannot open")); + break; + + case VDEV_AUX_BAD_GUID_SUM: + (void) printf(gettext("missing device")); + break; + + case VDEV_AUX_NO_REPLICAS: + (void) printf(gettext("insufficient replicas")); + break; + + case VDEV_AUX_VERSION_NEWER: + (void) printf(gettext("newer version")); + break; + + case VDEV_AUX_SPARED: + verify(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, + &cb.cb_guid) == 0); + if (zpool_iter(g_zfs, find_spare, &cb) == 1) { + if (strcmp(zpool_get_name(cb.cb_zhp), + zpool_get_name(zhp)) == 0) + (void) printf(gettext("currently in " + "use")); + else + (void) printf(gettext("in use by " + "pool '%s'"), + zpool_get_name(cb.cb_zhp)); + zpool_close(cb.cb_zhp); + } else { + (void) printf(gettext("currently in use")); + } + break; + + case VDEV_AUX_ERR_EXCEEDED: + (void) printf(gettext("too many errors")); + break; + + case VDEV_AUX_IO_FAILURE: + (void) printf(gettext("experienced I/O failures")); + break; + + case VDEV_AUX_BAD_LOG: + (void) printf(gettext("bad intent log")); + break; + + default: + (void) printf(gettext("corrupted data")); + break; + } + } else if (vs->vs_scrub_repaired != 0 && children == 0) { + /* + * Report bytes resilvered/repaired on leaf devices. + */ + zfs_nicenum(vs->vs_scrub_repaired, repaired, sizeof (repaired)); + (void) printf(gettext(" %s %s"), repaired, + (vs->vs_scrub_type == POOL_SCRUB_RESILVER) ? + "resilvered" : "repaired"); + } + + (void) printf("\n"); + + for (c = 0; c < children; c++) { + uint64_t is_log = B_FALSE; + + (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG, + &is_log); + if ((is_log && !print_logs) || (!is_log && print_logs)) + continue; + vname = zpool_vdev_name(g_zfs, zhp, child[c]); + print_status_config(zhp, vname, child[c], + namewidth, depth + 2, isspare, B_FALSE); + free(vname); + } +} + +static void +print_error_log(zpool_handle_t *zhp) +{ + nvlist_t *nverrlist = NULL; + nvpair_t *elem; + char *pathname; + size_t len = MAXPATHLEN * 2; + + if (zpool_get_errlog(zhp, &nverrlist) != 0) { + (void) printf("errors: List of errors unavailable " + "(insufficient privileges)\n"); + return; + } + + (void) printf("errors: Permanent errors have been " + "detected in the following files:\n\n"); + + pathname = safe_malloc(len); + elem = NULL; + while ((elem = nvlist_next_nvpair(nverrlist, elem)) != NULL) { + nvlist_t *nv; + uint64_t dsobj, obj; + + verify(nvpair_value_nvlist(elem, &nv) == 0); + verify(nvlist_lookup_uint64(nv, ZPOOL_ERR_DATASET, + &dsobj) == 0); + verify(nvlist_lookup_uint64(nv, ZPOOL_ERR_OBJECT, + &obj) == 0); + zpool_obj_to_path(zhp, dsobj, obj, pathname, len); + (void) printf("%7s %s\n", "", pathname); + } + free(pathname); + nvlist_free(nverrlist); +} + +static void +print_spares(zpool_handle_t *zhp, nvlist_t **spares, uint_t nspares, + int namewidth) +{ + uint_t i; + char *name; + + if (nspares == 0) + return; + + (void) printf(gettext("\tspares\n")); + + for (i = 0; i < nspares; i++) { + name = zpool_vdev_name(g_zfs, zhp, spares[i]); + print_status_config(zhp, name, spares[i], + namewidth, 2, B_TRUE, B_FALSE); + free(name); + } +} + +static void +print_l2cache(zpool_handle_t *zhp, nvlist_t **l2cache, uint_t nl2cache, + int namewidth) +{ + uint_t i; + char *name; + + if (nl2cache == 0) + return; + + (void) printf(gettext("\tcache\n")); + + for (i = 0; i < nl2cache; i++) { + name = zpool_vdev_name(g_zfs, zhp, l2cache[i]); + print_status_config(zhp, name, l2cache[i], + namewidth, 2, B_FALSE, B_FALSE); + free(name); + } +} + +/* + * Display a summary of pool status. Displays a summary such as: + * + * pool: tank + * status: DEGRADED + * reason: One or more devices ... + * see: http://www.sun.com/msg/ZFS-xxxx-01 + * config: + * mirror DEGRADED + * c1t0d0 OK + * c2t0d0 UNAVAIL + * + * When given the '-v' option, we print out the complete config. If the '-e' + * option is specified, then we print out error rate information as well. + */ +int +status_callback(zpool_handle_t *zhp, void *data) +{ + status_cbdata_t *cbp = data; + nvlist_t *config, *nvroot; + char *msgid; + int reason; + const char *health; + uint_t c; + vdev_stat_t *vs; + + config = zpool_get_config(zhp, NULL); + reason = zpool_get_status(zhp, &msgid); + + cbp->cb_count++; + + /* + * If we were given 'zpool status -x', only report those pools with + * problems. + */ + if (reason == ZPOOL_STATUS_OK && cbp->cb_explain) { + if (!cbp->cb_allpools) { + (void) printf(gettext("pool '%s' is healthy\n"), + zpool_get_name(zhp)); + if (cbp->cb_first) + cbp->cb_first = B_FALSE; + } + return (0); + } + + if (cbp->cb_first) + cbp->cb_first = B_FALSE; + else + (void) printf("\n"); + + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &nvroot) == 0); + verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_STATS, + (uint64_t **)&vs, &c) == 0); + health = zpool_state_to_name(vs->vs_state, vs->vs_aux); + + (void) printf(gettext(" pool: %s\n"), zpool_get_name(zhp)); + (void) printf(gettext(" state: %s\n"), health); + + switch (reason) { + case ZPOOL_STATUS_MISSING_DEV_R: + (void) printf(gettext("status: One or more devices could not " + "be opened. Sufficient replicas exist for\n\tthe pool to " + "continue functioning in a degraded state.\n")); + (void) printf(gettext("action: Attach the missing device and " + "online it using 'zpool online'.\n")); + break; + + case ZPOOL_STATUS_MISSING_DEV_NR: + (void) printf(gettext("status: One or more devices could not " + "be opened. There are insufficient\n\treplicas for the " + "pool to continue functioning.\n")); + (void) printf(gettext("action: Attach the missing device and " + "online it using 'zpool online'.\n")); + break; + + case ZPOOL_STATUS_CORRUPT_LABEL_R: + (void) printf(gettext("status: One or more devices could not " + "be used because the label is missing or\n\tinvalid. " + "Sufficient replicas exist for the pool to continue\n\t" + "functioning in a degraded state.\n")); + (void) printf(gettext("action: Replace the device using " + "'zpool replace'.\n")); + break; + + case ZPOOL_STATUS_CORRUPT_LABEL_NR: + (void) printf(gettext("status: One or more devices could not " + "be used because the label is missing \n\tor invalid. " + "There are insufficient replicas for the pool to " + "continue\n\tfunctioning.\n")); + (void) printf(gettext("action: Destroy and re-create the pool " + "from a backup source.\n")); + break; + + case ZPOOL_STATUS_FAILING_DEV: + (void) printf(gettext("status: One or more devices has " + "experienced an unrecoverable error. An\n\tattempt was " + "made to correct the error. Applications are " + "unaffected.\n")); + (void) printf(gettext("action: Determine if the device needs " + "to be replaced, and clear the errors\n\tusing " + "'zpool clear' or replace the device with 'zpool " + "replace'.\n")); + break; + + case ZPOOL_STATUS_OFFLINE_DEV: + (void) printf(gettext("status: One or more devices has " + "been taken offline by the administrator.\n\tSufficient " + "replicas exist for the pool to continue functioning in " + "a\n\tdegraded state.\n")); + (void) printf(gettext("action: Online the device using " + "'zpool online' or replace the device with\n\t'zpool " + "replace'.\n")); + break; + + case ZPOOL_STATUS_RESILVERING: + (void) printf(gettext("status: One or more devices is " + "currently being resilvered. The pool will\n\tcontinue " + "to function, possibly in a degraded state.\n")); + (void) printf(gettext("action: Wait for the resilver to " + "complete.\n")); + break; + + case ZPOOL_STATUS_CORRUPT_DATA: + (void) printf(gettext("status: One or more devices has " + "experienced an error resulting in data\n\tcorruption. " + "Applications may be affected.\n")); + (void) printf(gettext("action: Restore the file in question " + "if possible. Otherwise restore the\n\tentire pool from " + "backup.\n")); + break; + + case ZPOOL_STATUS_CORRUPT_POOL: + (void) printf(gettext("status: The pool metadata is corrupted " + "and the pool cannot be opened.\n")); + (void) printf(gettext("action: Destroy and re-create the pool " + "from a backup source.\n")); + break; + + case ZPOOL_STATUS_VERSION_OLDER: + (void) printf(gettext("status: The pool is formatted using an " + "older on-disk format. The pool can\n\tstill be used, but " + "some features are unavailable.\n")); + (void) printf(gettext("action: Upgrade the pool using 'zpool " + "upgrade'. Once this is done, the\n\tpool will no longer " + "be accessible on older software versions.\n")); + break; + + case ZPOOL_STATUS_VERSION_NEWER: + (void) printf(gettext("status: The pool has been upgraded to a " + "newer, incompatible on-disk version.\n\tThe pool cannot " + "be accessed on this system.\n")); + (void) printf(gettext("action: Access the pool from a system " + "running more recent software, or\n\trestore the pool from " + "backup.\n")); + break; + + case ZPOOL_STATUS_FAULTED_DEV_R: + (void) printf(gettext("status: One or more devices are " + "faulted in response to persistent errors.\n\tSufficient " + "replicas exist for the pool to continue functioning " + "in a\n\tdegraded state.\n")); + (void) printf(gettext("action: Replace the faulted device, " + "or use 'zpool clear' to mark the device\n\trepaired.\n")); + break; + + case ZPOOL_STATUS_FAULTED_DEV_NR: + (void) printf(gettext("status: One or more devices are " + "faulted in response to persistent errors. There are " + "insufficient replicas for the pool to\n\tcontinue " + "functioning.\n")); + (void) printf(gettext("action: Destroy and re-create the pool " + "from a backup source. Manually marking the device\n" + "\trepaired using 'zpool clear' may allow some data " + "to be recovered.\n")); + break; + + case ZPOOL_STATUS_IO_FAILURE_WAIT: + case ZPOOL_STATUS_IO_FAILURE_CONTINUE: + (void) printf(gettext("status: One or more devices are " + "faulted in response to IO failures.\n")); + (void) printf(gettext("action: Make sure the affected devices " + "are connected, then run 'zpool clear'.\n")); + break; + + case ZPOOL_STATUS_BAD_LOG: + (void) printf(gettext("status: An intent log record " + "could not be read.\n" + "\tWaiting for adminstrator intervention to fix the " + "faulted pool.\n")); + (void) printf(gettext("action: Either restore the affected " + "device(s) and run 'zpool online',\n" + "\tor ignore the intent log records by running " + "'zpool clear'.\n")); + break; + + default: + /* + * The remaining errors can't actually be generated, yet. + */ + assert(reason == ZPOOL_STATUS_OK); + } + + if (msgid != NULL) + (void) printf(gettext(" see: http://www.sun.com/msg/%s\n"), + msgid); + + if (config != NULL) { + int namewidth; + uint64_t nerr; + nvlist_t **spares, **l2cache; + uint_t nspares, nl2cache; + + + (void) printf(gettext(" scrub: ")); + print_scrub_status(nvroot); + + namewidth = max_width(zhp, nvroot, 0, 0); + if (namewidth < 10) + namewidth = 10; + + (void) printf(gettext("config:\n\n")); + (void) printf(gettext("\t%-*s %-8s %5s %5s %5s\n"), namewidth, + "NAME", "STATE", "READ", "WRITE", "CKSUM"); + print_status_config(zhp, zpool_get_name(zhp), nvroot, + namewidth, 0, B_FALSE, B_FALSE); + if (num_logs(nvroot) > 0) + print_status_config(zhp, "logs", nvroot, namewidth, 0, + B_FALSE, B_TRUE); + + if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, + &l2cache, &nl2cache) == 0) + print_l2cache(zhp, l2cache, nl2cache, namewidth); + + if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, + &spares, &nspares) == 0) + print_spares(zhp, spares, nspares, namewidth); + + if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_ERRCOUNT, + &nerr) == 0) { + nvlist_t *nverrlist = NULL; + + /* + * If the approximate error count is small, get a + * precise count by fetching the entire log and + * uniquifying the results. + */ + if (nerr > 0 && nerr < 100 && !cbp->cb_verbose && + zpool_get_errlog(zhp, &nverrlist) == 0) { + nvpair_t *elem; + + elem = NULL; + nerr = 0; + while ((elem = nvlist_next_nvpair(nverrlist, + elem)) != NULL) { + nerr++; + } + } + nvlist_free(nverrlist); + + (void) printf("\n"); + + if (nerr == 0) + (void) printf(gettext("errors: No known data " + "errors\n")); + else if (!cbp->cb_verbose) + (void) printf(gettext("errors: %llu data " + "errors, use '-v' for a list\n"), + (u_longlong_t)nerr); + else + print_error_log(zhp); + } + } else { + (void) printf(gettext("config: The configuration cannot be " + "determined.\n")); + } + + return (0); +} + +/* + * zpool status [-vx] [pool] ... + * + * -v Display complete error logs + * -x Display only pools with potential problems + * + * Describes the health status of all pools or some subset. + */ +int +zpool_do_status(int argc, char **argv) +{ + int c; + int ret; + status_cbdata_t cb = { 0 }; + + /* check options */ + while ((c = getopt(argc, argv, "vx")) != -1) { + switch (c) { + case 'v': + cb.cb_verbose = B_TRUE; + break; + case 'x': + cb.cb_explain = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + cb.cb_first = B_TRUE; + + if (argc == 0) + cb.cb_allpools = B_TRUE; + + ret = for_each_pool(argc, argv, B_TRUE, NULL, status_callback, &cb); + + if (argc == 0 && cb.cb_count == 0) + (void) printf(gettext("no pools available\n")); + else if (cb.cb_explain && cb.cb_first && cb.cb_allpools) + (void) printf(gettext("all pools are healthy\n")); + + return (ret); +} + +typedef struct upgrade_cbdata { + int cb_all; + int cb_first; + int cb_newer; + int cb_argc; + uint64_t cb_version; + char **cb_argv; +} upgrade_cbdata_t; + +static int +upgrade_cb(zpool_handle_t *zhp, void *arg) +{ + upgrade_cbdata_t *cbp = arg; + nvlist_t *config; + uint64_t version; + int ret = 0; + + config = zpool_get_config(zhp, NULL); + verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION, + &version) == 0); + + if (!cbp->cb_newer && version < SPA_VERSION) { + if (!cbp->cb_all) { + if (cbp->cb_first) { + (void) printf(gettext("The following pools are " + "out of date, and can be upgraded. After " + "being\nupgraded, these pools will no " + "longer be accessible by older software " + "versions.\n\n")); + (void) printf(gettext("VER POOL\n")); + (void) printf(gettext("--- ------------\n")); + cbp->cb_first = B_FALSE; + } + + (void) printf("%2llu %s\n", (u_longlong_t)version, + zpool_get_name(zhp)); + } else { + cbp->cb_first = B_FALSE; + ret = zpool_upgrade(zhp, cbp->cb_version); + if (!ret) { + (void) printf(gettext("Successfully upgraded " + "'%s'\n\n"), zpool_get_name(zhp)); + } + } + } else if (cbp->cb_newer && version > SPA_VERSION) { + assert(!cbp->cb_all); + + if (cbp->cb_first) { + (void) printf(gettext("The following pools are " + "formatted using a newer software version and\n" + "cannot be accessed on the current system.\n\n")); + (void) printf(gettext("VER POOL\n")); + (void) printf(gettext("--- ------------\n")); + cbp->cb_first = B_FALSE; + } + + (void) printf("%2llu %s\n", (u_longlong_t)version, + zpool_get_name(zhp)); + } + + zpool_close(zhp); + return (ret); +} + +/* ARGSUSED */ +static int +upgrade_one(zpool_handle_t *zhp, void *data) +{ + upgrade_cbdata_t *cbp = data; + uint64_t cur_version; + int ret; + + if (strcmp("log", zpool_get_name(zhp)) == 0) { + (void) printf(gettext("'log' is now a reserved word\n" + "Pool 'log' must be renamed using export and import" + " to upgrade.\n")); + return (1); + } + + cur_version = zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL); + if (cur_version > cbp->cb_version) { + (void) printf(gettext("Pool '%s' is already formatted " + "using more current version '%llu'.\n"), + zpool_get_name(zhp), cur_version); + return (0); + } + if (cur_version == cbp->cb_version) { + (void) printf(gettext("Pool '%s' is already formatted " + "using the current version.\n"), zpool_get_name(zhp)); + return (0); + } + + ret = zpool_upgrade(zhp, cbp->cb_version); + + if (!ret) { + (void) printf(gettext("Successfully upgraded '%s' " + "from version %llu to version %llu\n\n"), + zpool_get_name(zhp), (u_longlong_t)cur_version, + (u_longlong_t)cbp->cb_version); + } + + return (ret != 0); +} + +/* + * zpool upgrade + * zpool upgrade -v + * zpool upgrade [-V version] <-a | pool ...> + * + * With no arguments, display downrev'd ZFS pool available for upgrade. + * Individual pools can be upgraded by specifying the pool, and '-a' will + * upgrade all pools. + */ +int +zpool_do_upgrade(int argc, char **argv) +{ + int c; + upgrade_cbdata_t cb = { 0 }; + int ret = 0; + boolean_t showversions = B_FALSE; + char *end; + + + /* check options */ + while ((c = getopt(argc, argv, "avV:")) != -1) { + switch (c) { + case 'a': + cb.cb_all = B_TRUE; + break; + case 'v': + showversions = B_TRUE; + break; + case 'V': + cb.cb_version = strtoll(optarg, &end, 10); + if (*end != '\0' || cb.cb_version > SPA_VERSION || + cb.cb_version < SPA_VERSION_1) { + (void) fprintf(stderr, + gettext("invalid version '%s'\n"), optarg); + usage(B_FALSE); + } + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + cb.cb_argc = argc; + cb.cb_argv = argv; + argc -= optind; + argv += optind; + + if (cb.cb_version == 0) { + cb.cb_version = SPA_VERSION; + } else if (!cb.cb_all && argc == 0) { + (void) fprintf(stderr, gettext("-V option is " + "incompatible with other arguments\n")); + usage(B_FALSE); + } + + if (showversions) { + if (cb.cb_all || argc != 0) { + (void) fprintf(stderr, gettext("-v option is " + "incompatible with other arguments\n")); + usage(B_FALSE); + } + } else if (cb.cb_all) { + if (argc != 0) { + (void) fprintf(stderr, gettext("-a option should not " + "be used along with a pool name\n")); + usage(B_FALSE); + } + } + + (void) printf(gettext("This system is currently running " + "ZFS pool version %llu.\n\n"), SPA_VERSION); + cb.cb_first = B_TRUE; + if (showversions) { + (void) printf(gettext("The following versions are " + "supported:\n\n")); + (void) printf(gettext("VER DESCRIPTION\n")); + (void) printf("--- -----------------------------------------" + "---------------\n"); + (void) printf(gettext(" 1 Initial ZFS version\n")); + (void) printf(gettext(" 2 Ditto blocks " + "(replicated metadata)\n")); + (void) printf(gettext(" 3 Hot spares and double parity " + "RAID-Z\n")); + (void) printf(gettext(" 4 zpool history\n")); + (void) printf(gettext(" 5 Compression using the gzip " + "algorithm\n")); + (void) printf(gettext(" 6 bootfs pool property\n")); + (void) printf(gettext(" 7 Separate intent log devices\n")); + (void) printf(gettext(" 8 Delegated administration\n")); + (void) printf(gettext(" 9 refquota and refreservation " + "properties\n")); + (void) printf(gettext(" 10 Cache devices\n")); + (void) printf(gettext(" 11 Improved scrub performance\n")); + (void) printf(gettext(" 12 Snapshot properties\n")); + (void) printf(gettext(" 13 snapused property\n")); + (void) printf(gettext(" 14 passthrough-x aclinherit " + "support\n")); + (void) printf(gettext("For more information on a particular " + "version, including supported releases, see:\n\n")); + (void) printf("http://www.opensolaris.org/os/community/zfs/" + "version/N\n\n"); + (void) printf(gettext("Where 'N' is the version number.\n")); + } else if (argc == 0) { + int notfound; + + ret = zpool_iter(g_zfs, upgrade_cb, &cb); + notfound = cb.cb_first; + + if (!cb.cb_all && ret == 0) { + if (!cb.cb_first) + (void) printf("\n"); + cb.cb_first = B_TRUE; + cb.cb_newer = B_TRUE; + ret = zpool_iter(g_zfs, upgrade_cb, &cb); + if (!cb.cb_first) { + notfound = B_FALSE; + (void) printf("\n"); + } + } + + if (ret == 0) { + if (notfound) + (void) printf(gettext("All pools are formatted " + "using this version.\n")); + else if (!cb.cb_all) + (void) printf(gettext("Use 'zpool upgrade -v' " + "for a list of available versions and " + "their associated\nfeatures.\n")); + } + } else { + ret = for_each_pool(argc, argv, B_FALSE, NULL, + upgrade_one, &cb); + } + + return (ret); +} + +typedef struct hist_cbdata { + boolean_t first; + int longfmt; + int internal; +} hist_cbdata_t; + +char *hist_event_table[LOG_END] = { + "invalid event", + "pool create", + "vdev add", + "pool remove", + "pool destroy", + "pool export", + "pool import", + "vdev attach", + "vdev replace", + "vdev detach", + "vdev online", + "vdev offline", + "vdev upgrade", + "pool clear", + "pool scrub", + "pool property set", + "create", + "clone", + "destroy", + "destroy_begin_sync", + "inherit", + "property set", + "quota set", + "permission update", + "permission remove", + "permission who remove", + "promote", + "receive", + "rename", + "reservation set", + "replay_inc_sync", + "replay_full_sync", + "rollback", + "snapshot", + "filesystem version upgrade", + "refquota set", + "refreservation set", + "pool scrub done", +}; + +/* + * Print out the command history for a specific pool. + */ +static int +get_history_one(zpool_handle_t *zhp, void *data) +{ + nvlist_t *nvhis; + nvlist_t **records; + uint_t numrecords; + char *cmdstr; + char *pathstr; + uint64_t dst_time; + time_t tsec; + struct tm t; + char tbuf[30]; + int ret, i; + uint64_t who; + struct passwd *pwd; + char *hostname; + char *zonename; + char internalstr[MAXPATHLEN]; + hist_cbdata_t *cb = (hist_cbdata_t *)data; + uint64_t txg; + uint64_t ievent; + + cb->first = B_FALSE; + + (void) printf(gettext("History for '%s':\n"), zpool_get_name(zhp)); + + if ((ret = zpool_get_history(zhp, &nvhis)) != 0) + return (ret); + + verify(nvlist_lookup_nvlist_array(nvhis, ZPOOL_HIST_RECORD, + &records, &numrecords) == 0); + for (i = 0; i < numrecords; i++) { + if (nvlist_lookup_uint64(records[i], ZPOOL_HIST_TIME, + &dst_time) != 0) + continue; + + /* is it an internal event or a standard event? */ + if (nvlist_lookup_string(records[i], ZPOOL_HIST_CMD, + &cmdstr) != 0) { + if (cb->internal == 0) + continue; + + if (nvlist_lookup_uint64(records[i], + ZPOOL_HIST_INT_EVENT, &ievent) != 0) + continue; + verify(nvlist_lookup_uint64(records[i], + ZPOOL_HIST_TXG, &txg) == 0); + verify(nvlist_lookup_string(records[i], + ZPOOL_HIST_INT_STR, &pathstr) == 0); + if (ievent >= LOG_END) + continue; + (void) snprintf(internalstr, + sizeof (internalstr), + "[internal %s txg:%lld] %s", + hist_event_table[ievent], txg, + pathstr); + cmdstr = internalstr; + } + tsec = dst_time; + (void) localtime_r(&tsec, &t); + (void) strftime(tbuf, sizeof (tbuf), "%F.%T", &t); + (void) printf("%s %s", tbuf, cmdstr); + + if (!cb->longfmt) { + (void) printf("\n"); + continue; + } + (void) printf(" ["); + if (nvlist_lookup_uint64(records[i], + ZPOOL_HIST_WHO, &who) == 0) { + pwd = getpwuid((uid_t)who); + if (pwd) + (void) printf("user %s on", + pwd->pw_name); + else + (void) printf("user %d on", + (int)who); + } else { + (void) printf(gettext("no info]\n")); + continue; + } + if (nvlist_lookup_string(records[i], + ZPOOL_HIST_HOST, &hostname) == 0) { + (void) printf(" %s", hostname); + } + if (nvlist_lookup_string(records[i], + ZPOOL_HIST_ZONE, &zonename) == 0) { + (void) printf(":%s", zonename); + } + + (void) printf("]"); + (void) printf("\n"); + } + (void) printf("\n"); + nvlist_free(nvhis); + + return (ret); +} + +/* + * zpool history <pool> + * + * Displays the history of commands that modified pools. + */ + + +int +zpool_do_history(int argc, char **argv) +{ + hist_cbdata_t cbdata = { 0 }; + int ret; + int c; + + cbdata.first = B_TRUE; + /* check options */ + while ((c = getopt(argc, argv, "li")) != -1) { + switch (c) { + case 'l': + cbdata.longfmt = 1; + break; + case 'i': + cbdata.internal = 1; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + argc -= optind; + argv += optind; + + ret = for_each_pool(argc, argv, B_FALSE, NULL, get_history_one, + &cbdata); + + if (argc == 0 && cbdata.first == B_TRUE) { + (void) printf(gettext("no pools available\n")); + return (0); + } + + return (ret); +} + +static int +get_callback(zpool_handle_t *zhp, void *data) +{ + zprop_get_cbdata_t *cbp = (zprop_get_cbdata_t *)data; + char value[MAXNAMELEN]; + zprop_source_t srctype; + zprop_list_t *pl; + + for (pl = cbp->cb_proplist; pl != NULL; pl = pl->pl_next) { + + /* + * Skip the special fake placeholder. This will also skip + * over the name property when 'all' is specified. + */ + if (pl->pl_prop == ZPOOL_PROP_NAME && + pl == cbp->cb_proplist) + continue; + + if (zpool_get_prop(zhp, pl->pl_prop, + value, sizeof (value), &srctype) != 0) + continue; + + zprop_print_one_property(zpool_get_name(zhp), cbp, + zpool_prop_to_name(pl->pl_prop), value, srctype, NULL); + } + return (0); +} + +int +zpool_do_get(int argc, char **argv) +{ + zprop_get_cbdata_t cb = { 0 }; + zprop_list_t fake_name = { 0 }; + int ret; + + if (argc < 3) + usage(B_FALSE); + + cb.cb_first = B_TRUE; + cb.cb_sources = ZPROP_SRC_ALL; + cb.cb_columns[0] = GET_COL_NAME; + cb.cb_columns[1] = GET_COL_PROPERTY; + cb.cb_columns[2] = GET_COL_VALUE; + cb.cb_columns[3] = GET_COL_SOURCE; + cb.cb_type = ZFS_TYPE_POOL; + + if (zprop_get_list(g_zfs, argv[1], &cb.cb_proplist, + ZFS_TYPE_POOL) != 0) + usage(B_FALSE); + + if (cb.cb_proplist != NULL) { + fake_name.pl_prop = ZPOOL_PROP_NAME; + fake_name.pl_width = strlen(gettext("NAME")); + fake_name.pl_next = cb.cb_proplist; + cb.cb_proplist = &fake_name; + } + + ret = for_each_pool(argc - 2, argv + 2, B_TRUE, &cb.cb_proplist, + get_callback, &cb); + + if (cb.cb_proplist == &fake_name) + zprop_free_list(fake_name.pl_next); + else + zprop_free_list(cb.cb_proplist); + + return (ret); +} + +typedef struct set_cbdata { + char *cb_propname; + char *cb_value; + boolean_t cb_any_successful; +} set_cbdata_t; + +int +set_callback(zpool_handle_t *zhp, void *data) +{ + int error; + set_cbdata_t *cb = (set_cbdata_t *)data; + + error = zpool_set_prop(zhp, cb->cb_propname, cb->cb_value); + + if (!error) + cb->cb_any_successful = B_TRUE; + + return (error); +} + +int +zpool_do_set(int argc, char **argv) +{ + set_cbdata_t cb = { 0 }; + int error; + + if (argc > 1 && argv[1][0] == '-') { + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + argv[1][1]); + usage(B_FALSE); + } + + if (argc < 2) { + (void) fprintf(stderr, gettext("missing property=value " + "argument\n")); + usage(B_FALSE); + } + + if (argc < 3) { + (void) fprintf(stderr, gettext("missing pool name\n")); + usage(B_FALSE); + } + + if (argc > 3) { + (void) fprintf(stderr, gettext("too many pool names\n")); + usage(B_FALSE); + } + + cb.cb_propname = argv[1]; + cb.cb_value = strchr(cb.cb_propname, '='); + if (cb.cb_value == NULL) { + (void) fprintf(stderr, gettext("missing value in " + "property=value argument\n")); + usage(B_FALSE); + } + + *(cb.cb_value) = '\0'; + cb.cb_value++; + + error = for_each_pool(argc - 2, argv + 2, B_TRUE, NULL, + set_callback, &cb); + + return (error); +} + +static int +find_command_idx(char *command, int *idx) +{ + int i; + + for (i = 0; i < NCOMMAND; i++) { + if (command_table[i].name == NULL) + continue; + + if (strcmp(command, command_table[i].name) == 0) { + *idx = i; + return (0); + } + } + return (1); +} + +int +main(int argc, char **argv) +{ + int ret; + int i; + char *cmdname; + + (void) setlocale(LC_ALL, ""); + (void) textdomain(TEXT_DOMAIN); + + if ((g_zfs = libzfs_init()) == NULL) { + (void) fprintf(stderr, gettext("internal error: failed to " + "initialize ZFS library\n")); + return (1); + } + + libzfs_print_on_error(g_zfs, B_TRUE); + + opterr = 0; + + /* + * Make sure the user has specified some command. + */ + if (argc < 2) { + (void) fprintf(stderr, gettext("missing command\n")); + usage(B_FALSE); + } + + cmdname = argv[1]; + + /* + * Special case '-?' + */ + if (strcmp(cmdname, "-?") == 0) + usage(B_TRUE); + + zpool_set_history_str("zpool", argc, argv, history_str); + verify(zpool_stage_history(g_zfs, history_str) == 0); + + /* + * Run the appropriate command. + */ + if (find_command_idx(cmdname, &i) == 0) { + current_command = &command_table[i]; + ret = command_table[i].func(argc - 1, argv + 1); + } else if (strchr(cmdname, '=')) { + verify(find_command_idx("set", &i) == 0); + current_command = &command_table[i]; + ret = command_table[i].func(argc, argv); + } else if (strcmp(cmdname, "freeze") == 0 && argc == 3) { + /* + * 'freeze' is a vile debugging abomination, so we treat + * it as such. + */ + char buf[16384]; + int fd = open(ZFS_DEV, O_RDWR); + (void) strcpy((void *)buf, argv[2]); + return (!!ioctl(fd, ZFS_IOC_POOL_FREEZE, buf)); + } else { + (void) fprintf(stderr, gettext("unrecognized " + "command '%s'\n"), cmdname); + usage(B_FALSE); + } + + libzfs_fini(g_zfs); + + /* + * The 'ZFS_ABORT' environment variable causes us to dump core on exit + * for the purposes of running ::findleaks. + */ + if (getenv("ZFS_ABORT") != NULL) { + (void) printf("dumping core by request\n"); + abort(); + } + + return (ret); +} diff --git a/cmd/zpool/zpool_util.c b/cmd/zpool/zpool_util.c new file mode 100644 index 000000000..f44da4ff6 --- /dev/null +++ b/cmd/zpool/zpool_util.c @@ -0,0 +1,104 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ +/* + * Copyright 2007 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#pragma ident "%Z%%M% %I% %E% SMI" + +#include <errno.h> +#include <libgen.h> +#include <libintl.h> +#include <stdio.h> +#include <stdlib.h> +#include <strings.h> + +#include "zpool_util.h" + +/* + * Utility function to guarantee malloc() success. + */ +void * +safe_malloc(size_t size) +{ + void *data; + + if ((data = calloc(1, size)) == NULL) { + (void) fprintf(stderr, "internal error: out of memory\n"); + exit(1); + } + + return (data); +} + +/* + * Same as above, but for strdup() + */ +char * +safe_strdup(const char *str) +{ + char *ret; + + if ((ret = strdup(str)) == NULL) { + (void) fprintf(stderr, "internal error: out of memory\n"); + exit(1); + } + + return (ret); +} + +/* + * Display an out of memory error message and abort the current program. + */ +void +zpool_no_memory(void) +{ + assert(errno == ENOMEM); + (void) fprintf(stderr, + gettext("internal error: out of memory\n")); + exit(1); +} + +/* + * Return the number of logs in supplied nvlist + */ +uint_t +num_logs(nvlist_t *nv) +{ + uint_t nlogs = 0; + uint_t c, children; + nvlist_t **child; + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) + return (0); + + for (c = 0; c < children; c++) { + uint64_t is_log = B_FALSE; + + (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG, + &is_log); + if (is_log) + nlogs++; + } + return (nlogs); +} diff --git a/cmd/zpool/zpool_util.h b/cmd/zpool/zpool_util.h new file mode 100644 index 000000000..e82f3202a --- /dev/null +++ b/cmd/zpool/zpool_util.h @@ -0,0 +1,72 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ +/* + * Copyright 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#ifndef ZPOOL_UTIL_H +#define ZPOOL_UTIL_H + +#include <libnvpair.h> +#include <libzfs.h> + +#ifdef __cplusplus +extern "C" { +#endif + +/* + * Basic utility functions + */ +void *safe_malloc(size_t); +char *safe_strdup(const char *); +void zpool_no_memory(void); +uint_t num_logs(nvlist_t *nv); + +/* + * Virtual device functions + */ + +nvlist_t *make_root_vdev(zpool_handle_t *zhp, int force, int check_rep, + boolean_t isreplace, boolean_t dryrun, int argc, char **argv); + +/* + * Pool list functions + */ +int for_each_pool(int, char **, boolean_t unavail, zprop_list_t **, + zpool_iter_f, void *); + +typedef struct zpool_list zpool_list_t; + +zpool_list_t *pool_list_get(int, char **, zprop_list_t **, int *); +void pool_list_update(zpool_list_t *); +int pool_list_iter(zpool_list_t *, int unavail, zpool_iter_f, void *); +void pool_list_free(zpool_list_t *); +int pool_list_count(zpool_list_t *); +void pool_list_remove(zpool_list_t *, zpool_handle_t *); + +libzfs_handle_t *g_zfs; + +#ifdef __cplusplus +} +#endif + +#endif /* ZPOOL_UTIL_H */ diff --git a/cmd/zpool/zpool_vdev.c b/cmd/zpool/zpool_vdev.c new file mode 100644 index 000000000..10007c149 --- /dev/null +++ b/cmd/zpool/zpool_vdev.c @@ -0,0 +1,1401 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ + +/* + * Copyright 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +/* + * Functions to convert between a list of vdevs and an nvlist representing the + * configuration. Each entry in the list can be one of: + * + * Device vdevs + * disk=(path=..., devid=...) + * file=(path=...) + * + * Group vdevs + * raidz[1|2]=(...) + * mirror=(...) + * + * Hot spares + * + * While the underlying implementation supports it, group vdevs cannot contain + * other group vdevs. All userland verification of devices is contained within + * this file. If successful, the nvlist returned can be passed directly to the + * kernel; we've done as much verification as possible in userland. + * + * Hot spares are a special case, and passed down as an array of disk vdevs, at + * the same level as the root of the vdev tree. + * + * The only function exported by this file is 'make_root_vdev'. The + * function performs several passes: + * + * 1. Construct the vdev specification. Performs syntax validation and + * makes sure each device is valid. + * 2. Check for devices in use. Using libdiskmgt, makes sure that no + * devices are also in use. Some can be overridden using the 'force' + * flag, others cannot. + * 3. Check for replication errors if the 'force' flag is not specified. + * validates that the replication level is consistent across the + * entire pool. + * 4. Call libzfs to label any whole disks with an EFI label. + */ + +#include <assert.h> +#include <devid.h> +#include <errno.h> +#include <fcntl.h> +#include <libdiskmgt.h> +#include <libintl.h> +#include <libnvpair.h> +#include <stdio.h> +#include <string.h> +#include <unistd.h> +#include <sys/efi_partition.h> +#include <sys/stat.h> +#include <sys/vtoc.h> +#include <sys/mntent.h> + +#include "zpool_util.h" + +#define DISK_ROOT "/dev/dsk" +#define RDISK_ROOT "/dev/rdsk" +#define BACKUP_SLICE "s2" + +/* + * For any given vdev specification, we can have multiple errors. The + * vdev_error() function keeps track of whether we have seen an error yet, and + * prints out a header if its the first error we've seen. + */ +boolean_t error_seen; +boolean_t is_force; + +/*PRINTFLIKE1*/ +static void +vdev_error(const char *fmt, ...) +{ + va_list ap; + + if (!error_seen) { + (void) fprintf(stderr, gettext("invalid vdev specification\n")); + if (!is_force) + (void) fprintf(stderr, gettext("use '-f' to override " + "the following errors:\n")); + else + (void) fprintf(stderr, gettext("the following errors " + "must be manually repaired:\n")); + error_seen = B_TRUE; + } + + va_start(ap, fmt); + (void) vfprintf(stderr, fmt, ap); + va_end(ap); +} + +static void +libdiskmgt_error(int error) +{ + /* + * ENXIO/ENODEV is a valid error message if the device doesn't live in + * /dev/dsk. Don't bother printing an error message in this case. + */ + if (error == ENXIO || error == ENODEV) + return; + + (void) fprintf(stderr, gettext("warning: device in use checking " + "failed: %s\n"), strerror(error)); +} + +/* + * Validate a device, passing the bulk of the work off to libdiskmgt. + */ +static int +check_slice(const char *path, int force, boolean_t wholedisk, boolean_t isspare) +{ + char *msg; + int error = 0; + dm_who_type_t who; + + if (force) + who = DM_WHO_ZPOOL_FORCE; + else if (isspare) + who = DM_WHO_ZPOOL_SPARE; + else + who = DM_WHO_ZPOOL; + + if (dm_inuse((char *)path, &msg, who, &error) || error) { + if (error != 0) { + libdiskmgt_error(error); + return (0); + } else { + vdev_error("%s", msg); + free(msg); + return (-1); + } + } + + /* + * If we're given a whole disk, ignore overlapping slices since we're + * about to label it anyway. + */ + error = 0; + if (!wholedisk && !force && + (dm_isoverlapping((char *)path, &msg, &error) || error)) { + if (error == 0) { + /* dm_isoverlapping returned -1 */ + vdev_error(gettext("%s overlaps with %s\n"), path, msg); + free(msg); + return (-1); + } else if (error != ENODEV) { + /* libdiskmgt's devcache only handles physical drives */ + libdiskmgt_error(error); + return (0); + } + } + + return (0); +} + + +/* + * Validate a whole disk. Iterate over all slices on the disk and make sure + * that none is in use by calling check_slice(). + */ +static int +check_disk(const char *name, dm_descriptor_t disk, int force, int isspare) +{ + dm_descriptor_t *drive, *media, *slice; + int err = 0; + int i; + int ret; + + /* + * Get the drive associated with this disk. This should never fail, + * because we already have an alias handle open for the device. + */ + if ((drive = dm_get_associated_descriptors(disk, DM_DRIVE, + &err)) == NULL || *drive == NULL) { + if (err) + libdiskmgt_error(err); + return (0); + } + + if ((media = dm_get_associated_descriptors(*drive, DM_MEDIA, + &err)) == NULL) { + dm_free_descriptors(drive); + if (err) + libdiskmgt_error(err); + return (0); + } + + dm_free_descriptors(drive); + + /* + * It is possible that the user has specified a removable media drive, + * and the media is not present. + */ + if (*media == NULL) { + dm_free_descriptors(media); + vdev_error(gettext("'%s' has no media in drive\n"), name); + return (-1); + } + + if ((slice = dm_get_associated_descriptors(*media, DM_SLICE, + &err)) == NULL) { + dm_free_descriptors(media); + if (err) + libdiskmgt_error(err); + return (0); + } + + dm_free_descriptors(media); + + ret = 0; + + /* + * Iterate over all slices and report any errors. We don't care about + * overlapping slices because we are using the whole disk. + */ + for (i = 0; slice[i] != NULL; i++) { + char *name = dm_get_name(slice[i], &err); + + if (check_slice(name, force, B_TRUE, isspare) != 0) + ret = -1; + + dm_free_name(name); + } + + dm_free_descriptors(slice); + return (ret); +} + +/* + * Validate a device. + */ +static int +check_device(const char *path, boolean_t force, boolean_t isspare) +{ + dm_descriptor_t desc; + int err; + char *dev; + + /* + * For whole disks, libdiskmgt does not include the leading dev path. + */ + dev = strrchr(path, '/'); + assert(dev != NULL); + dev++; + if ((desc = dm_get_descriptor_by_name(DM_ALIAS, dev, &err)) != NULL) { + err = check_disk(path, desc, force, isspare); + dm_free_descriptor(desc); + return (err); + } + + return (check_slice(path, force, B_FALSE, isspare)); +} + +/* + * Check that a file is valid. All we can do in this case is check that it's + * not in use by another pool, and not in use by swap. + */ +static int +check_file(const char *file, boolean_t force, boolean_t isspare) +{ + char *name; + int fd; + int ret = 0; + int err; + pool_state_t state; + boolean_t inuse; + + if (dm_inuse_swap(file, &err)) { + if (err) + libdiskmgt_error(err); + else + vdev_error(gettext("%s is currently used by swap. " + "Please see swap(1M).\n"), file); + return (-1); + } + + if ((fd = open(file, O_RDONLY)) < 0) + return (0); + + if (zpool_in_use(g_zfs, fd, &state, &name, &inuse) == 0 && inuse) { + const char *desc; + + switch (state) { + case POOL_STATE_ACTIVE: + desc = gettext("active"); + break; + + case POOL_STATE_EXPORTED: + desc = gettext("exported"); + break; + + case POOL_STATE_POTENTIALLY_ACTIVE: + desc = gettext("potentially active"); + break; + + default: + desc = gettext("unknown"); + break; + } + + /* + * Allow hot spares to be shared between pools. + */ + if (state == POOL_STATE_SPARE && isspare) + return (0); + + if (state == POOL_STATE_ACTIVE || + state == POOL_STATE_SPARE || !force) { + switch (state) { + case POOL_STATE_SPARE: + vdev_error(gettext("%s is reserved as a hot " + "spare for pool %s\n"), file, name); + break; + default: + vdev_error(gettext("%s is part of %s pool " + "'%s'\n"), file, desc, name); + break; + } + ret = -1; + } + + free(name); + } + + (void) close(fd); + return (ret); +} + + +/* + * By "whole disk" we mean an entire physical disk (something we can + * label, toggle the write cache on, etc.) as opposed to the full + * capacity of a pseudo-device such as lofi or did. We act as if we + * are labeling the disk, which should be a pretty good test of whether + * it's a viable device or not. Returns B_TRUE if it is and B_FALSE if + * it isn't. + */ +static boolean_t +is_whole_disk(const char *arg) +{ + struct dk_gpt *label; + int fd; + char path[MAXPATHLEN]; + + (void) snprintf(path, sizeof (path), "%s%s%s", + RDISK_ROOT, strrchr(arg, '/'), BACKUP_SLICE); + if ((fd = open(path, O_RDWR | O_NDELAY)) < 0) + return (B_FALSE); + if (efi_alloc_and_init(fd, EFI_NUMPAR, &label) != 0) { + (void) close(fd); + return (B_FALSE); + } + efi_free(label); + (void) close(fd); + return (B_TRUE); +} + +/* + * Create a leaf vdev. Determine if this is a file or a device. If it's a + * device, fill in the device id to make a complete nvlist. Valid forms for a + * leaf vdev are: + * + * /dev/dsk/xxx Complete disk path + * /xxx Full path to file + * xxx Shorthand for /dev/dsk/xxx + */ +static nvlist_t * +make_leaf_vdev(const char *arg, uint64_t is_log) +{ + char path[MAXPATHLEN]; + struct stat64 statbuf; + nvlist_t *vdev = NULL; + char *type = NULL; + boolean_t wholedisk = B_FALSE; + + /* + * Determine what type of vdev this is, and put the full path into + * 'path'. We detect whether this is a device of file afterwards by + * checking the st_mode of the file. + */ + if (arg[0] == '/') { + /* + * Complete device or file path. Exact type is determined by + * examining the file descriptor afterwards. + */ + wholedisk = is_whole_disk(arg); + if (!wholedisk && (stat64(arg, &statbuf) != 0)) { + (void) fprintf(stderr, + gettext("cannot open '%s': %s\n"), + arg, strerror(errno)); + return (NULL); + } + + (void) strlcpy(path, arg, sizeof (path)); + } else { + /* + * This may be a short path for a device, or it could be total + * gibberish. Check to see if it's a known device in + * /dev/dsk/. As part of this check, see if we've been given a + * an entire disk (minus the slice number). + */ + (void) snprintf(path, sizeof (path), "%s/%s", DISK_ROOT, + arg); + wholedisk = is_whole_disk(path); + if (!wholedisk && (stat64(path, &statbuf) != 0)) { + /* + * If we got ENOENT, then the user gave us + * gibberish, so try to direct them with a + * reasonable error message. Otherwise, + * regurgitate strerror() since it's the best we + * can do. + */ + if (errno == ENOENT) { + (void) fprintf(stderr, + gettext("cannot open '%s': no such " + "device in %s\n"), arg, DISK_ROOT); + (void) fprintf(stderr, + gettext("must be a full path or " + "shorthand device name\n")); + return (NULL); + } else { + (void) fprintf(stderr, + gettext("cannot open '%s': %s\n"), + path, strerror(errno)); + return (NULL); + } + } + } + + /* + * Determine whether this is a device or a file. + */ + if (wholedisk || S_ISBLK(statbuf.st_mode)) { + type = VDEV_TYPE_DISK; + } else if (S_ISREG(statbuf.st_mode)) { + type = VDEV_TYPE_FILE; + } else { + (void) fprintf(stderr, gettext("cannot use '%s': must be a " + "block device or regular file\n"), path); + return (NULL); + } + + /* + * Finally, we have the complete device or file, and we know that it is + * acceptable to use. Construct the nvlist to describe this vdev. All + * vdevs have a 'path' element, and devices also have a 'devid' element. + */ + verify(nvlist_alloc(&vdev, NV_UNIQUE_NAME, 0) == 0); + verify(nvlist_add_string(vdev, ZPOOL_CONFIG_PATH, path) == 0); + verify(nvlist_add_string(vdev, ZPOOL_CONFIG_TYPE, type) == 0); + verify(nvlist_add_uint64(vdev, ZPOOL_CONFIG_IS_LOG, is_log) == 0); + if (strcmp(type, VDEV_TYPE_DISK) == 0) + verify(nvlist_add_uint64(vdev, ZPOOL_CONFIG_WHOLE_DISK, + (uint64_t)wholedisk) == 0); + + /* + * For a whole disk, defer getting its devid until after labeling it. + */ + if (S_ISBLK(statbuf.st_mode) && !wholedisk) { + /* + * Get the devid for the device. + */ + int fd; + ddi_devid_t devid; + char *minor = NULL, *devid_str = NULL; + + if ((fd = open(path, O_RDONLY)) < 0) { + (void) fprintf(stderr, gettext("cannot open '%s': " + "%s\n"), path, strerror(errno)); + nvlist_free(vdev); + return (NULL); + } + + if (devid_get(fd, &devid) == 0) { + if (devid_get_minor_name(fd, &minor) == 0 && + (devid_str = devid_str_encode(devid, minor)) != + NULL) { + verify(nvlist_add_string(vdev, + ZPOOL_CONFIG_DEVID, devid_str) == 0); + } + if (devid_str != NULL) + devid_str_free(devid_str); + if (minor != NULL) + devid_str_free(minor); + devid_free(devid); + } + + (void) close(fd); + } + + return (vdev); +} + +/* + * Go through and verify the replication level of the pool is consistent. + * Performs the following checks: + * + * For the new spec, verifies that devices in mirrors and raidz are the + * same size. + * + * If the current configuration already has inconsistent replication + * levels, ignore any other potential problems in the new spec. + * + * Otherwise, make sure that the current spec (if there is one) and the new + * spec have consistent replication levels. + */ +typedef struct replication_level { + char *zprl_type; + uint64_t zprl_children; + uint64_t zprl_parity; +} replication_level_t; + +#define ZPOOL_FUZZ (16 * 1024 * 1024) + +/* + * Given a list of toplevel vdevs, return the current replication level. If + * the config is inconsistent, then NULL is returned. If 'fatal' is set, then + * an error message will be displayed for each self-inconsistent vdev. + */ +static replication_level_t * +get_replication(nvlist_t *nvroot, boolean_t fatal) +{ + nvlist_t **top; + uint_t t, toplevels; + nvlist_t **child; + uint_t c, children; + nvlist_t *nv; + char *type; + replication_level_t lastrep, rep, *ret; + boolean_t dontreport; + + ret = safe_malloc(sizeof (replication_level_t)); + + verify(nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, + &top, &toplevels) == 0); + + lastrep.zprl_type = NULL; + for (t = 0; t < toplevels; t++) { + uint64_t is_log = B_FALSE; + + nv = top[t]; + + /* + * For separate logs we ignore the top level vdev replication + * constraints. + */ + (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_LOG, &is_log); + if (is_log) + continue; + + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, + &type) == 0); + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) { + /* + * This is a 'file' or 'disk' vdev. + */ + rep.zprl_type = type; + rep.zprl_children = 1; + rep.zprl_parity = 0; + } else { + uint64_t vdev_size; + + /* + * This is a mirror or RAID-Z vdev. Go through and make + * sure the contents are all the same (files vs. disks), + * keeping track of the number of elements in the + * process. + * + * We also check that the size of each vdev (if it can + * be determined) is the same. + */ + rep.zprl_type = type; + rep.zprl_children = 0; + + if (strcmp(type, VDEV_TYPE_RAIDZ) == 0) { + verify(nvlist_lookup_uint64(nv, + ZPOOL_CONFIG_NPARITY, + &rep.zprl_parity) == 0); + assert(rep.zprl_parity != 0); + } else { + rep.zprl_parity = 0; + } + + /* + * The 'dontreport' variable indicates that we've + * already reported an error for this spec, so don't + * bother doing it again. + */ + type = NULL; + dontreport = 0; + vdev_size = -1ULL; + for (c = 0; c < children; c++) { + nvlist_t *cnv = child[c]; + char *path; + struct stat64 statbuf; + uint64_t size = -1ULL; + char *childtype; + int fd, err; + + rep.zprl_children++; + + verify(nvlist_lookup_string(cnv, + ZPOOL_CONFIG_TYPE, &childtype) == 0); + + /* + * If this is a replacing or spare vdev, then + * get the real first child of the vdev. + */ + if (strcmp(childtype, + VDEV_TYPE_REPLACING) == 0 || + strcmp(childtype, VDEV_TYPE_SPARE) == 0) { + nvlist_t **rchild; + uint_t rchildren; + + verify(nvlist_lookup_nvlist_array(cnv, + ZPOOL_CONFIG_CHILDREN, &rchild, + &rchildren) == 0); + assert(rchildren == 2); + cnv = rchild[0]; + + verify(nvlist_lookup_string(cnv, + ZPOOL_CONFIG_TYPE, + &childtype) == 0); + } + + verify(nvlist_lookup_string(cnv, + ZPOOL_CONFIG_PATH, &path) == 0); + + /* + * If we have a raidz/mirror that combines disks + * with files, report it as an error. + */ + if (!dontreport && type != NULL && + strcmp(type, childtype) != 0) { + if (ret != NULL) + free(ret); + ret = NULL; + if (fatal) + vdev_error(gettext( + "mismatched replication " + "level: %s contains both " + "files and devices\n"), + rep.zprl_type); + else + return (NULL); + dontreport = B_TRUE; + } + + /* + * According to stat(2), the value of 'st_size' + * is undefined for block devices and character + * devices. But there is no effective way to + * determine the real size in userland. + * + * Instead, we'll take advantage of an + * implementation detail of spec_size(). If the + * device is currently open, then we (should) + * return a valid size. + * + * If we still don't get a valid size (indicated + * by a size of 0 or MAXOFFSET_T), then ignore + * this device altogether. + */ + if ((fd = open(path, O_RDONLY)) >= 0) { + err = fstat64(fd, &statbuf); + (void) close(fd); + } else { + err = stat64(path, &statbuf); + } + + if (err != 0 || + statbuf.st_size == 0 || + statbuf.st_size == MAXOFFSET_T) + continue; + + size = statbuf.st_size; + + /* + * Also make sure that devices and + * slices have a consistent size. If + * they differ by a significant amount + * (~16MB) then report an error. + */ + if (!dontreport && + (vdev_size != -1ULL && + (labs(size - vdev_size) > + ZPOOL_FUZZ))) { + if (ret != NULL) + free(ret); + ret = NULL; + if (fatal) + vdev_error(gettext( + "%s contains devices of " + "different sizes\n"), + rep.zprl_type); + else + return (NULL); + dontreport = B_TRUE; + } + + type = childtype; + vdev_size = size; + } + } + + /* + * At this point, we have the replication of the last toplevel + * vdev in 'rep'. Compare it to 'lastrep' to see if its + * different. + */ + if (lastrep.zprl_type != NULL) { + if (strcmp(lastrep.zprl_type, rep.zprl_type) != 0) { + if (ret != NULL) + free(ret); + ret = NULL; + if (fatal) + vdev_error(gettext( + "mismatched replication level: " + "both %s and %s vdevs are " + "present\n"), + lastrep.zprl_type, rep.zprl_type); + else + return (NULL); + } else if (lastrep.zprl_parity != rep.zprl_parity) { + if (ret) + free(ret); + ret = NULL; + if (fatal) + vdev_error(gettext( + "mismatched replication level: " + "both %llu and %llu device parity " + "%s vdevs are present\n"), + lastrep.zprl_parity, + rep.zprl_parity, + rep.zprl_type); + else + return (NULL); + } else if (lastrep.zprl_children != rep.zprl_children) { + if (ret) + free(ret); + ret = NULL; + if (fatal) + vdev_error(gettext( + "mismatched replication level: " + "both %llu-way and %llu-way %s " + "vdevs are present\n"), + lastrep.zprl_children, + rep.zprl_children, + rep.zprl_type); + else + return (NULL); + } + } + lastrep = rep; + } + + if (ret != NULL) + *ret = rep; + + return (ret); +} + +/* + * Check the replication level of the vdev spec against the current pool. Calls + * get_replication() to make sure the new spec is self-consistent. If the pool + * has a consistent replication level, then we ignore any errors. Otherwise, + * report any difference between the two. + */ +static int +check_replication(nvlist_t *config, nvlist_t *newroot) +{ + nvlist_t **child; + uint_t children; + replication_level_t *current = NULL, *new; + int ret; + + /* + * If we have a current pool configuration, check to see if it's + * self-consistent. If not, simply return success. + */ + if (config != NULL) { + nvlist_t *nvroot; + + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &nvroot) == 0); + if ((current = get_replication(nvroot, B_FALSE)) == NULL) + return (0); + } + /* + * for spares there may be no children, and therefore no + * replication level to check + */ + if ((nvlist_lookup_nvlist_array(newroot, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) || (children == 0)) { + free(current); + return (0); + } + + /* + * If all we have is logs then there's no replication level to check. + */ + if (num_logs(newroot) == children) { + free(current); + return (0); + } + + /* + * Get the replication level of the new vdev spec, reporting any + * inconsistencies found. + */ + if ((new = get_replication(newroot, B_TRUE)) == NULL) { + free(current); + return (-1); + } + + /* + * Check to see if the new vdev spec matches the replication level of + * the current pool. + */ + ret = 0; + if (current != NULL) { + if (strcmp(current->zprl_type, new->zprl_type) != 0) { + vdev_error(gettext( + "mismatched replication level: pool uses %s " + "and new vdev is %s\n"), + current->zprl_type, new->zprl_type); + ret = -1; + } else if (current->zprl_parity != new->zprl_parity) { + vdev_error(gettext( + "mismatched replication level: pool uses %llu " + "device parity and new vdev uses %llu\n"), + current->zprl_parity, new->zprl_parity); + ret = -1; + } else if (current->zprl_children != new->zprl_children) { + vdev_error(gettext( + "mismatched replication level: pool uses %llu-way " + "%s and new vdev uses %llu-way %s\n"), + current->zprl_children, current->zprl_type, + new->zprl_children, new->zprl_type); + ret = -1; + } + } + + free(new); + if (current != NULL) + free(current); + + return (ret); +} + +/* + * Go through and find any whole disks in the vdev specification, labelling them + * as appropriate. When constructing the vdev spec, we were unable to open this + * device in order to provide a devid. Now that we have labelled the disk and + * know that slice 0 is valid, we can construct the devid now. + * + * If the disk was already labeled with an EFI label, we will have gotten the + * devid already (because we were able to open the whole disk). Otherwise, we + * need to get the devid after we label the disk. + */ +static int +make_disks(zpool_handle_t *zhp, nvlist_t *nv) +{ + nvlist_t **child; + uint_t c, children; + char *type, *path, *diskname; + char buf[MAXPATHLEN]; + uint64_t wholedisk; + int fd; + int ret; + ddi_devid_t devid; + char *minor = NULL, *devid_str = NULL; + + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) == 0); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) { + + if (strcmp(type, VDEV_TYPE_DISK) != 0) + return (0); + + /* + * We have a disk device. Get the path to the device + * and see if it's a whole disk by appending the backup + * slice and stat()ing the device. + */ + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) == 0); + if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK, + &wholedisk) != 0 || !wholedisk) + return (0); + + diskname = strrchr(path, '/'); + assert(diskname != NULL); + diskname++; + if (zpool_label_disk(g_zfs, zhp, diskname) == -1) + return (-1); + + /* + * Fill in the devid, now that we've labeled the disk. + */ + (void) snprintf(buf, sizeof (buf), "%ss0", path); + if ((fd = open(buf, O_RDONLY)) < 0) { + (void) fprintf(stderr, + gettext("cannot open '%s': %s\n"), + buf, strerror(errno)); + return (-1); + } + + if (devid_get(fd, &devid) == 0) { + if (devid_get_minor_name(fd, &minor) == 0 && + (devid_str = devid_str_encode(devid, minor)) != + NULL) { + verify(nvlist_add_string(nv, + ZPOOL_CONFIG_DEVID, devid_str) == 0); + } + if (devid_str != NULL) + devid_str_free(devid_str); + if (minor != NULL) + devid_str_free(minor); + devid_free(devid); + } + + /* + * Update the path to refer to the 's0' slice. The presence of + * the 'whole_disk' field indicates to the CLI that we should + * chop off the slice number when displaying the device in + * future output. + */ + verify(nvlist_add_string(nv, ZPOOL_CONFIG_PATH, buf) == 0); + + (void) close(fd); + + return (0); + } + + for (c = 0; c < children; c++) + if ((ret = make_disks(zhp, child[c])) != 0) + return (ret); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES, + &child, &children) == 0) + for (c = 0; c < children; c++) + if ((ret = make_disks(zhp, child[c])) != 0) + return (ret); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE, + &child, &children) == 0) + for (c = 0; c < children; c++) + if ((ret = make_disks(zhp, child[c])) != 0) + return (ret); + + return (0); +} + +/* + * Determine if the given path is a hot spare within the given configuration. + */ +static boolean_t +is_spare(nvlist_t *config, const char *path) +{ + int fd; + pool_state_t state; + char *name = NULL; + nvlist_t *label; + uint64_t guid, spareguid; + nvlist_t *nvroot; + nvlist_t **spares; + uint_t i, nspares; + boolean_t inuse; + + if ((fd = open(path, O_RDONLY)) < 0) + return (B_FALSE); + + if (zpool_in_use(g_zfs, fd, &state, &name, &inuse) != 0 || + !inuse || + state != POOL_STATE_SPARE || + zpool_read_label(fd, &label) != 0) { + free(name); + (void) close(fd); + return (B_FALSE); + } + free(name); + + (void) close(fd); + verify(nvlist_lookup_uint64(label, ZPOOL_CONFIG_GUID, &guid) == 0); + nvlist_free(label); + + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &nvroot) == 0); + if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, + &spares, &nspares) == 0) { + for (i = 0; i < nspares; i++) { + verify(nvlist_lookup_uint64(spares[i], + ZPOOL_CONFIG_GUID, &spareguid) == 0); + if (spareguid == guid) + return (B_TRUE); + } + } + + return (B_FALSE); +} + +/* + * Go through and find any devices that are in use. We rely on libdiskmgt for + * the majority of this task. + */ +static int +check_in_use(nvlist_t *config, nvlist_t *nv, int force, int isreplacing, + int isspare) +{ + nvlist_t **child; + uint_t c, children; + char *type, *path; + int ret; + char buf[MAXPATHLEN]; + uint64_t wholedisk; + + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) == 0); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) { + + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) == 0); + + /* + * As a generic check, we look to see if this is a replace of a + * hot spare within the same pool. If so, we allow it + * regardless of what libdiskmgt or zpool_in_use() says. + */ + if (isreplacing) { + if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK, + &wholedisk) == 0 && wholedisk) + (void) snprintf(buf, sizeof (buf), "%ss0", + path); + else + (void) strlcpy(buf, path, sizeof (buf)); + if (is_spare(config, buf)) + return (0); + } + + if (strcmp(type, VDEV_TYPE_DISK) == 0) + ret = check_device(path, force, isspare); + + if (strcmp(type, VDEV_TYPE_FILE) == 0) + ret = check_file(path, force, isspare); + + return (ret); + } + + for (c = 0; c < children; c++) + if ((ret = check_in_use(config, child[c], force, + isreplacing, B_FALSE)) != 0) + return (ret); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES, + &child, &children) == 0) + for (c = 0; c < children; c++) + if ((ret = check_in_use(config, child[c], force, + isreplacing, B_TRUE)) != 0) + return (ret); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE, + &child, &children) == 0) + for (c = 0; c < children; c++) + if ((ret = check_in_use(config, child[c], force, + isreplacing, B_FALSE)) != 0) + return (ret); + + return (0); +} + +static const char * +is_grouping(const char *type, int *mindev) +{ + if (strcmp(type, "raidz") == 0 || strcmp(type, "raidz1") == 0) { + if (mindev != NULL) + *mindev = 2; + return (VDEV_TYPE_RAIDZ); + } + + if (strcmp(type, "raidz2") == 0) { + if (mindev != NULL) + *mindev = 3; + return (VDEV_TYPE_RAIDZ); + } + + if (strcmp(type, "mirror") == 0) { + if (mindev != NULL) + *mindev = 2; + return (VDEV_TYPE_MIRROR); + } + + if (strcmp(type, "spare") == 0) { + if (mindev != NULL) + *mindev = 1; + return (VDEV_TYPE_SPARE); + } + + if (strcmp(type, "log") == 0) { + if (mindev != NULL) + *mindev = 1; + return (VDEV_TYPE_LOG); + } + + if (strcmp(type, "cache") == 0) { + if (mindev != NULL) + *mindev = 1; + return (VDEV_TYPE_L2CACHE); + } + + return (NULL); +} + +/* + * Construct a syntactically valid vdev specification, + * and ensure that all devices and files exist and can be opened. + * Note: we don't bother freeing anything in the error paths + * because the program is just going to exit anyway. + */ +nvlist_t * +construct_spec(int argc, char **argv) +{ + nvlist_t *nvroot, *nv, **top, **spares, **l2cache; + int t, toplevels, mindev, nspares, nlogs, nl2cache; + const char *type; + uint64_t is_log; + boolean_t seen_logs; + + top = NULL; + toplevels = 0; + spares = NULL; + l2cache = NULL; + nspares = 0; + nlogs = 0; + nl2cache = 0; + is_log = B_FALSE; + seen_logs = B_FALSE; + + while (argc > 0) { + nv = NULL; + + /* + * If it's a mirror or raidz, the subsequent arguments are + * its leaves -- until we encounter the next mirror or raidz. + */ + if ((type = is_grouping(argv[0], &mindev)) != NULL) { + nvlist_t **child = NULL; + int c, children = 0; + + if (strcmp(type, VDEV_TYPE_SPARE) == 0) { + if (spares != NULL) { + (void) fprintf(stderr, + gettext("invalid vdev " + "specification: 'spare' can be " + "specified only once\n")); + return (NULL); + } + is_log = B_FALSE; + } + + if (strcmp(type, VDEV_TYPE_LOG) == 0) { + if (seen_logs) { + (void) fprintf(stderr, + gettext("invalid vdev " + "specification: 'log' can be " + "specified only once\n")); + return (NULL); + } + seen_logs = B_TRUE; + is_log = B_TRUE; + argc--; + argv++; + /* + * A log is not a real grouping device. + * We just set is_log and continue. + */ + continue; + } + + if (strcmp(type, VDEV_TYPE_L2CACHE) == 0) { + if (l2cache != NULL) { + (void) fprintf(stderr, + gettext("invalid vdev " + "specification: 'cache' can be " + "specified only once\n")); + return (NULL); + } + is_log = B_FALSE; + } + + if (is_log) { + if (strcmp(type, VDEV_TYPE_MIRROR) != 0) { + (void) fprintf(stderr, + gettext("invalid vdev " + "specification: unsupported 'log' " + "device: %s\n"), type); + return (NULL); + } + nlogs++; + } + + for (c = 1; c < argc; c++) { + if (is_grouping(argv[c], NULL) != NULL) + break; + children++; + child = realloc(child, + children * sizeof (nvlist_t *)); + if (child == NULL) + zpool_no_memory(); + if ((nv = make_leaf_vdev(argv[c], B_FALSE)) + == NULL) + return (NULL); + child[children - 1] = nv; + } + + if (children < mindev) { + (void) fprintf(stderr, gettext("invalid vdev " + "specification: %s requires at least %d " + "devices\n"), argv[0], mindev); + return (NULL); + } + + argc -= c; + argv += c; + + if (strcmp(type, VDEV_TYPE_SPARE) == 0) { + spares = child; + nspares = children; + continue; + } else if (strcmp(type, VDEV_TYPE_L2CACHE) == 0) { + l2cache = child; + nl2cache = children; + continue; + } else { + verify(nvlist_alloc(&nv, NV_UNIQUE_NAME, + 0) == 0); + verify(nvlist_add_string(nv, ZPOOL_CONFIG_TYPE, + type) == 0); + verify(nvlist_add_uint64(nv, + ZPOOL_CONFIG_IS_LOG, is_log) == 0); + if (strcmp(type, VDEV_TYPE_RAIDZ) == 0) { + verify(nvlist_add_uint64(nv, + ZPOOL_CONFIG_NPARITY, + mindev - 1) == 0); + } + verify(nvlist_add_nvlist_array(nv, + ZPOOL_CONFIG_CHILDREN, child, + children) == 0); + + for (c = 0; c < children; c++) + nvlist_free(child[c]); + free(child); + } + } else { + /* + * We have a device. Pass off to make_leaf_vdev() to + * construct the appropriate nvlist describing the vdev. + */ + if ((nv = make_leaf_vdev(argv[0], is_log)) == NULL) + return (NULL); + if (is_log) + nlogs++; + argc--; + argv++; + } + + toplevels++; + top = realloc(top, toplevels * sizeof (nvlist_t *)); + if (top == NULL) + zpool_no_memory(); + top[toplevels - 1] = nv; + } + + if (toplevels == 0 && nspares == 0 && nl2cache == 0) { + (void) fprintf(stderr, gettext("invalid vdev " + "specification: at least one toplevel vdev must be " + "specified\n")); + return (NULL); + } + + if (seen_logs && nlogs == 0) { + (void) fprintf(stderr, gettext("invalid vdev specification: " + "log requires at least 1 device\n")); + return (NULL); + } + + /* + * Finally, create nvroot and add all top-level vdevs to it. + */ + verify(nvlist_alloc(&nvroot, NV_UNIQUE_NAME, 0) == 0); + verify(nvlist_add_string(nvroot, ZPOOL_CONFIG_TYPE, + VDEV_TYPE_ROOT) == 0); + verify(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, + top, toplevels) == 0); + if (nspares != 0) + verify(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, + spares, nspares) == 0); + if (nl2cache != 0) + verify(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, + l2cache, nl2cache) == 0); + + for (t = 0; t < toplevels; t++) + nvlist_free(top[t]); + for (t = 0; t < nspares; t++) + nvlist_free(spares[t]); + for (t = 0; t < nl2cache; t++) + nvlist_free(l2cache[t]); + if (spares) + free(spares); + if (l2cache) + free(l2cache); + free(top); + + return (nvroot); +} + + +/* + * Get and validate the contents of the given vdev specification. This ensures + * that the nvlist returned is well-formed, that all the devices exist, and that + * they are not currently in use by any other known consumer. The 'poolconfig' + * parameter is the current configuration of the pool when adding devices + * existing pool, and is used to perform additional checks, such as changing the + * replication level of the pool. It can be 'NULL' to indicate that this is a + * new pool. The 'force' flag controls whether devices should be forcefully + * added, even if they appear in use. + */ +nvlist_t * +make_root_vdev(zpool_handle_t *zhp, int force, int check_rep, + boolean_t isreplacing, boolean_t dryrun, int argc, char **argv) +{ + nvlist_t *newroot; + nvlist_t *poolconfig = NULL; + is_force = force; + + /* + * Construct the vdev specification. If this is successful, we know + * that we have a valid specification, and that all devices can be + * opened. + */ + if ((newroot = construct_spec(argc, argv)) == NULL) + return (NULL); + + if (zhp && ((poolconfig = zpool_get_config(zhp, NULL)) == NULL)) + return (NULL); + + /* + * Validate each device to make sure that its not shared with another + * subsystem. We do this even if 'force' is set, because there are some + * uses (such as a dedicated dump device) that even '-f' cannot + * override. + */ + if (check_in_use(poolconfig, newroot, force, isreplacing, + B_FALSE) != 0) { + nvlist_free(newroot); + return (NULL); + } + + /* + * Check the replication level of the given vdevs and report any errors + * found. We include the existing pool spec, if any, as we need to + * catch changes against the existing replication level. + */ + if (check_rep && check_replication(poolconfig, newroot) != 0) { + nvlist_free(newroot); + return (NULL); + } + + /* + * Run through the vdev specification and label any whole disks found. + */ + if (!dryrun && make_disks(zhp, newroot) != 0) { + nvlist_free(newroot); + return (NULL); + } + + return (newroot); +} diff --git a/cmd/ztest/ztest.c b/cmd/ztest/ztest.c new file mode 100644 index 000000000..4aba99126 --- /dev/null +++ b/cmd/ztest/ztest.c @@ -0,0 +1,3476 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ +/* + * Copyright 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +/* + * The objective of this program is to provide a DMU/ZAP/SPA stress test + * that runs entirely in userland, is easy to use, and easy to extend. + * + * The overall design of the ztest program is as follows: + * + * (1) For each major functional area (e.g. adding vdevs to a pool, + * creating and destroying datasets, reading and writing objects, etc) + * we have a simple routine to test that functionality. These + * individual routines do not have to do anything "stressful". + * + * (2) We turn these simple functionality tests into a stress test by + * running them all in parallel, with as many threads as desired, + * and spread across as many datasets, objects, and vdevs as desired. + * + * (3) While all this is happening, we inject faults into the pool to + * verify that self-healing data really works. + * + * (4) Every time we open a dataset, we change its checksum and compression + * functions. Thus even individual objects vary from block to block + * in which checksum they use and whether they're compressed. + * + * (5) To verify that we never lose on-disk consistency after a crash, + * we run the entire test in a child of the main process. + * At random times, the child self-immolates with a SIGKILL. + * This is the software equivalent of pulling the power cord. + * The parent then runs the test again, using the existing + * storage pool, as many times as desired. + * + * (6) To verify that we don't have future leaks or temporal incursions, + * many of the functional tests record the transaction group number + * as part of their data. When reading old data, they verify that + * the transaction group number is less than the current, open txg. + * If you add a new test, please do this if applicable. + * + * When run with no arguments, ztest runs for about five minutes and + * produces no output if successful. To get a little bit of information, + * specify -V. To get more information, specify -VV, and so on. + * + * To turn this into an overnight stress test, use -T to specify run time. + * + * You can ask more more vdevs [-v], datasets [-d], or threads [-t] + * to increase the pool capacity, fanout, and overall stress level. + * + * The -N(okill) option will suppress kills, so each child runs to completion. + * This can be useful when you're trying to distinguish temporal incursions + * from plain old race conditions. + */ + +#include <sys/zfs_context.h> +#include <sys/spa.h> +#include <sys/dmu.h> +#include <sys/txg.h> +#include <sys/zap.h> +#include <sys/dmu_objset.h> +#include <sys/poll.h> +#include <sys/stat.h> +#include <sys/time.h> +#include <sys/wait.h> +#include <sys/mman.h> +#include <sys/resource.h> +#include <sys/zio.h> +#include <sys/zio_checksum.h> +#include <sys/zio_compress.h> +#include <sys/zil.h> +#include <sys/vdev_impl.h> +#include <sys/vdev_file.h> +#include <sys/spa_impl.h> +#include <sys/dsl_prop.h> +#include <sys/refcount.h> +#include <stdio.h> +#include <stdio_ext.h> +#include <stdlib.h> +#include <unistd.h> +#include <signal.h> +#include <umem.h> +#include <dlfcn.h> +#include <ctype.h> +#include <math.h> +#include <sys/fs/zfs.h> + +static char cmdname[] = "ztest"; +static char *zopt_pool = cmdname; + +static uint64_t zopt_vdevs = 5; +static uint64_t zopt_vdevtime; +static int zopt_ashift = SPA_MINBLOCKSHIFT; +static int zopt_mirrors = 2; +static int zopt_raidz = 4; +static int zopt_raidz_parity = 1; +static size_t zopt_vdev_size = SPA_MINDEVSIZE; +static int zopt_datasets = 7; +static int zopt_threads = 23; +static uint64_t zopt_passtime = 60; /* 60 seconds */ +static uint64_t zopt_killrate = 70; /* 70% kill rate */ +static int zopt_verbose = 0; +static int zopt_init = 1; +static char *zopt_dir = "/tmp"; +static uint64_t zopt_time = 300; /* 5 minutes */ +static int zopt_maxfaults; + +typedef struct ztest_block_tag { + uint64_t bt_objset; + uint64_t bt_object; + uint64_t bt_offset; + uint64_t bt_txg; + uint64_t bt_thread; + uint64_t bt_seq; +} ztest_block_tag_t; + +typedef struct ztest_args { + char za_pool[MAXNAMELEN]; + spa_t *za_spa; + objset_t *za_os; + zilog_t *za_zilog; + pthread_t za_thread; + uint64_t za_instance; + uint64_t za_random; + uint64_t za_diroff; + uint64_t za_diroff_shared; + uint64_t za_zil_seq; + hrtime_t za_start; + hrtime_t za_stop; + hrtime_t za_kill; + /* + * Thread-local variables can go here to aid debugging. + */ + ztest_block_tag_t za_rbt; + ztest_block_tag_t za_wbt; + dmu_object_info_t za_doi; + dmu_buf_t *za_dbuf; +} ztest_args_t; + +typedef void ztest_func_t(ztest_args_t *); + +/* + * Note: these aren't static because we want dladdr() to work. + */ +ztest_func_t ztest_dmu_read_write; +ztest_func_t ztest_dmu_write_parallel; +ztest_func_t ztest_dmu_object_alloc_free; +ztest_func_t ztest_zap; +ztest_func_t ztest_zap_parallel; +ztest_func_t ztest_traverse; +ztest_func_t ztest_dsl_prop_get_set; +ztest_func_t ztest_dmu_objset_create_destroy; +ztest_func_t ztest_dmu_snapshot_create_destroy; +ztest_func_t ztest_spa_create_destroy; +ztest_func_t ztest_fault_inject; +ztest_func_t ztest_spa_rename; +ztest_func_t ztest_vdev_attach_detach; +ztest_func_t ztest_vdev_LUN_growth; +ztest_func_t ztest_vdev_add_remove; +ztest_func_t ztest_vdev_aux_add_remove; +ztest_func_t ztest_scrub; + +typedef struct ztest_info { + ztest_func_t *zi_func; /* test function */ + uint64_t zi_iters; /* iterations per execution */ + uint64_t *zi_interval; /* execute every <interval> seconds */ + uint64_t zi_calls; /* per-pass count */ + uint64_t zi_call_time; /* per-pass time */ + uint64_t zi_call_total; /* cumulative total */ + uint64_t zi_call_target; /* target cumulative total */ +} ztest_info_t; + +uint64_t zopt_always = 0; /* all the time */ +uint64_t zopt_often = 1; /* every second */ +uint64_t zopt_sometimes = 10; /* every 10 seconds */ +uint64_t zopt_rarely = 60; /* every 60 seconds */ + +ztest_info_t ztest_info[] = { + { ztest_dmu_read_write, 1, &zopt_always }, + { ztest_dmu_write_parallel, 30, &zopt_always }, + { ztest_dmu_object_alloc_free, 1, &zopt_always }, + { ztest_zap, 30, &zopt_always }, + { ztest_zap_parallel, 100, &zopt_always }, + { ztest_dsl_prop_get_set, 1, &zopt_sometimes }, + { ztest_dmu_objset_create_destroy, 1, &zopt_sometimes }, + { ztest_dmu_snapshot_create_destroy, 1, &zopt_sometimes }, + { ztest_spa_create_destroy, 1, &zopt_sometimes }, + { ztest_fault_inject, 1, &zopt_sometimes }, + { ztest_spa_rename, 1, &zopt_rarely }, + { ztest_vdev_attach_detach, 1, &zopt_rarely }, + { ztest_vdev_LUN_growth, 1, &zopt_rarely }, + { ztest_vdev_add_remove, 1, &zopt_vdevtime }, + { ztest_vdev_aux_add_remove, 1, &zopt_vdevtime }, + { ztest_scrub, 1, &zopt_vdevtime }, +}; + +#define ZTEST_FUNCS (sizeof (ztest_info) / sizeof (ztest_info_t)) + +#define ZTEST_SYNC_LOCKS 16 + +/* + * Stuff we need to share writably between parent and child. + */ +typedef struct ztest_shared { + pthread_mutex_t zs_vdev_lock; + pthread_rwlock_t zs_name_lock; + uint64_t zs_vdev_primaries; + uint64_t zs_vdev_aux; + uint64_t zs_enospc_count; + hrtime_t zs_start_time; + hrtime_t zs_stop_time; + uint64_t zs_alloc; + uint64_t zs_space; + ztest_info_t zs_info[ZTEST_FUNCS]; + pthread_mutex_t zs_sync_lock[ZTEST_SYNC_LOCKS]; + uint64_t zs_seq[ZTEST_SYNC_LOCKS]; +} ztest_shared_t; + +static char ztest_dev_template[] = "%s/%s.%llua"; +static char ztest_aux_template[] = "%s/%s.%s.%llu"; +static ztest_shared_t *ztest_shared; + +static int ztest_random_fd; +static int ztest_dump_core = 1; + +static boolean_t ztest_exiting; + +extern uint64_t metaslab_gang_bang; + +#define ZTEST_DIROBJ 1 +#define ZTEST_MICROZAP_OBJ 2 +#define ZTEST_FATZAP_OBJ 3 + +#define ZTEST_DIROBJ_BLOCKSIZE (1 << 10) +#define ZTEST_DIRSIZE 256 + +static void usage(boolean_t) __NORETURN; + +/* + * These libumem hooks provide a reasonable set of defaults for the allocator's + * debugging facilities. + */ +const char * +_umem_debug_init() +{ + return ("default,verbose"); /* $UMEM_DEBUG setting */ +} + +const char * +_umem_logging_init(void) +{ + return ("fail,contents"); /* $UMEM_LOGGING setting */ +} + +#define FATAL_MSG_SZ 1024 + +char *fatal_msg; + +static void +fatal(int do_perror, char *message, ...) +{ + va_list args; + int save_errno = errno; + char buf[FATAL_MSG_SZ]; + + (void) fflush(stdout); + + va_start(args, message); + (void) sprintf(buf, "ztest: "); + /* LINTED */ + (void) vsprintf(buf + strlen(buf), message, args); + va_end(args); + if (do_perror) { + (void) snprintf(buf + strlen(buf), FATAL_MSG_SZ - strlen(buf), + ": %s", strerror(save_errno)); + } + (void) fprintf(stderr, "%s\n", buf); + fatal_msg = buf; /* to ease debugging */ + if (ztest_dump_core) + abort(); + exit(3); +} + +static int +str2shift(const char *buf) +{ + const char *ends = "BKMGTPEZ"; + int i; + + if (buf[0] == '\0') + return (0); + for (i = 0; i < strlen(ends); i++) { + if (toupper(buf[0]) == ends[i]) + break; + } + if (i == strlen(ends)) { + (void) fprintf(stderr, "ztest: invalid bytes suffix: %s\n", + buf); + usage(B_FALSE); + } + if (buf[1] == '\0' || (toupper(buf[1]) == 'B' && buf[2] == '\0')) { + return (10*i); + } + (void) fprintf(stderr, "ztest: invalid bytes suffix: %s\n", buf); + usage(B_FALSE); + /* NOTREACHED */ +} + +static uint64_t +nicenumtoull(const char *buf) +{ + char *end; + uint64_t val; + + val = strtoull(buf, &end, 0); + if (end == buf) { + (void) fprintf(stderr, "ztest: bad numeric value: %s\n", buf); + usage(B_FALSE); + } else if (end[0] == '.') { + double fval = strtod(buf, &end); + fval *= pow(2, str2shift(end)); + if (fval > UINT64_MAX) { + (void) fprintf(stderr, "ztest: value too large: %s\n", + buf); + usage(B_FALSE); + } + val = (uint64_t)fval; + } else { + int shift = str2shift(end); + if (shift >= 64 || (val << shift) >> shift != val) { + (void) fprintf(stderr, "ztest: value too large: %s\n", + buf); + usage(B_FALSE); + } + val <<= shift; + } + return (val); +} + +static void +usage(boolean_t requested) +{ + char nice_vdev_size[10]; + char nice_gang_bang[10]; + FILE *fp = requested ? stdout : stderr; + + nicenum(zopt_vdev_size, nice_vdev_size); + nicenum(metaslab_gang_bang, nice_gang_bang); + + (void) fprintf(fp, "Usage: %s\n" + "\t[-v vdevs (default: %llu)]\n" + "\t[-s size_of_each_vdev (default: %s)]\n" + "\t[-a alignment_shift (default: %d) (use 0 for random)]\n" + "\t[-m mirror_copies (default: %d)]\n" + "\t[-r raidz_disks (default: %d)]\n" + "\t[-R raidz_parity (default: %d)]\n" + "\t[-d datasets (default: %d)]\n" + "\t[-t threads (default: %d)]\n" + "\t[-g gang_block_threshold (default: %s)]\n" + "\t[-i initialize pool i times (default: %d)]\n" + "\t[-k kill percentage (default: %llu%%)]\n" + "\t[-p pool_name (default: %s)]\n" + "\t[-f file directory for vdev files (default: %s)]\n" + "\t[-V(erbose)] (use multiple times for ever more blather)\n" + "\t[-E(xisting)] (use existing pool instead of creating new one)\n" + "\t[-T time] total run time (default: %llu sec)\n" + "\t[-P passtime] time per pass (default: %llu sec)\n" + "\t[-h] (print help)\n" + "", + cmdname, + (u_longlong_t)zopt_vdevs, /* -v */ + nice_vdev_size, /* -s */ + zopt_ashift, /* -a */ + zopt_mirrors, /* -m */ + zopt_raidz, /* -r */ + zopt_raidz_parity, /* -R */ + zopt_datasets, /* -d */ + zopt_threads, /* -t */ + nice_gang_bang, /* -g */ + zopt_init, /* -i */ + (u_longlong_t)zopt_killrate, /* -k */ + zopt_pool, /* -p */ + zopt_dir, /* -f */ + (u_longlong_t)zopt_time, /* -T */ + (u_longlong_t)zopt_passtime); /* -P */ + exit(requested ? 0 : 1); +} + +static uint64_t +ztest_random(uint64_t range) +{ + uint64_t r; + + if (range == 0) + return (0); + + if (read(ztest_random_fd, &r, sizeof (r)) != sizeof (r)) + fatal(1, "short read from /dev/urandom"); + + return (r % range); +} + +static void +ztest_record_enospc(char *s) +{ + dprintf("ENOSPC doing: %s\n", s ? s : "<unknown>"); + ztest_shared->zs_enospc_count++; +} + +static void +process_options(int argc, char **argv) +{ + int opt; + uint64_t value; + + /* By default, test gang blocks for blocks 32K and greater */ + metaslab_gang_bang = 32 << 10; + + while ((opt = getopt(argc, argv, + "v:s:a:m:r:R:d:t:g:i:k:p:f:VET:P:h")) != EOF) { + value = 0; + switch (opt) { + case 'v': + case 's': + case 'a': + case 'm': + case 'r': + case 'R': + case 'd': + case 't': + case 'g': + case 'i': + case 'k': + case 'T': + case 'P': + value = nicenumtoull(optarg); + } + switch (opt) { + case 'v': + zopt_vdevs = value; + break; + case 's': + zopt_vdev_size = MAX(SPA_MINDEVSIZE, value); + break; + case 'a': + zopt_ashift = value; + break; + case 'm': + zopt_mirrors = value; + break; + case 'r': + zopt_raidz = MAX(1, value); + break; + case 'R': + zopt_raidz_parity = MIN(MAX(value, 1), 2); + break; + case 'd': + zopt_datasets = MAX(1, value); + break; + case 't': + zopt_threads = MAX(1, value); + break; + case 'g': + metaslab_gang_bang = MAX(SPA_MINBLOCKSIZE << 1, value); + break; + case 'i': + zopt_init = value; + break; + case 'k': + zopt_killrate = value; + break; + case 'p': + zopt_pool = strdup(optarg); + break; + case 'f': + zopt_dir = strdup(optarg); + break; + case 'V': + zopt_verbose++; + break; + case 'E': + zopt_init = 0; + break; + case 'T': + zopt_time = value; + break; + case 'P': + zopt_passtime = MAX(1, value); + break; + case 'h': + usage(B_TRUE); + break; + case '?': + default: + usage(B_FALSE); + break; + } + } + + zopt_raidz_parity = MIN(zopt_raidz_parity, zopt_raidz - 1); + + zopt_vdevtime = (zopt_vdevs > 0 ? zopt_time / zopt_vdevs : UINT64_MAX); + zopt_maxfaults = MAX(zopt_mirrors, 1) * (zopt_raidz_parity + 1) - 1; +} + +static uint64_t +ztest_get_ashift(void) +{ + if (zopt_ashift == 0) + return (SPA_MINBLOCKSHIFT + ztest_random(3)); + return (zopt_ashift); +} + +static nvlist_t * +make_vdev_file(char *path, char *aux, size_t size, uint64_t ashift) +{ + char pathbuf[MAXPATHLEN]; + uint64_t vdev; + nvlist_t *file; + + if (ashift == 0) + ashift = ztest_get_ashift(); + + if (path == NULL) { + path = pathbuf; + + if (aux != NULL) { + vdev = ztest_shared->zs_vdev_aux; + (void) sprintf(path, ztest_aux_template, + zopt_dir, zopt_pool, aux, vdev); + } else { + vdev = ztest_shared->zs_vdev_primaries++; + (void) sprintf(path, ztest_dev_template, + zopt_dir, zopt_pool, vdev); + } + } + + if (size != 0) { + int fd = open(path, O_RDWR | O_CREAT | O_TRUNC, 0666); + if (fd == -1) + fatal(1, "can't open %s", path); + if (ftruncate(fd, size) != 0) + fatal(1, "can't ftruncate %s", path); + (void) close(fd); + } + + VERIFY(nvlist_alloc(&file, NV_UNIQUE_NAME, 0) == 0); + VERIFY(nvlist_add_string(file, ZPOOL_CONFIG_TYPE, VDEV_TYPE_FILE) == 0); + VERIFY(nvlist_add_string(file, ZPOOL_CONFIG_PATH, path) == 0); + VERIFY(nvlist_add_uint64(file, ZPOOL_CONFIG_ASHIFT, ashift) == 0); + + return (file); +} + +static nvlist_t * +make_vdev_raidz(char *path, char *aux, size_t size, uint64_t ashift, int r) +{ + nvlist_t *raidz, **child; + int c; + + if (r < 2) + return (make_vdev_file(path, aux, size, ashift)); + child = umem_alloc(r * sizeof (nvlist_t *), UMEM_NOFAIL); + + for (c = 0; c < r; c++) + child[c] = make_vdev_file(path, aux, size, ashift); + + VERIFY(nvlist_alloc(&raidz, NV_UNIQUE_NAME, 0) == 0); + VERIFY(nvlist_add_string(raidz, ZPOOL_CONFIG_TYPE, + VDEV_TYPE_RAIDZ) == 0); + VERIFY(nvlist_add_uint64(raidz, ZPOOL_CONFIG_NPARITY, + zopt_raidz_parity) == 0); + VERIFY(nvlist_add_nvlist_array(raidz, ZPOOL_CONFIG_CHILDREN, + child, r) == 0); + + for (c = 0; c < r; c++) + nvlist_free(child[c]); + + umem_free(child, r * sizeof (nvlist_t *)); + + return (raidz); +} + +static nvlist_t * +make_vdev_mirror(char *path, char *aux, size_t size, uint64_t ashift, + int r, int m) +{ + nvlist_t *mirror, **child; + int c; + + if (m < 1) + return (make_vdev_raidz(path, aux, size, ashift, r)); + + child = umem_alloc(m * sizeof (nvlist_t *), UMEM_NOFAIL); + + for (c = 0; c < m; c++) + child[c] = make_vdev_raidz(path, aux, size, ashift, r); + + VERIFY(nvlist_alloc(&mirror, NV_UNIQUE_NAME, 0) == 0); + VERIFY(nvlist_add_string(mirror, ZPOOL_CONFIG_TYPE, + VDEV_TYPE_MIRROR) == 0); + VERIFY(nvlist_add_nvlist_array(mirror, ZPOOL_CONFIG_CHILDREN, + child, m) == 0); + + for (c = 0; c < m; c++) + nvlist_free(child[c]); + + umem_free(child, m * sizeof (nvlist_t *)); + + return (mirror); +} + +static nvlist_t * +make_vdev_root(char *path, char *aux, size_t size, uint64_t ashift, + int log, int r, int m, int t) +{ + nvlist_t *root, **child; + int c; + + ASSERT(t > 0); + + child = umem_alloc(t * sizeof (nvlist_t *), UMEM_NOFAIL); + + for (c = 0; c < t; c++) { + child[c] = make_vdev_mirror(path, aux, size, ashift, r, m); + VERIFY(nvlist_add_uint64(child[c], ZPOOL_CONFIG_IS_LOG, + log) == 0); + } + + VERIFY(nvlist_alloc(&root, NV_UNIQUE_NAME, 0) == 0); + VERIFY(nvlist_add_string(root, ZPOOL_CONFIG_TYPE, VDEV_TYPE_ROOT) == 0); + VERIFY(nvlist_add_nvlist_array(root, aux ? aux : ZPOOL_CONFIG_CHILDREN, + child, t) == 0); + + for (c = 0; c < t; c++) + nvlist_free(child[c]); + + umem_free(child, t * sizeof (nvlist_t *)); + + return (root); +} + +static void +ztest_set_random_blocksize(objset_t *os, uint64_t object, dmu_tx_t *tx) +{ + int bs = SPA_MINBLOCKSHIFT + + ztest_random(SPA_MAXBLOCKSHIFT - SPA_MINBLOCKSHIFT + 1); + int ibs = DN_MIN_INDBLKSHIFT + + ztest_random(DN_MAX_INDBLKSHIFT - DN_MIN_INDBLKSHIFT + 1); + int error; + + error = dmu_object_set_blocksize(os, object, 1ULL << bs, ibs, tx); + if (error) { + char osname[300]; + dmu_objset_name(os, osname); + fatal(0, "dmu_object_set_blocksize('%s', %llu, %d, %d) = %d", + osname, object, 1 << bs, ibs, error); + } +} + +static uint8_t +ztest_random_checksum(void) +{ + uint8_t checksum; + + do { + checksum = ztest_random(ZIO_CHECKSUM_FUNCTIONS); + } while (zio_checksum_table[checksum].ci_zbt); + + if (checksum == ZIO_CHECKSUM_OFF) + checksum = ZIO_CHECKSUM_ON; + + return (checksum); +} + +static uint8_t +ztest_random_compress(void) +{ + return ((uint8_t)ztest_random(ZIO_COMPRESS_FUNCTIONS)); +} + +typedef struct ztest_replay { + objset_t *zr_os; + uint64_t zr_assign; +} ztest_replay_t; + +static int +ztest_replay_create(ztest_replay_t *zr, lr_create_t *lr, boolean_t byteswap) +{ + objset_t *os = zr->zr_os; + dmu_tx_t *tx; + int error; + + if (byteswap) + byteswap_uint64_array(lr, sizeof (*lr)); + + tx = dmu_tx_create(os); + dmu_tx_hold_bonus(tx, DMU_NEW_OBJECT); + error = dmu_tx_assign(tx, zr->zr_assign); + if (error) { + dmu_tx_abort(tx); + return (error); + } + + error = dmu_object_claim(os, lr->lr_doid, lr->lr_mode, 0, + DMU_OT_NONE, 0, tx); + ASSERT3U(error, ==, 0); + dmu_tx_commit(tx); + + if (zopt_verbose >= 5) { + char osname[MAXNAMELEN]; + dmu_objset_name(os, osname); + (void) printf("replay create of %s object %llu" + " in txg %llu = %d\n", + osname, (u_longlong_t)lr->lr_doid, + (u_longlong_t)zr->zr_assign, error); + } + + return (error); +} + +static int +ztest_replay_remove(ztest_replay_t *zr, lr_remove_t *lr, boolean_t byteswap) +{ + objset_t *os = zr->zr_os; + dmu_tx_t *tx; + int error; + + if (byteswap) + byteswap_uint64_array(lr, sizeof (*lr)); + + tx = dmu_tx_create(os); + dmu_tx_hold_free(tx, lr->lr_doid, 0, DMU_OBJECT_END); + error = dmu_tx_assign(tx, zr->zr_assign); + if (error) { + dmu_tx_abort(tx); + return (error); + } + + error = dmu_object_free(os, lr->lr_doid, tx); + dmu_tx_commit(tx); + + return (error); +} + +zil_replay_func_t *ztest_replay_vector[TX_MAX_TYPE] = { + NULL, /* 0 no such transaction type */ + ztest_replay_create, /* TX_CREATE */ + NULL, /* TX_MKDIR */ + NULL, /* TX_MKXATTR */ + NULL, /* TX_SYMLINK */ + ztest_replay_remove, /* TX_REMOVE */ + NULL, /* TX_RMDIR */ + NULL, /* TX_LINK */ + NULL, /* TX_RENAME */ + NULL, /* TX_WRITE */ + NULL, /* TX_TRUNCATE */ + NULL, /* TX_SETATTR */ + NULL, /* TX_ACL */ +}; + +/* + * Verify that we can't destroy an active pool, create an existing pool, + * or create a pool with a bad vdev spec. + */ +void +ztest_spa_create_destroy(ztest_args_t *za) +{ + int error; + spa_t *spa; + nvlist_t *nvroot; + + /* + * Attempt to create using a bad file. + */ + nvroot = make_vdev_root("/dev/bogus", NULL, 0, 0, 0, 0, 0, 1); + error = spa_create("ztest_bad_file", nvroot, NULL, NULL, NULL); + nvlist_free(nvroot); + if (error != ENOENT) + fatal(0, "spa_create(bad_file) = %d", error); + + /* + * Attempt to create using a bad mirror. + */ + nvroot = make_vdev_root("/dev/bogus", NULL, 0, 0, 0, 0, 2, 1); + error = spa_create("ztest_bad_mirror", nvroot, NULL, NULL, NULL); + nvlist_free(nvroot); + if (error != ENOENT) + fatal(0, "spa_create(bad_mirror) = %d", error); + + /* + * Attempt to create an existing pool. It shouldn't matter + * what's in the nvroot; we should fail with EEXIST. + */ + (void) pthread_rwlock_rdlock(&ztest_shared->zs_name_lock); + nvroot = make_vdev_root("/dev/bogus", NULL, 0, 0, 0, 0, 0, 1); + error = spa_create(za->za_pool, nvroot, NULL, NULL, NULL); + nvlist_free(nvroot); + if (error != EEXIST) + fatal(0, "spa_create(whatever) = %d", error); + + error = spa_open(za->za_pool, &spa, FTAG); + if (error) + fatal(0, "spa_open() = %d", error); + + error = spa_destroy(za->za_pool); + if (error != EBUSY) + fatal(0, "spa_destroy() = %d", error); + + spa_close(spa, FTAG); + (void) pthread_rwlock_unlock(&ztest_shared->zs_name_lock); +} + +static vdev_t * +vdev_lookup_by_path(vdev_t *vd, const char *path) +{ + vdev_t *mvd; + + if (vd->vdev_path != NULL && strcmp(path, vd->vdev_path) == 0) + return (vd); + + for (int c = 0; c < vd->vdev_children; c++) + if ((mvd = vdev_lookup_by_path(vd->vdev_child[c], path)) != + NULL) + return (mvd); + + return (NULL); +} + +/* + * Verify that vdev_add() works as expected. + */ +void +ztest_vdev_add_remove(ztest_args_t *za) +{ + spa_t *spa = za->za_spa; + uint64_t leaves = MAX(zopt_mirrors, 1) * zopt_raidz; + nvlist_t *nvroot; + int error; + + (void) pthread_mutex_lock(&ztest_shared->zs_vdev_lock); + + spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); + + ztest_shared->zs_vdev_primaries = + spa->spa_root_vdev->vdev_children * leaves; + + spa_config_exit(spa, SCL_VDEV, FTAG); + + /* + * Make 1/4 of the devices be log devices. + */ + nvroot = make_vdev_root(NULL, NULL, zopt_vdev_size, 0, + ztest_random(4) == 0, zopt_raidz, zopt_mirrors, 1); + + error = spa_vdev_add(spa, nvroot); + nvlist_free(nvroot); + + (void) pthread_mutex_unlock(&ztest_shared->zs_vdev_lock); + + if (error == ENOSPC) + ztest_record_enospc("spa_vdev_add"); + else if (error != 0) + fatal(0, "spa_vdev_add() = %d", error); +} + +/* + * Verify that adding/removing aux devices (l2arc, hot spare) works as expected. + */ +void +ztest_vdev_aux_add_remove(ztest_args_t *za) +{ + spa_t *spa = za->za_spa; + vdev_t *rvd = spa->spa_root_vdev; + spa_aux_vdev_t *sav; + char *aux; + uint64_t guid = 0; + int error; + + if (ztest_random(2) == 0) { + sav = &spa->spa_spares; + aux = ZPOOL_CONFIG_SPARES; + } else { + sav = &spa->spa_l2cache; + aux = ZPOOL_CONFIG_L2CACHE; + } + + (void) pthread_mutex_lock(&ztest_shared->zs_vdev_lock); + + spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); + + if (sav->sav_count != 0 && ztest_random(4) == 0) { + /* + * Pick a random device to remove. + */ + guid = sav->sav_vdevs[ztest_random(sav->sav_count)]->vdev_guid; + } else { + /* + * Find an unused device we can add. + */ + ztest_shared->zs_vdev_aux = 0; + for (;;) { + char path[MAXPATHLEN]; + int c; + (void) sprintf(path, ztest_aux_template, zopt_dir, + zopt_pool, aux, ztest_shared->zs_vdev_aux); + for (c = 0; c < sav->sav_count; c++) + if (strcmp(sav->sav_vdevs[c]->vdev_path, + path) == 0) + break; + if (c == sav->sav_count && + vdev_lookup_by_path(rvd, path) == NULL) + break; + ztest_shared->zs_vdev_aux++; + } + } + + spa_config_exit(spa, SCL_VDEV, FTAG); + + if (guid == 0) { + /* + * Add a new device. + */ + nvlist_t *nvroot = make_vdev_root(NULL, aux, + (zopt_vdev_size * 5) / 4, 0, 0, 0, 0, 1); + error = spa_vdev_add(spa, nvroot); + if (error != 0) + fatal(0, "spa_vdev_add(%p) = %d", nvroot, error); + nvlist_free(nvroot); + } else { + /* + * Remove an existing device. Sometimes, dirty its + * vdev state first to make sure we handle removal + * of devices that have pending state changes. + */ + if (ztest_random(2) == 0) + (void) vdev_online(spa, guid, B_FALSE, NULL); + + error = spa_vdev_remove(spa, guid, B_FALSE); + if (error != 0 && error != EBUSY) + fatal(0, "spa_vdev_remove(%llu) = %d", guid, error); + } + + (void) pthread_mutex_unlock(&ztest_shared->zs_vdev_lock); +} + +/* + * Verify that we can attach and detach devices. + */ +void +ztest_vdev_attach_detach(ztest_args_t *za) +{ + spa_t *spa = za->za_spa; + spa_aux_vdev_t *sav = &spa->spa_spares; + vdev_t *rvd = spa->spa_root_vdev; + vdev_t *oldvd, *newvd, *pvd; + nvlist_t *root; + uint64_t leaves = MAX(zopt_mirrors, 1) * zopt_raidz; + uint64_t leaf, top; + uint64_t ashift = ztest_get_ashift(); + uint64_t oldguid; + size_t oldsize, newsize; + char oldpath[MAXPATHLEN], newpath[MAXPATHLEN]; + int replacing; + int oldvd_has_siblings = B_FALSE; + int newvd_is_spare = B_FALSE; + int oldvd_is_log; + int error, expected_error; + + (void) mutex_lock(&ztest_shared->zs_vdev_lock); + + spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); + + /* + * Decide whether to do an attach or a replace. + */ + replacing = ztest_random(2); + + /* + * Pick a random top-level vdev. + */ + top = ztest_random(rvd->vdev_children); + + /* + * Pick a random leaf within it. + */ + leaf = ztest_random(leaves); + + /* + * Locate this vdev. + */ + oldvd = rvd->vdev_child[top]; + if (zopt_mirrors >= 1) + oldvd = oldvd->vdev_child[leaf / zopt_raidz]; + if (zopt_raidz > 1) + oldvd = oldvd->vdev_child[leaf % zopt_raidz]; + + /* + * If we're already doing an attach or replace, oldvd may be a + * mirror vdev -- in which case, pick a random child. + */ + while (oldvd->vdev_children != 0) { + oldvd_has_siblings = B_TRUE; + ASSERT(oldvd->vdev_children == 2); + oldvd = oldvd->vdev_child[ztest_random(2)]; + } + + oldguid = oldvd->vdev_guid; + oldsize = vdev_get_rsize(oldvd); + oldvd_is_log = oldvd->vdev_top->vdev_islog; + (void) strcpy(oldpath, oldvd->vdev_path); + pvd = oldvd->vdev_parent; + + /* + * If oldvd has siblings, then half of the time, detach it. + */ + if (oldvd_has_siblings && ztest_random(2) == 0) { + spa_config_exit(spa, SCL_VDEV, FTAG); + error = spa_vdev_detach(spa, oldguid, B_FALSE); + if (error != 0 && error != ENODEV && error != EBUSY) + fatal(0, "detach (%s) returned %d", + oldpath, error); + (void) mutex_unlock(&ztest_shared->zs_vdev_lock); + return; + } + + /* + * For the new vdev, choose with equal probability between the two + * standard paths (ending in either 'a' or 'b') or a random hot spare. + */ + if (sav->sav_count != 0 && ztest_random(3) == 0) { + newvd = sav->sav_vdevs[ztest_random(sav->sav_count)]; + newvd_is_spare = B_TRUE; + (void) strcpy(newpath, newvd->vdev_path); + } else { + (void) snprintf(newpath, sizeof (newpath), ztest_dev_template, + zopt_dir, zopt_pool, top * leaves + leaf); + if (ztest_random(2) == 0) + newpath[strlen(newpath) - 1] = 'b'; + newvd = vdev_lookup_by_path(rvd, newpath); + } + + if (newvd) { + newsize = vdev_get_rsize(newvd); + } else { + /* + * Make newsize a little bigger or smaller than oldsize. + * If it's smaller, the attach should fail. + * If it's larger, and we're doing a replace, + * we should get dynamic LUN growth when we're done. + */ + newsize = 10 * oldsize / (9 + ztest_random(3)); + } + + /* + * If pvd is not a mirror or root, the attach should fail with ENOTSUP, + * unless it's a replace; in that case any non-replacing parent is OK. + * + * If newvd is already part of the pool, it should fail with EBUSY. + * + * If newvd is too small, it should fail with EOVERFLOW. + */ + if (pvd->vdev_ops != &vdev_mirror_ops && + pvd->vdev_ops != &vdev_root_ops && (!replacing || + pvd->vdev_ops == &vdev_replacing_ops || + pvd->vdev_ops == &vdev_spare_ops)) + expected_error = ENOTSUP; + else if (newvd_is_spare && (!replacing || oldvd_is_log)) + expected_error = ENOTSUP; + else if (newvd == oldvd) + expected_error = replacing ? 0 : EBUSY; + else if (vdev_lookup_by_path(rvd, newpath) != NULL) + expected_error = EBUSY; + else if (newsize < oldsize) + expected_error = EOVERFLOW; + else if (ashift > oldvd->vdev_top->vdev_ashift) + expected_error = EDOM; + else + expected_error = 0; + + spa_config_exit(spa, SCL_VDEV, FTAG); + + /* + * Build the nvlist describing newpath. + */ + root = make_vdev_root(newpath, NULL, newvd == NULL ? newsize : 0, + ashift, 0, 0, 0, 1); + + error = spa_vdev_attach(spa, oldguid, root, replacing); + + nvlist_free(root); + + /* + * If our parent was the replacing vdev, but the replace completed, + * then instead of failing with ENOTSUP we may either succeed, + * fail with ENODEV, or fail with EOVERFLOW. + */ + if (expected_error == ENOTSUP && + (error == 0 || error == ENODEV || error == EOVERFLOW)) + expected_error = error; + + /* + * If someone grew the LUN, the replacement may be too small. + */ + if (error == EOVERFLOW || error == EBUSY) + expected_error = error; + + /* XXX workaround 6690467 */ + if (error != expected_error && expected_error != EBUSY) { + fatal(0, "attach (%s %llu, %s %llu, %d) " + "returned %d, expected %d", + oldpath, (longlong_t)oldsize, newpath, + (longlong_t)newsize, replacing, error, expected_error); + } + + (void) mutex_unlock(&ztest_shared->zs_vdev_lock); +} + +/* + * Verify that dynamic LUN growth works as expected. + */ +/* ARGSUSED */ +void +ztest_vdev_LUN_growth(ztest_args_t *za) +{ + spa_t *spa = za->za_spa; + char dev_name[MAXPATHLEN]; + uint64_t leaves = MAX(zopt_mirrors, 1) * zopt_raidz; + uint64_t vdev; + size_t fsize; + int fd; + + (void) pthread_mutex_lock(&ztest_shared->zs_vdev_lock); + + /* + * Pick a random leaf vdev. + */ + spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); + vdev = ztest_random(spa->spa_root_vdev->vdev_children * leaves); + spa_config_exit(spa, SCL_VDEV, FTAG); + + (void) sprintf(dev_name, ztest_dev_template, zopt_dir, zopt_pool, vdev); + + if ((fd = open(dev_name, O_RDWR)) != -1) { + /* + * Determine the size. + */ + fsize = lseek(fd, 0, SEEK_END); + + /* + * If it's less than 2x the original size, grow by around 3%. + */ + if (fsize < 2 * zopt_vdev_size) { + size_t newsize = fsize + ztest_random(fsize / 32); + (void) ftruncate(fd, newsize); + if (zopt_verbose >= 6) { + (void) printf("%s grew from %lu to %lu bytes\n", + dev_name, (ulong_t)fsize, (ulong_t)newsize); + } + } + (void) close(fd); + } + + (void) pthread_mutex_unlock(&ztest_shared->zs_vdev_lock); +} + +/* ARGSUSED */ +static void +ztest_create_cb(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx) +{ + /* + * Create the directory object. + */ + VERIFY(dmu_object_claim(os, ZTEST_DIROBJ, + DMU_OT_UINT64_OTHER, ZTEST_DIROBJ_BLOCKSIZE, + DMU_OT_UINT64_OTHER, 5 * sizeof (ztest_block_tag_t), tx) == 0); + + VERIFY(zap_create_claim(os, ZTEST_MICROZAP_OBJ, + DMU_OT_ZAP_OTHER, DMU_OT_NONE, 0, tx) == 0); + + VERIFY(zap_create_claim(os, ZTEST_FATZAP_OBJ, + DMU_OT_ZAP_OTHER, DMU_OT_NONE, 0, tx) == 0); +} + +static int +ztest_destroy_cb(char *name, void *arg) +{ + ztest_args_t *za = arg; + objset_t *os; + dmu_object_info_t *doi = &za->za_doi; + int error; + + /* + * Verify that the dataset contains a directory object. + */ + error = dmu_objset_open(name, DMU_OST_OTHER, + DS_MODE_USER | DS_MODE_READONLY, &os); + ASSERT3U(error, ==, 0); + error = dmu_object_info(os, ZTEST_DIROBJ, doi); + if (error != ENOENT) { + /* We could have crashed in the middle of destroying it */ + ASSERT3U(error, ==, 0); + ASSERT3U(doi->doi_type, ==, DMU_OT_UINT64_OTHER); + ASSERT3S(doi->doi_physical_blks, >=, 0); + } + dmu_objset_close(os); + + /* + * Destroy the dataset. + */ + error = dmu_objset_destroy(name); + if (error) { + (void) dmu_objset_open(name, DMU_OST_OTHER, + DS_MODE_USER | DS_MODE_READONLY, &os); + fatal(0, "dmu_objset_destroy(os=%p) = %d\n", &os, error); + } + return (0); +} + +/* + * Verify that dmu_objset_{create,destroy,open,close} work as expected. + */ +static uint64_t +ztest_log_create(zilog_t *zilog, dmu_tx_t *tx, uint64_t object, int mode) +{ + itx_t *itx; + lr_create_t *lr; + size_t namesize; + char name[24]; + + (void) sprintf(name, "ZOBJ_%llu", (u_longlong_t)object); + namesize = strlen(name) + 1; + + itx = zil_itx_create(TX_CREATE, sizeof (*lr) + namesize + + ztest_random(ZIL_MAX_BLKSZ)); + lr = (lr_create_t *)&itx->itx_lr; + bzero(lr + 1, lr->lr_common.lrc_reclen - sizeof (*lr)); + lr->lr_doid = object; + lr->lr_foid = 0; + lr->lr_mode = mode; + lr->lr_uid = 0; + lr->lr_gid = 0; + lr->lr_gen = dmu_tx_get_txg(tx); + lr->lr_crtime[0] = time(NULL); + lr->lr_crtime[1] = 0; + lr->lr_rdev = 0; + bcopy(name, (char *)(lr + 1), namesize); + + return (zil_itx_assign(zilog, itx, tx)); +} + +void +ztest_dmu_objset_create_destroy(ztest_args_t *za) +{ + int error; + objset_t *os, *os2; + char name[100]; + int basemode, expected_error; + zilog_t *zilog; + uint64_t seq; + uint64_t objects; + ztest_replay_t zr; + + (void) pthread_rwlock_rdlock(&ztest_shared->zs_name_lock); + (void) snprintf(name, 100, "%s/%s_temp_%llu", za->za_pool, za->za_pool, + (u_longlong_t)za->za_instance); + + basemode = DS_MODE_TYPE(za->za_instance); + if (basemode != DS_MODE_USER && basemode != DS_MODE_OWNER) + basemode = DS_MODE_USER; + + /* + * If this dataset exists from a previous run, process its replay log + * half of the time. If we don't replay it, then dmu_objset_destroy() + * (invoked from ztest_destroy_cb() below) should just throw it away. + */ + if (ztest_random(2) == 0 && + dmu_objset_open(name, DMU_OST_OTHER, DS_MODE_OWNER, &os) == 0) { + zr.zr_os = os; + zil_replay(os, &zr, &zr.zr_assign, ztest_replay_vector, NULL); + dmu_objset_close(os); + } + + /* + * There may be an old instance of the dataset we're about to + * create lying around from a previous run. If so, destroy it + * and all of its snapshots. + */ + (void) dmu_objset_find(name, ztest_destroy_cb, za, + DS_FIND_CHILDREN | DS_FIND_SNAPSHOTS); + + /* + * Verify that the destroyed dataset is no longer in the namespace. + */ + error = dmu_objset_open(name, DMU_OST_OTHER, basemode, &os); + if (error != ENOENT) + fatal(1, "dmu_objset_open(%s) found destroyed dataset %p", + name, os); + + /* + * Verify that we can create a new dataset. + */ + error = dmu_objset_create(name, DMU_OST_OTHER, NULL, 0, + ztest_create_cb, NULL); + if (error) { + if (error == ENOSPC) { + ztest_record_enospc("dmu_objset_create"); + (void) pthread_rwlock_unlock(&ztest_shared->zs_name_lock); + return; + } + fatal(0, "dmu_objset_create(%s) = %d", name, error); + } + + error = dmu_objset_open(name, DMU_OST_OTHER, basemode, &os); + if (error) { + fatal(0, "dmu_objset_open(%s) = %d", name, error); + } + + /* + * Open the intent log for it. + */ + zilog = zil_open(os, NULL); + + /* + * Put a random number of objects in there. + */ + objects = ztest_random(20); + seq = 0; + while (objects-- != 0) { + uint64_t object; + dmu_tx_t *tx = dmu_tx_create(os); + dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, sizeof (name)); + error = dmu_tx_assign(tx, TXG_WAIT); + if (error) { + dmu_tx_abort(tx); + } else { + object = dmu_object_alloc(os, DMU_OT_UINT64_OTHER, 0, + DMU_OT_NONE, 0, tx); + ztest_set_random_blocksize(os, object, tx); + seq = ztest_log_create(zilog, tx, object, + DMU_OT_UINT64_OTHER); + dmu_write(os, object, 0, sizeof (name), name, tx); + dmu_tx_commit(tx); + } + if (ztest_random(5) == 0) { + zil_commit(zilog, seq, object); + } + if (ztest_random(100) == 0) { + error = zil_suspend(zilog); + if (error == 0) { + zil_resume(zilog); + } + } + } + + /* + * Verify that we cannot create an existing dataset. + */ + error = dmu_objset_create(name, DMU_OST_OTHER, NULL, 0, NULL, NULL); + if (error != EEXIST) + fatal(0, "created existing dataset, error = %d", error); + + /* + * Verify that multiple dataset holds are allowed, but only when + * the new access mode is compatible with the base mode. + */ + if (basemode == DS_MODE_OWNER) { + error = dmu_objset_open(name, DMU_OST_OTHER, DS_MODE_USER, + &os2); + if (error) + fatal(0, "dmu_objset_open('%s') = %d", name, error); + else + dmu_objset_close(os2); + } + error = dmu_objset_open(name, DMU_OST_OTHER, DS_MODE_OWNER, &os2); + expected_error = (basemode == DS_MODE_OWNER) ? EBUSY : 0; + if (error != expected_error) + fatal(0, "dmu_objset_open('%s') = %d, expected %d", + name, error, expected_error); + if (error == 0) + dmu_objset_close(os2); + + zil_close(zilog); + dmu_objset_close(os); + + error = dmu_objset_destroy(name); + if (error) + fatal(0, "dmu_objset_destroy(%s) = %d", name, error); + + (void) pthread_rwlock_unlock(&ztest_shared->zs_name_lock); +} + +/* + * Verify that dmu_snapshot_{create,destroy,open,close} work as expected. + */ +void +ztest_dmu_snapshot_create_destroy(ztest_args_t *za) +{ + int error; + objset_t *os = za->za_os; + char snapname[100]; + char osname[MAXNAMELEN]; + + (void) pthread_rwlock_rdlock(&ztest_shared->zs_name_lock); + dmu_objset_name(os, osname); + (void) snprintf(snapname, 100, "%s@%llu", osname, + (u_longlong_t)za->za_instance); + + error = dmu_objset_destroy(snapname); + if (error != 0 && error != ENOENT) + fatal(0, "dmu_objset_destroy() = %d", error); + error = dmu_objset_snapshot(osname, strchr(snapname, '@')+1, FALSE); + if (error == ENOSPC) + ztest_record_enospc("dmu_take_snapshot"); + else if (error != 0 && error != EEXIST) + fatal(0, "dmu_take_snapshot() = %d", error); + (void) pthread_rwlock_unlock(&ztest_shared->zs_name_lock); +} + +/* + * Verify that dmu_object_{alloc,free} work as expected. + */ +void +ztest_dmu_object_alloc_free(ztest_args_t *za) +{ + objset_t *os = za->za_os; + dmu_buf_t *db; + dmu_tx_t *tx; + uint64_t batchobj, object, batchsize, endoff, temp; + int b, c, error, bonuslen; + dmu_object_info_t *doi = &za->za_doi; + char osname[MAXNAMELEN]; + + dmu_objset_name(os, osname); + + endoff = -8ULL; + batchsize = 2; + + /* + * Create a batch object if necessary, and record it in the directory. + */ + VERIFY3U(0, ==, dmu_read(os, ZTEST_DIROBJ, za->za_diroff, + sizeof (uint64_t), &batchobj)); + if (batchobj == 0) { + tx = dmu_tx_create(os); + dmu_tx_hold_write(tx, ZTEST_DIROBJ, za->za_diroff, + sizeof (uint64_t)); + dmu_tx_hold_bonus(tx, DMU_NEW_OBJECT); + error = dmu_tx_assign(tx, TXG_WAIT); + if (error) { + ztest_record_enospc("create a batch object"); + dmu_tx_abort(tx); + return; + } + batchobj = dmu_object_alloc(os, DMU_OT_UINT64_OTHER, 0, + DMU_OT_NONE, 0, tx); + ztest_set_random_blocksize(os, batchobj, tx); + dmu_write(os, ZTEST_DIROBJ, za->za_diroff, + sizeof (uint64_t), &batchobj, tx); + dmu_tx_commit(tx); + } + + /* + * Destroy the previous batch of objects. + */ + for (b = 0; b < batchsize; b++) { + VERIFY3U(0, ==, dmu_read(os, batchobj, b * sizeof (uint64_t), + sizeof (uint64_t), &object)); + if (object == 0) + continue; + /* + * Read and validate contents. + * We expect the nth byte of the bonus buffer to be n. + */ + VERIFY(0 == dmu_bonus_hold(os, object, FTAG, &db)); + za->za_dbuf = db; + + dmu_object_info_from_db(db, doi); + ASSERT(doi->doi_type == DMU_OT_UINT64_OTHER); + ASSERT(doi->doi_bonus_type == DMU_OT_PLAIN_OTHER); + ASSERT3S(doi->doi_physical_blks, >=, 0); + + bonuslen = doi->doi_bonus_size; + + for (c = 0; c < bonuslen; c++) { + if (((uint8_t *)db->db_data)[c] != + (uint8_t)(c + bonuslen)) { + fatal(0, + "bad bonus: %s, obj %llu, off %d: %u != %u", + osname, object, c, + ((uint8_t *)db->db_data)[c], + (uint8_t)(c + bonuslen)); + } + } + + dmu_buf_rele(db, FTAG); + za->za_dbuf = NULL; + + /* + * We expect the word at endoff to be our object number. + */ + VERIFY(0 == dmu_read(os, object, endoff, + sizeof (uint64_t), &temp)); + + if (temp != object) { + fatal(0, "bad data in %s, got %llu, expected %llu", + osname, temp, object); + } + + /* + * Destroy old object and clear batch entry. + */ + tx = dmu_tx_create(os); + dmu_tx_hold_write(tx, batchobj, + b * sizeof (uint64_t), sizeof (uint64_t)); + dmu_tx_hold_free(tx, object, 0, DMU_OBJECT_END); + error = dmu_tx_assign(tx, TXG_WAIT); + if (error) { + ztest_record_enospc("free object"); + dmu_tx_abort(tx); + return; + } + error = dmu_object_free(os, object, tx); + if (error) { + fatal(0, "dmu_object_free('%s', %llu) = %d", + osname, object, error); + } + object = 0; + + dmu_object_set_checksum(os, batchobj, + ztest_random_checksum(), tx); + dmu_object_set_compress(os, batchobj, + ztest_random_compress(), tx); + + dmu_write(os, batchobj, b * sizeof (uint64_t), + sizeof (uint64_t), &object, tx); + + dmu_tx_commit(tx); + } + + /* + * Before creating the new batch of objects, generate a bunch of churn. + */ + for (b = ztest_random(100); b > 0; b--) { + tx = dmu_tx_create(os); + dmu_tx_hold_bonus(tx, DMU_NEW_OBJECT); + error = dmu_tx_assign(tx, TXG_WAIT); + if (error) { + ztest_record_enospc("churn objects"); + dmu_tx_abort(tx); + return; + } + object = dmu_object_alloc(os, DMU_OT_UINT64_OTHER, 0, + DMU_OT_NONE, 0, tx); + ztest_set_random_blocksize(os, object, tx); + error = dmu_object_free(os, object, tx); + if (error) { + fatal(0, "dmu_object_free('%s', %llu) = %d", + osname, object, error); + } + dmu_tx_commit(tx); + } + + /* + * Create a new batch of objects with randomly chosen + * blocksizes and record them in the batch directory. + */ + for (b = 0; b < batchsize; b++) { + uint32_t va_blksize; + u_longlong_t va_nblocks; + + tx = dmu_tx_create(os); + dmu_tx_hold_write(tx, batchobj, b * sizeof (uint64_t), + sizeof (uint64_t)); + dmu_tx_hold_bonus(tx, DMU_NEW_OBJECT); + dmu_tx_hold_write(tx, DMU_NEW_OBJECT, endoff, + sizeof (uint64_t)); + error = dmu_tx_assign(tx, TXG_WAIT); + if (error) { + ztest_record_enospc("create batchobj"); + dmu_tx_abort(tx); + return; + } + bonuslen = (int)ztest_random(dmu_bonus_max()) + 1; + + object = dmu_object_alloc(os, DMU_OT_UINT64_OTHER, 0, + DMU_OT_PLAIN_OTHER, bonuslen, tx); + + ztest_set_random_blocksize(os, object, tx); + + dmu_object_set_checksum(os, object, + ztest_random_checksum(), tx); + dmu_object_set_compress(os, object, + ztest_random_compress(), tx); + + dmu_write(os, batchobj, b * sizeof (uint64_t), + sizeof (uint64_t), &object, tx); + + /* + * Write to both the bonus buffer and the regular data. + */ + VERIFY(dmu_bonus_hold(os, object, FTAG, &db) == 0); + za->za_dbuf = db; + ASSERT3U(bonuslen, <=, db->db_size); + + dmu_object_size_from_db(db, &va_blksize, &va_nblocks); + ASSERT3S(va_nblocks, >=, 0); + + dmu_buf_will_dirty(db, tx); + + /* + * See comments above regarding the contents of + * the bonus buffer and the word at endoff. + */ + for (c = 0; c < bonuslen; c++) + ((uint8_t *)db->db_data)[c] = (uint8_t)(c + bonuslen); + + dmu_buf_rele(db, FTAG); + za->za_dbuf = NULL; + + /* + * Write to a large offset to increase indirection. + */ + dmu_write(os, object, endoff, sizeof (uint64_t), &object, tx); + + dmu_tx_commit(tx); + } +} + +/* + * Verify that dmu_{read,write} work as expected. + */ +typedef struct bufwad { + uint64_t bw_index; + uint64_t bw_txg; + uint64_t bw_data; +} bufwad_t; + +typedef struct dmu_read_write_dir { + uint64_t dd_packobj; + uint64_t dd_bigobj; + uint64_t dd_chunk; +} dmu_read_write_dir_t; + +void +ztest_dmu_read_write(ztest_args_t *za) +{ + objset_t *os = za->za_os; + dmu_read_write_dir_t dd; + dmu_tx_t *tx; + int i, freeit, error; + uint64_t n, s, txg; + bufwad_t *packbuf, *bigbuf, *pack, *bigH, *bigT; + uint64_t packoff, packsize, bigoff, bigsize; + uint64_t regions = 997; + uint64_t stride = 123456789ULL; + uint64_t width = 40; + int free_percent = 5; + + /* + * This test uses two objects, packobj and bigobj, that are always + * updated together (i.e. in the same tx) so that their contents are + * in sync and can be compared. Their contents relate to each other + * in a simple way: packobj is a dense array of 'bufwad' structures, + * while bigobj is a sparse array of the same bufwads. Specifically, + * for any index n, there are three bufwads that should be identical: + * + * packobj, at offset n * sizeof (bufwad_t) + * bigobj, at the head of the nth chunk + * bigobj, at the tail of the nth chunk + * + * The chunk size is arbitrary. It doesn't have to be a power of two, + * and it doesn't have any relation to the object blocksize. + * The only requirement is that it can hold at least two bufwads. + * + * Normally, we write the bufwad to each of these locations. + * However, free_percent of the time we instead write zeroes to + * packobj and perform a dmu_free_range() on bigobj. By comparing + * bigobj to packobj, we can verify that the DMU is correctly + * tracking which parts of an object are allocated and free, + * and that the contents of the allocated blocks are correct. + */ + + /* + * Read the directory info. If it's the first time, set things up. + */ + VERIFY(0 == dmu_read(os, ZTEST_DIROBJ, za->za_diroff, + sizeof (dd), &dd)); + if (dd.dd_chunk == 0) { + ASSERT(dd.dd_packobj == 0); + ASSERT(dd.dd_bigobj == 0); + tx = dmu_tx_create(os); + dmu_tx_hold_write(tx, ZTEST_DIROBJ, za->za_diroff, sizeof (dd)); + dmu_tx_hold_bonus(tx, DMU_NEW_OBJECT); + error = dmu_tx_assign(tx, TXG_WAIT); + if (error) { + ztest_record_enospc("create r/w directory"); + dmu_tx_abort(tx); + return; + } + + dd.dd_packobj = dmu_object_alloc(os, DMU_OT_UINT64_OTHER, 0, + DMU_OT_NONE, 0, tx); + dd.dd_bigobj = dmu_object_alloc(os, DMU_OT_UINT64_OTHER, 0, + DMU_OT_NONE, 0, tx); + dd.dd_chunk = (1000 + ztest_random(1000)) * sizeof (uint64_t); + + ztest_set_random_blocksize(os, dd.dd_packobj, tx); + ztest_set_random_blocksize(os, dd.dd_bigobj, tx); + + dmu_write(os, ZTEST_DIROBJ, za->za_diroff, sizeof (dd), &dd, + tx); + dmu_tx_commit(tx); + } + + /* + * Prefetch a random chunk of the big object. + * Our aim here is to get some async reads in flight + * for blocks that we may free below; the DMU should + * handle this race correctly. + */ + n = ztest_random(regions) * stride + ztest_random(width); + s = 1 + ztest_random(2 * width - 1); + dmu_prefetch(os, dd.dd_bigobj, n * dd.dd_chunk, s * dd.dd_chunk); + + /* + * Pick a random index and compute the offsets into packobj and bigobj. + */ + n = ztest_random(regions) * stride + ztest_random(width); + s = 1 + ztest_random(width - 1); + + packoff = n * sizeof (bufwad_t); + packsize = s * sizeof (bufwad_t); + + bigoff = n * dd.dd_chunk; + bigsize = s * dd.dd_chunk; + + packbuf = umem_alloc(packsize, UMEM_NOFAIL); + bigbuf = umem_alloc(bigsize, UMEM_NOFAIL); + + /* + * free_percent of the time, free a range of bigobj rather than + * overwriting it. + */ + freeit = (ztest_random(100) < free_percent); + + /* + * Read the current contents of our objects. + */ + error = dmu_read(os, dd.dd_packobj, packoff, packsize, packbuf); + ASSERT3U(error, ==, 0); + error = dmu_read(os, dd.dd_bigobj, bigoff, bigsize, bigbuf); + ASSERT3U(error, ==, 0); + + /* + * Get a tx for the mods to both packobj and bigobj. + */ + tx = dmu_tx_create(os); + + dmu_tx_hold_write(tx, dd.dd_packobj, packoff, packsize); + + if (freeit) + dmu_tx_hold_free(tx, dd.dd_bigobj, bigoff, bigsize); + else + dmu_tx_hold_write(tx, dd.dd_bigobj, bigoff, bigsize); + + error = dmu_tx_assign(tx, TXG_WAIT); + + if (error) { + ztest_record_enospc("dmu r/w range"); + dmu_tx_abort(tx); + umem_free(packbuf, packsize); + umem_free(bigbuf, bigsize); + return; + } + + txg = dmu_tx_get_txg(tx); + + /* + * For each index from n to n + s, verify that the existing bufwad + * in packobj matches the bufwads at the head and tail of the + * corresponding chunk in bigobj. Then update all three bufwads + * with the new values we want to write out. + */ + for (i = 0; i < s; i++) { + /* LINTED */ + pack = (bufwad_t *)((char *)packbuf + i * sizeof (bufwad_t)); + /* LINTED */ + bigH = (bufwad_t *)((char *)bigbuf + i * dd.dd_chunk); + /* LINTED */ + bigT = (bufwad_t *)((char *)bigH + dd.dd_chunk) - 1; + + ASSERT((uintptr_t)bigH - (uintptr_t)bigbuf < bigsize); + ASSERT((uintptr_t)bigT - (uintptr_t)bigbuf < bigsize); + + if (pack->bw_txg > txg) + fatal(0, "future leak: got %llx, open txg is %llx", + pack->bw_txg, txg); + + if (pack->bw_data != 0 && pack->bw_index != n + i) + fatal(0, "wrong index: got %llx, wanted %llx+%llx", + pack->bw_index, n, i); + + if (bcmp(pack, bigH, sizeof (bufwad_t)) != 0) + fatal(0, "pack/bigH mismatch in %p/%p", pack, bigH); + + if (bcmp(pack, bigT, sizeof (bufwad_t)) != 0) + fatal(0, "pack/bigT mismatch in %p/%p", pack, bigT); + + if (freeit) { + bzero(pack, sizeof (bufwad_t)); + } else { + pack->bw_index = n + i; + pack->bw_txg = txg; + pack->bw_data = 1 + ztest_random(-2ULL); + } + *bigH = *pack; + *bigT = *pack; + } + + /* + * We've verified all the old bufwads, and made new ones. + * Now write them out. + */ + dmu_write(os, dd.dd_packobj, packoff, packsize, packbuf, tx); + + if (freeit) { + if (zopt_verbose >= 6) { + (void) printf("freeing offset %llx size %llx" + " txg %llx\n", + (u_longlong_t)bigoff, + (u_longlong_t)bigsize, + (u_longlong_t)txg); + } + VERIFY(0 == dmu_free_range(os, dd.dd_bigobj, bigoff, + bigsize, tx)); + } else { + if (zopt_verbose >= 6) { + (void) printf("writing offset %llx size %llx" + " txg %llx\n", + (u_longlong_t)bigoff, + (u_longlong_t)bigsize, + (u_longlong_t)txg); + } + dmu_write(os, dd.dd_bigobj, bigoff, bigsize, bigbuf, tx); + } + + dmu_tx_commit(tx); + + /* + * Sanity check the stuff we just wrote. + */ + { + void *packcheck = umem_alloc(packsize, UMEM_NOFAIL); + void *bigcheck = umem_alloc(bigsize, UMEM_NOFAIL); + + VERIFY(0 == dmu_read(os, dd.dd_packobj, packoff, + packsize, packcheck)); + VERIFY(0 == dmu_read(os, dd.dd_bigobj, bigoff, + bigsize, bigcheck)); + + ASSERT(bcmp(packbuf, packcheck, packsize) == 0); + ASSERT(bcmp(bigbuf, bigcheck, bigsize) == 0); + + umem_free(packcheck, packsize); + umem_free(bigcheck, bigsize); + } + + umem_free(packbuf, packsize); + umem_free(bigbuf, bigsize); +} + +void +ztest_dmu_check_future_leak(ztest_args_t *za) +{ + objset_t *os = za->za_os; + dmu_buf_t *db; + ztest_block_tag_t *bt; + dmu_object_info_t *doi = &za->za_doi; + + /* + * Make sure that, if there is a write record in the bonus buffer + * of the ZTEST_DIROBJ, that the txg for this record is <= the + * last synced txg of the pool. + */ + VERIFY(dmu_bonus_hold(os, ZTEST_DIROBJ, FTAG, &db) == 0); + za->za_dbuf = db; + VERIFY(dmu_object_info(os, ZTEST_DIROBJ, doi) == 0); + ASSERT3U(doi->doi_bonus_size, >=, sizeof (*bt)); + ASSERT3U(doi->doi_bonus_size, <=, db->db_size); + ASSERT3U(doi->doi_bonus_size % sizeof (*bt), ==, 0); + bt = (void *)((char *)db->db_data + doi->doi_bonus_size - sizeof (*bt)); + if (bt->bt_objset != 0) { + ASSERT3U(bt->bt_objset, ==, dmu_objset_id(os)); + ASSERT3U(bt->bt_object, ==, ZTEST_DIROBJ); + ASSERT3U(bt->bt_offset, ==, -1ULL); + ASSERT3U(bt->bt_txg, <, spa_first_txg(za->za_spa)); + } + dmu_buf_rele(db, FTAG); + za->za_dbuf = NULL; +} + +void +ztest_dmu_write_parallel(ztest_args_t *za) +{ + objset_t *os = za->za_os; + ztest_block_tag_t *rbt = &za->za_rbt; + ztest_block_tag_t *wbt = &za->za_wbt; + const size_t btsize = sizeof (ztest_block_tag_t); + dmu_buf_t *db; + int b, error; + int bs = ZTEST_DIROBJ_BLOCKSIZE; + int do_free = 0; + uint64_t off, txg, txg_how; + pthread_mutex_t *lp; + char osname[MAXNAMELEN]; + char iobuf[SPA_MAXBLOCKSIZE]; + blkptr_t blk = { 0 }; + uint64_t blkoff; + zbookmark_t zb; + dmu_tx_t *tx = dmu_tx_create(os); + + dmu_objset_name(os, osname); + + /* + * Have multiple threads write to large offsets in ZTEST_DIROBJ + * to verify that having multiple threads writing to the same object + * in parallel doesn't cause any trouble. + */ + if (ztest_random(4) == 0) { + /* + * Do the bonus buffer instead of a regular block. + * We need a lock to serialize resize vs. others, + * so we hash on the objset ID. + */ + b = dmu_objset_id(os) % ZTEST_SYNC_LOCKS; + off = -1ULL; + dmu_tx_hold_bonus(tx, ZTEST_DIROBJ); + } else { + b = ztest_random(ZTEST_SYNC_LOCKS); + off = za->za_diroff_shared + (b << SPA_MAXBLOCKSHIFT); + if (ztest_random(4) == 0) { + do_free = 1; + dmu_tx_hold_free(tx, ZTEST_DIROBJ, off, bs); + } else { + dmu_tx_hold_write(tx, ZTEST_DIROBJ, off, bs); + } + } + + txg_how = ztest_random(2) == 0 ? TXG_WAIT : TXG_NOWAIT; + error = dmu_tx_assign(tx, txg_how); + if (error) { + if (error == ERESTART) { + ASSERT(txg_how == TXG_NOWAIT); + dmu_tx_wait(tx); + } else { + ztest_record_enospc("dmu write parallel"); + } + dmu_tx_abort(tx); + return; + } + txg = dmu_tx_get_txg(tx); + + lp = &ztest_shared->zs_sync_lock[b]; + (void) pthread_mutex_lock(lp); + + wbt->bt_objset = dmu_objset_id(os); + wbt->bt_object = ZTEST_DIROBJ; + wbt->bt_offset = off; + wbt->bt_txg = txg; + wbt->bt_thread = za->za_instance; + wbt->bt_seq = ztest_shared->zs_seq[b]++; /* protected by lp */ + + /* + * Occasionally, write an all-zero block to test the behavior + * of blocks that compress into holes. + */ + if (off != -1ULL && ztest_random(8) == 0) + bzero(wbt, btsize); + + if (off == -1ULL) { + dmu_object_info_t *doi = &za->za_doi; + char *dboff; + + VERIFY(dmu_bonus_hold(os, ZTEST_DIROBJ, FTAG, &db) == 0); + za->za_dbuf = db; + dmu_object_info_from_db(db, doi); + ASSERT3U(doi->doi_bonus_size, <=, db->db_size); + ASSERT3U(doi->doi_bonus_size, >=, btsize); + ASSERT3U(doi->doi_bonus_size % btsize, ==, 0); + dboff = (char *)db->db_data + doi->doi_bonus_size - btsize; + bcopy(dboff, rbt, btsize); + if (rbt->bt_objset != 0) { + ASSERT3U(rbt->bt_objset, ==, wbt->bt_objset); + ASSERT3U(rbt->bt_object, ==, wbt->bt_object); + ASSERT3U(rbt->bt_offset, ==, wbt->bt_offset); + ASSERT3U(rbt->bt_txg, <=, wbt->bt_txg); + } + if (ztest_random(10) == 0) { + int newsize = (ztest_random(db->db_size / + btsize) + 1) * btsize; + + ASSERT3U(newsize, >=, btsize); + ASSERT3U(newsize, <=, db->db_size); + VERIFY3U(dmu_set_bonus(db, newsize, tx), ==, 0); + dboff = (char *)db->db_data + newsize - btsize; + } + dmu_buf_will_dirty(db, tx); + bcopy(wbt, dboff, btsize); + dmu_buf_rele(db, FTAG); + za->za_dbuf = NULL; + } else if (do_free) { + VERIFY(dmu_free_range(os, ZTEST_DIROBJ, off, bs, tx) == 0); + } else { + dmu_write(os, ZTEST_DIROBJ, off, btsize, wbt, tx); + } + + (void) pthread_mutex_unlock(lp); + + if (ztest_random(1000) == 0) + (void) poll(NULL, 0, 1); /* open dn_notxholds window */ + + dmu_tx_commit(tx); + + if (ztest_random(10000) == 0) + txg_wait_synced(dmu_objset_pool(os), txg); + + if (off == -1ULL || do_free) + return; + + if (ztest_random(2) != 0) + return; + + /* + * dmu_sync() the block we just wrote. + */ + (void) pthread_mutex_lock(lp); + + blkoff = P2ALIGN_TYPED(off, bs, uint64_t); + error = dmu_buf_hold(os, ZTEST_DIROBJ, blkoff, FTAG, &db); + za->za_dbuf = db; + if (error) { + dprintf("dmu_buf_hold(%s, %d, %llx) = %d\n", + osname, ZTEST_DIROBJ, blkoff, error); + (void) pthread_mutex_unlock(lp); + return; + } + blkoff = off - blkoff; + error = dmu_sync(NULL, db, &blk, txg, NULL, NULL); + dmu_buf_rele(db, FTAG); + za->za_dbuf = NULL; + + (void) pthread_mutex_unlock(lp); + + if (error) { + dprintf("dmu_sync(%s, %d, %llx) = %d\n", + osname, ZTEST_DIROBJ, off, error); + return; + } + + if (blk.blk_birth == 0) /* concurrent free */ + return; + + txg_suspend(dmu_objset_pool(os)); + + ASSERT(blk.blk_fill == 1); + ASSERT3U(BP_GET_TYPE(&blk), ==, DMU_OT_UINT64_OTHER); + ASSERT3U(BP_GET_LEVEL(&blk), ==, 0); + ASSERT3U(BP_GET_LSIZE(&blk), ==, bs); + + /* + * Read the block that dmu_sync() returned to make sure its contents + * match what we wrote. We do this while still txg_suspend()ed + * to ensure that the block can't be reused before we read it. + */ + zb.zb_objset = dmu_objset_id(os); + zb.zb_object = ZTEST_DIROBJ; + zb.zb_level = 0; + zb.zb_blkid = off / bs; + error = zio_wait(zio_read(NULL, za->za_spa, &blk, iobuf, bs, + NULL, NULL, ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_MUSTSUCCEED, &zb)); + ASSERT3U(error, ==, 0); + + txg_resume(dmu_objset_pool(os)); + + bcopy(&iobuf[blkoff], rbt, btsize); + + if (rbt->bt_objset == 0) /* concurrent free */ + return; + + if (wbt->bt_objset == 0) /* all-zero overwrite */ + return; + + ASSERT3U(rbt->bt_objset, ==, wbt->bt_objset); + ASSERT3U(rbt->bt_object, ==, wbt->bt_object); + ASSERT3U(rbt->bt_offset, ==, wbt->bt_offset); + + /* + * The semantic of dmu_sync() is that we always push the most recent + * version of the data, so in the face of concurrent updates we may + * see a newer version of the block. That's OK. + */ + ASSERT3U(rbt->bt_txg, >=, wbt->bt_txg); + if (rbt->bt_thread == wbt->bt_thread) + ASSERT3U(rbt->bt_seq, ==, wbt->bt_seq); + else + ASSERT3U(rbt->bt_seq, >, wbt->bt_seq); +} + +/* + * Verify that zap_{create,destroy,add,remove,update} work as expected. + */ +#define ZTEST_ZAP_MIN_INTS 1 +#define ZTEST_ZAP_MAX_INTS 4 +#define ZTEST_ZAP_MAX_PROPS 1000 + +void +ztest_zap(ztest_args_t *za) +{ + objset_t *os = za->za_os; + uint64_t object; + uint64_t txg, last_txg; + uint64_t value[ZTEST_ZAP_MAX_INTS]; + uint64_t zl_ints, zl_intsize, prop; + int i, ints; + dmu_tx_t *tx; + char propname[100], txgname[100]; + int error; + char osname[MAXNAMELEN]; + char *hc[2] = { "s.acl.h", ".s.open.h.hyLZlg" }; + + dmu_objset_name(os, osname); + + /* + * Create a new object if necessary, and record it in the directory. + */ + VERIFY(0 == dmu_read(os, ZTEST_DIROBJ, za->za_diroff, + sizeof (uint64_t), &object)); + + if (object == 0) { + tx = dmu_tx_create(os); + dmu_tx_hold_write(tx, ZTEST_DIROBJ, za->za_diroff, + sizeof (uint64_t)); + dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, TRUE, NULL); + error = dmu_tx_assign(tx, TXG_WAIT); + if (error) { + ztest_record_enospc("create zap test obj"); + dmu_tx_abort(tx); + return; + } + object = zap_create(os, DMU_OT_ZAP_OTHER, DMU_OT_NONE, 0, tx); + if (error) { + fatal(0, "zap_create('%s', %llu) = %d", + osname, object, error); + } + ASSERT(object != 0); + dmu_write(os, ZTEST_DIROBJ, za->za_diroff, + sizeof (uint64_t), &object, tx); + /* + * Generate a known hash collision, and verify that + * we can lookup and remove both entries. + */ + for (i = 0; i < 2; i++) { + value[i] = i; + error = zap_add(os, object, hc[i], sizeof (uint64_t), + 1, &value[i], tx); + ASSERT3U(error, ==, 0); + } + for (i = 0; i < 2; i++) { + error = zap_add(os, object, hc[i], sizeof (uint64_t), + 1, &value[i], tx); + ASSERT3U(error, ==, EEXIST); + error = zap_length(os, object, hc[i], + &zl_intsize, &zl_ints); + ASSERT3U(error, ==, 0); + ASSERT3U(zl_intsize, ==, sizeof (uint64_t)); + ASSERT3U(zl_ints, ==, 1); + } + for (i = 0; i < 2; i++) { + error = zap_remove(os, object, hc[i], tx); + ASSERT3U(error, ==, 0); + } + + dmu_tx_commit(tx); + } + + ints = MAX(ZTEST_ZAP_MIN_INTS, object % ZTEST_ZAP_MAX_INTS); + + prop = ztest_random(ZTEST_ZAP_MAX_PROPS); + (void) sprintf(propname, "prop_%llu", (u_longlong_t)prop); + (void) sprintf(txgname, "txg_%llu", (u_longlong_t)prop); + bzero(value, sizeof (value)); + last_txg = 0; + + /* + * If these zap entries already exist, validate their contents. + */ + error = zap_length(os, object, txgname, &zl_intsize, &zl_ints); + if (error == 0) { + ASSERT3U(zl_intsize, ==, sizeof (uint64_t)); + ASSERT3U(zl_ints, ==, 1); + + VERIFY(zap_lookup(os, object, txgname, zl_intsize, + zl_ints, &last_txg) == 0); + + VERIFY(zap_length(os, object, propname, &zl_intsize, + &zl_ints) == 0); + + ASSERT3U(zl_intsize, ==, sizeof (uint64_t)); + ASSERT3U(zl_ints, ==, ints); + + VERIFY(zap_lookup(os, object, propname, zl_intsize, + zl_ints, value) == 0); + + for (i = 0; i < ints; i++) { + ASSERT3U(value[i], ==, last_txg + object + i); + } + } else { + ASSERT3U(error, ==, ENOENT); + } + + /* + * Atomically update two entries in our zap object. + * The first is named txg_%llu, and contains the txg + * in which the property was last updated. The second + * is named prop_%llu, and the nth element of its value + * should be txg + object + n. + */ + tx = dmu_tx_create(os); + dmu_tx_hold_zap(tx, object, TRUE, NULL); + error = dmu_tx_assign(tx, TXG_WAIT); + if (error) { + ztest_record_enospc("create zap entry"); + dmu_tx_abort(tx); + return; + } + txg = dmu_tx_get_txg(tx); + + if (last_txg > txg) + fatal(0, "zap future leak: old %llu new %llu", last_txg, txg); + + for (i = 0; i < ints; i++) + value[i] = txg + object + i; + + error = zap_update(os, object, txgname, sizeof (uint64_t), 1, &txg, tx); + if (error) + fatal(0, "zap_update('%s', %llu, '%s') = %d", + osname, object, txgname, error); + + error = zap_update(os, object, propname, sizeof (uint64_t), + ints, value, tx); + if (error) + fatal(0, "zap_update('%s', %llu, '%s') = %d", + osname, object, propname, error); + + dmu_tx_commit(tx); + + /* + * Remove a random pair of entries. + */ + prop = ztest_random(ZTEST_ZAP_MAX_PROPS); + (void) sprintf(propname, "prop_%llu", (u_longlong_t)prop); + (void) sprintf(txgname, "txg_%llu", (u_longlong_t)prop); + + error = zap_length(os, object, txgname, &zl_intsize, &zl_ints); + + if (error == ENOENT) + return; + + ASSERT3U(error, ==, 0); + + tx = dmu_tx_create(os); + dmu_tx_hold_zap(tx, object, TRUE, NULL); + error = dmu_tx_assign(tx, TXG_WAIT); + if (error) { + ztest_record_enospc("remove zap entry"); + dmu_tx_abort(tx); + return; + } + error = zap_remove(os, object, txgname, tx); + if (error) + fatal(0, "zap_remove('%s', %llu, '%s') = %d", + osname, object, txgname, error); + + error = zap_remove(os, object, propname, tx); + if (error) + fatal(0, "zap_remove('%s', %llu, '%s') = %d", + osname, object, propname, error); + + dmu_tx_commit(tx); + + /* + * Once in a while, destroy the object. + */ + if (ztest_random(1000) != 0) + return; + + tx = dmu_tx_create(os); + dmu_tx_hold_write(tx, ZTEST_DIROBJ, za->za_diroff, sizeof (uint64_t)); + dmu_tx_hold_free(tx, object, 0, DMU_OBJECT_END); + error = dmu_tx_assign(tx, TXG_WAIT); + if (error) { + ztest_record_enospc("destroy zap object"); + dmu_tx_abort(tx); + return; + } + error = zap_destroy(os, object, tx); + if (error) + fatal(0, "zap_destroy('%s', %llu) = %d", + osname, object, error); + object = 0; + dmu_write(os, ZTEST_DIROBJ, za->za_diroff, sizeof (uint64_t), + &object, tx); + dmu_tx_commit(tx); +} + +void +ztest_zap_parallel(ztest_args_t *za) +{ + objset_t *os = za->za_os; + uint64_t txg, object, count, wsize, wc, zl_wsize, zl_wc; + dmu_tx_t *tx; + int i, namelen, error; + char name[20], string_value[20]; + void *data; + + /* + * Generate a random name of the form 'xxx.....' where each + * x is a random printable character and the dots are dots. + * There are 94 such characters, and the name length goes from + * 6 to 20, so there are 94^3 * 15 = 12,458,760 possible names. + */ + namelen = ztest_random(sizeof (name) - 5) + 5 + 1; + + for (i = 0; i < 3; i++) + name[i] = '!' + ztest_random('~' - '!' + 1); + for (; i < namelen - 1; i++) + name[i] = '.'; + name[i] = '\0'; + + if (ztest_random(2) == 0) + object = ZTEST_MICROZAP_OBJ; + else + object = ZTEST_FATZAP_OBJ; + + if ((namelen & 1) || object == ZTEST_MICROZAP_OBJ) { + wsize = sizeof (txg); + wc = 1; + data = &txg; + } else { + wsize = 1; + wc = namelen; + data = string_value; + } + + count = -1ULL; + VERIFY(zap_count(os, object, &count) == 0); + ASSERT(count != -1ULL); + + /* + * Select an operation: length, lookup, add, update, remove. + */ + i = ztest_random(5); + + if (i >= 2) { + tx = dmu_tx_create(os); + dmu_tx_hold_zap(tx, object, TRUE, NULL); + error = dmu_tx_assign(tx, TXG_WAIT); + if (error) { + ztest_record_enospc("zap parallel"); + dmu_tx_abort(tx); + return; + } + txg = dmu_tx_get_txg(tx); + bcopy(name, string_value, namelen); + } else { + tx = NULL; + txg = 0; + bzero(string_value, namelen); + } + + switch (i) { + + case 0: + error = zap_length(os, object, name, &zl_wsize, &zl_wc); + if (error == 0) { + ASSERT3U(wsize, ==, zl_wsize); + ASSERT3U(wc, ==, zl_wc); + } else { + ASSERT3U(error, ==, ENOENT); + } + break; + + case 1: + error = zap_lookup(os, object, name, wsize, wc, data); + if (error == 0) { + if (data == string_value && + bcmp(name, data, namelen) != 0) + fatal(0, "name '%s' != val '%s' len %d", + name, data, namelen); + } else { + ASSERT3U(error, ==, ENOENT); + } + break; + + case 2: + error = zap_add(os, object, name, wsize, wc, data, tx); + ASSERT(error == 0 || error == EEXIST); + break; + + case 3: + VERIFY(zap_update(os, object, name, wsize, wc, data, tx) == 0); + break; + + case 4: + error = zap_remove(os, object, name, tx); + ASSERT(error == 0 || error == ENOENT); + break; + } + + if (tx != NULL) + dmu_tx_commit(tx); +} + +void +ztest_dsl_prop_get_set(ztest_args_t *za) +{ + objset_t *os = za->za_os; + int i, inherit; + uint64_t value; + const char *prop, *valname; + char setpoint[MAXPATHLEN]; + char osname[MAXNAMELEN]; + int error; + + (void) pthread_rwlock_rdlock(&ztest_shared->zs_name_lock); + + dmu_objset_name(os, osname); + + for (i = 0; i < 2; i++) { + if (i == 0) { + prop = "checksum"; + value = ztest_random_checksum(); + inherit = (value == ZIO_CHECKSUM_INHERIT); + } else { + prop = "compression"; + value = ztest_random_compress(); + inherit = (value == ZIO_COMPRESS_INHERIT); + } + + error = dsl_prop_set(osname, prop, sizeof (value), + !inherit, &value); + + if (error == ENOSPC) { + ztest_record_enospc("dsl_prop_set"); + break; + } + + ASSERT3U(error, ==, 0); + + VERIFY3U(dsl_prop_get(osname, prop, sizeof (value), + 1, &value, setpoint), ==, 0); + + if (i == 0) + valname = zio_checksum_table[value].ci_name; + else + valname = zio_compress_table[value].ci_name; + + if (zopt_verbose >= 6) { + (void) printf("%s %s = %s for '%s'\n", + osname, prop, valname, setpoint); + } + } + + (void) pthread_rwlock_unlock(&ztest_shared->zs_name_lock); +} + +/* + * Inject random faults into the on-disk data. + */ +void +ztest_fault_inject(ztest_args_t *za) +{ + int fd; + uint64_t offset; + uint64_t leaves = MAX(zopt_mirrors, 1) * zopt_raidz; + uint64_t bad = 0x1990c0ffeedecade; + uint64_t top, leaf; + char path0[MAXPATHLEN]; + char pathrand[MAXPATHLEN]; + size_t fsize; + spa_t *spa = za->za_spa; + int bshift = SPA_MAXBLOCKSHIFT + 2; /* don't scrog all labels */ + int iters = 1000; + int maxfaults = zopt_maxfaults; + vdev_t *vd0 = NULL; + uint64_t guid0 = 0; + + ASSERT(leaves >= 1); + + /* + * We need SCL_STATE here because we're going to look at vd0->vdev_tsd. + */ + spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); + + if (ztest_random(2) == 0) { + /* + * Inject errors on a normal data device. + */ + top = ztest_random(spa->spa_root_vdev->vdev_children); + leaf = ztest_random(leaves); + + /* + * Generate paths to the first leaf in this top-level vdev, + * and to the random leaf we selected. We'll induce transient + * write failures and random online/offline activity on leaf 0, + * and we'll write random garbage to the randomly chosen leaf. + */ + (void) snprintf(path0, sizeof (path0), ztest_dev_template, + zopt_dir, zopt_pool, top * leaves + 0); + (void) snprintf(pathrand, sizeof (pathrand), ztest_dev_template, + zopt_dir, zopt_pool, top * leaves + leaf); + + vd0 = vdev_lookup_by_path(spa->spa_root_vdev, path0); + if (vd0 != NULL && maxfaults != 1) { + /* + * Make vd0 explicitly claim to be unreadable, + * or unwriteable, or reach behind its back + * and close the underlying fd. We can do this if + * maxfaults == 0 because we'll fail and reexecute, + * and we can do it if maxfaults >= 2 because we'll + * have enough redundancy. If maxfaults == 1, the + * combination of this with injection of random data + * corruption below exceeds the pool's fault tolerance. + */ + vdev_file_t *vf = vd0->vdev_tsd; + + if (vf != NULL && ztest_random(3) == 0) { + (void) close(vf->vf_vnode->v_fd); + vf->vf_vnode->v_fd = -1; + } else if (ztest_random(2) == 0) { + vd0->vdev_cant_read = B_TRUE; + } else { + vd0->vdev_cant_write = B_TRUE; + } + guid0 = vd0->vdev_guid; + } + } else { + /* + * Inject errors on an l2cache device. + */ + spa_aux_vdev_t *sav = &spa->spa_l2cache; + + if (sav->sav_count == 0) { + spa_config_exit(spa, SCL_STATE, FTAG); + return; + } + vd0 = sav->sav_vdevs[ztest_random(sav->sav_count)]; + guid0 = vd0->vdev_guid; + (void) strcpy(path0, vd0->vdev_path); + (void) strcpy(pathrand, vd0->vdev_path); + + leaf = 0; + leaves = 1; + maxfaults = INT_MAX; /* no limit on cache devices */ + } + + dprintf("damaging %s and %s\n", path0, pathrand); + + spa_config_exit(spa, SCL_STATE, FTAG); + + if (maxfaults == 0) + return; + + /* + * If we can tolerate two or more faults, randomly online/offline vd0. + */ + if (maxfaults >= 2 && guid0 != 0) { + if (ztest_random(10) < 6) + (void) vdev_offline(spa, guid0, B_TRUE); + else + (void) vdev_online(spa, guid0, B_FALSE, NULL); + } + + /* + * We have at least single-fault tolerance, so inject data corruption. + */ + fd = open(pathrand, O_RDWR); + + if (fd == -1) /* we hit a gap in the device namespace */ + return; + + fsize = lseek(fd, 0, SEEK_END); + + while (--iters != 0) { + offset = ztest_random(fsize / (leaves << bshift)) * + (leaves << bshift) + (leaf << bshift) + + (ztest_random(1ULL << (bshift - 1)) & -8ULL); + + if (offset >= fsize) + continue; + + if (zopt_verbose >= 6) + (void) printf("injecting bad word into %s," + " offset 0x%llx\n", pathrand, (u_longlong_t)offset); + + if (pwrite(fd, &bad, sizeof (bad), offset) != sizeof (bad)) + fatal(1, "can't inject bad word at 0x%llx in %s", + offset, pathrand); + } + + (void) close(fd); +} + +/* + * Scrub the pool. + */ +void +ztest_scrub(ztest_args_t *za) +{ + spa_t *spa = za->za_spa; + + (void) spa_scrub(spa, POOL_SCRUB_EVERYTHING); + (void) poll(NULL, 0, 1000); /* wait a second, then force a restart */ + (void) spa_scrub(spa, POOL_SCRUB_EVERYTHING); +} + +/* + * Rename the pool to a different name and then rename it back. + */ +void +ztest_spa_rename(ztest_args_t *za) +{ + char *oldname, *newname; + int error; + spa_t *spa; + + (void) pthread_rwlock_wrlock(&ztest_shared->zs_name_lock); + + oldname = za->za_pool; + newname = umem_alloc(strlen(oldname) + 5, UMEM_NOFAIL); + (void) strcpy(newname, oldname); + (void) strcat(newname, "_tmp"); + + /* + * Do the rename + */ + error = spa_rename(oldname, newname); + if (error) + fatal(0, "spa_rename('%s', '%s') = %d", oldname, + newname, error); + + /* + * Try to open it under the old name, which shouldn't exist + */ + error = spa_open(oldname, &spa, FTAG); + if (error != ENOENT) + fatal(0, "spa_open('%s') = %d", oldname, error); + + /* + * Open it under the new name and make sure it's still the same spa_t. + */ + error = spa_open(newname, &spa, FTAG); + if (error != 0) + fatal(0, "spa_open('%s') = %d", newname, error); + + ASSERT(spa == za->za_spa); + spa_close(spa, FTAG); + + /* + * Rename it back to the original + */ + error = spa_rename(newname, oldname); + if (error) + fatal(0, "spa_rename('%s', '%s') = %d", newname, + oldname, error); + + /* + * Make sure it can still be opened + */ + error = spa_open(oldname, &spa, FTAG); + if (error != 0) + fatal(0, "spa_open('%s') = %d", oldname, error); + + ASSERT(spa == za->za_spa); + spa_close(spa, FTAG); + + umem_free(newname, strlen(newname) + 1); + + (void) pthread_rwlock_unlock(&ztest_shared->zs_name_lock); +} + + +/* + * Completely obliterate one disk. + */ +static void +ztest_obliterate_one_disk(uint64_t vdev) +{ + int fd; + char dev_name[MAXPATHLEN], copy_name[MAXPATHLEN]; + size_t fsize; + + if (zopt_maxfaults < 2) + return; + + (void) sprintf(dev_name, ztest_dev_template, zopt_dir, zopt_pool, vdev); + (void) snprintf(copy_name, MAXPATHLEN, "%s.old", dev_name); + + fd = open(dev_name, O_RDWR); + + if (fd == -1) + fatal(1, "can't open %s", dev_name); + + /* + * Determine the size. + */ + fsize = lseek(fd, 0, SEEK_END); + + (void) close(fd); + + /* + * Rename the old device to dev_name.old (useful for debugging). + */ + VERIFY(rename(dev_name, copy_name) == 0); + + /* + * Create a new one. + */ + VERIFY((fd = open(dev_name, O_RDWR | O_CREAT | O_TRUNC, 0666)) >= 0); + VERIFY(ftruncate(fd, fsize) == 0); + (void) close(fd); +} + +static void +ztest_replace_one_disk(spa_t *spa, uint64_t vdev) +{ + char dev_name[MAXPATHLEN]; + nvlist_t *root; + int error; + uint64_t guid; + vdev_t *vd; + + (void) sprintf(dev_name, ztest_dev_template, zopt_dir, zopt_pool, vdev); + + /* + * Build the nvlist describing dev_name. + */ + root = make_vdev_root(dev_name, NULL, 0, 0, 0, 0, 0, 1); + + spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); + if ((vd = vdev_lookup_by_path(spa->spa_root_vdev, dev_name)) == NULL) + guid = 0; + else + guid = vd->vdev_guid; + spa_config_exit(spa, SCL_VDEV, FTAG); + error = spa_vdev_attach(spa, guid, root, B_TRUE); + if (error != 0 && + error != EBUSY && + error != ENOTSUP && + error != ENODEV && + error != EDOM) + fatal(0, "spa_vdev_attach(in-place) = %d", error); + + nvlist_free(root); +} + +static void +ztest_verify_blocks(char *pool) +{ + int status; + char zdb[MAXPATHLEN + MAXNAMELEN + 20]; + char zbuf[1024]; + char *bin; + char *ztest; + char *isa; + int isalen; + FILE *fp; + + (void) realpath(getexecname(), zdb); + + /* zdb lives in /usr/sbin, while ztest lives in /usr/bin */ + bin = strstr(zdb, "/usr/bin/"); + ztest = strstr(bin, "/ztest"); + isa = bin + 8; + isalen = ztest - isa; + isa = strdup(isa); + /* LINTED */ + (void) sprintf(bin, + "/usr/sbin%.*s/zdb -bc%s%s -U /tmp/zpool.cache %s", + isalen, + isa, + zopt_verbose >= 3 ? "s" : "", + zopt_verbose >= 4 ? "v" : "", + pool); + free(isa); + + if (zopt_verbose >= 5) + (void) printf("Executing %s\n", strstr(zdb, "zdb ")); + + fp = popen(zdb, "r"); + + while (fgets(zbuf, sizeof (zbuf), fp) != NULL) + if (zopt_verbose >= 3) + (void) printf("%s", zbuf); + + status = pclose(fp); + + if (status == 0) + return; + + ztest_dump_core = 0; + if (WIFEXITED(status)) + fatal(0, "'%s' exit code %d", zdb, WEXITSTATUS(status)); + else + fatal(0, "'%s' died with signal %d", zdb, WTERMSIG(status)); +} + +static void +ztest_walk_pool_directory(char *header) +{ + spa_t *spa = NULL; + + if (zopt_verbose >= 6) + (void) printf("%s\n", header); + + mutex_enter(&spa_namespace_lock); + while ((spa = spa_next(spa)) != NULL) + if (zopt_verbose >= 6) + (void) printf("\t%s\n", spa_name(spa)); + mutex_exit(&spa_namespace_lock); +} + +static void +ztest_spa_import_export(char *oldname, char *newname) +{ + nvlist_t *config; + uint64_t pool_guid; + spa_t *spa; + int error; + + if (zopt_verbose >= 4) { + (void) printf("import/export: old = %s, new = %s\n", + oldname, newname); + } + + /* + * Clean up from previous runs. + */ + (void) spa_destroy(newname); + + /* + * Get the pool's configuration and guid. + */ + error = spa_open(oldname, &spa, FTAG); + if (error) + fatal(0, "spa_open('%s') = %d", oldname, error); + + pool_guid = spa_guid(spa); + spa_close(spa, FTAG); + + ztest_walk_pool_directory("pools before export"); + + /* + * Export it. + */ + error = spa_export(oldname, &config, B_FALSE); + if (error) + fatal(0, "spa_export('%s') = %d", oldname, error); + + ztest_walk_pool_directory("pools after export"); + + /* + * Import it under the new name. + */ + error = spa_import(newname, config, NULL); + if (error) + fatal(0, "spa_import('%s') = %d", newname, error); + + ztest_walk_pool_directory("pools after import"); + + /* + * Try to import it again -- should fail with EEXIST. + */ + error = spa_import(newname, config, NULL); + if (error != EEXIST) + fatal(0, "spa_import('%s') twice", newname); + + /* + * Try to import it under a different name -- should fail with EEXIST. + */ + error = spa_import(oldname, config, NULL); + if (error != EEXIST) + fatal(0, "spa_import('%s') under multiple names", newname); + + /* + * Verify that the pool is no longer visible under the old name. + */ + error = spa_open(oldname, &spa, FTAG); + if (error != ENOENT) + fatal(0, "spa_open('%s') = %d", newname, error); + + /* + * Verify that we can open and close the pool using the new name. + */ + error = spa_open(newname, &spa, FTAG); + if (error) + fatal(0, "spa_open('%s') = %d", newname, error); + ASSERT(pool_guid == spa_guid(spa)); + spa_close(spa, FTAG); + + nvlist_free(config); +} + +static void * +ztest_resume(void *arg) +{ + spa_t *spa = arg; + + while (!ztest_exiting) { + (void) poll(NULL, 0, 1000); + + if (!spa_suspended(spa)) + continue; + + spa_vdev_state_enter(spa); + vdev_clear(spa, NULL); + (void) spa_vdev_state_exit(spa, NULL, 0); + + zio_resume(spa); + } + return (NULL); +} + +static void * +ztest_thread(void *arg) +{ + ztest_args_t *za = arg; + ztest_shared_t *zs = ztest_shared; + hrtime_t now, functime; + ztest_info_t *zi; + int f, i; + + while ((now = gethrtime()) < za->za_stop) { + /* + * See if it's time to force a crash. + */ + if (now > za->za_kill) { + zs->zs_alloc = spa_get_alloc(za->za_spa); + zs->zs_space = spa_get_space(za->za_spa); + (void) kill(getpid(), SIGKILL); + } + + /* + * Pick a random function. + */ + f = ztest_random(ZTEST_FUNCS); + zi = &zs->zs_info[f]; + + /* + * Decide whether to call it, based on the requested frequency. + */ + if (zi->zi_call_target == 0 || + (double)zi->zi_call_total / zi->zi_call_target > + (double)(now - zs->zs_start_time) / (zopt_time * NANOSEC)) + continue; + + atomic_add_64(&zi->zi_calls, 1); + atomic_add_64(&zi->zi_call_total, 1); + + za->za_diroff = (za->za_instance * ZTEST_FUNCS + f) * + ZTEST_DIRSIZE; + za->za_diroff_shared = (1ULL << 63); + + for (i = 0; i < zi->zi_iters; i++) + zi->zi_func(za); + + functime = gethrtime() - now; + + atomic_add_64(&zi->zi_call_time, functime); + + if (zopt_verbose >= 4) { + Dl_info dli; + (void) dladdr((void *)zi->zi_func, &dli); + (void) printf("%6.2f sec in %s\n", + (double)functime / NANOSEC, dli.dli_sname); + } + + /* + * If we're getting ENOSPC with some regularity, stop. + */ + if (zs->zs_enospc_count > 10) + break; + } + + return (NULL); +} + +/* + * Kick off threads to run tests on all datasets in parallel. + */ +static void +ztest_run(char *pool) +{ + int t, d, error; + ztest_shared_t *zs = ztest_shared; + ztest_args_t *za; + spa_t *spa; + char name[100]; + pthread_t resume_tid; + + ztest_exiting = B_FALSE; + + (void) pthread_mutex_init(&zs->zs_vdev_lock, NULL); + (void) pthread_rwlock_init(&zs->zs_name_lock, NULL); + + for (t = 0; t < ZTEST_SYNC_LOCKS; t++) + (void) pthread_mutex_init(&zs->zs_sync_lock[t], NULL); + + + /* + * Destroy one disk before we even start. + * It's mirrored, so everything should work just fine. + * This makes us exercise fault handling very early in spa_load(). + */ + ztest_obliterate_one_disk(0); + + /* + * Verify that the sum of the sizes of all blocks in the pool + * equals the SPA's allocated space total. + */ + ztest_verify_blocks(pool); + + /* + * Kick off a replacement of the disk we just obliterated. + */ + kernel_init(FREAD | FWRITE); + VERIFY(spa_open(pool, &spa, FTAG) == 0); + ztest_replace_one_disk(spa, 0); + if (zopt_verbose >= 5) + show_pool_stats(spa); + spa_close(spa, FTAG); + kernel_fini(); + + kernel_init(FREAD | FWRITE); + + /* + * Verify that we can export the pool and reimport it under a + * different name. + */ + if (ztest_random(2) == 0) { + (void) snprintf(name, 100, "%s_import", pool); + ztest_spa_import_export(pool, name); + ztest_spa_import_export(name, pool); + } + + /* + * Verify that we can loop over all pools. + */ + mutex_enter(&spa_namespace_lock); + for (spa = spa_next(NULL); spa != NULL; spa = spa_next(spa)) { + if (zopt_verbose > 3) { + (void) printf("spa_next: found %s\n", spa_name(spa)); + } + } + mutex_exit(&spa_namespace_lock); + + /* + * Open our pool. + */ + VERIFY(spa_open(pool, &spa, FTAG) == 0); + + /* + * Create a thread to periodically resume suspended I/O. + */ + VERIFY(pthread_create(&resume_tid, NULL, ztest_suspend_monitor, + NULL) == 0); + + /* + * Verify that we can safely inquire about about any object, + * whether it's allocated or not. To make it interesting, + * we probe a 5-wide window around each power of two. + * This hits all edge cases, including zero and the max. + */ + for (t = 0; t < 64; t++) { + for (d = -5; d <= 5; d++) { + error = dmu_object_info(spa->spa_meta_objset, + (1ULL << t) + d, NULL); + ASSERT(error == 0 || error == ENOENT || + error == EINVAL); + } + } + + /* + * Now kick off all the tests that run in parallel. + */ + zs->zs_enospc_count = 0; + + za = umem_zalloc(zopt_threads * sizeof (ztest_args_t), UMEM_NOFAIL); + + if (zopt_verbose >= 4) + (void) printf("starting main threads...\n"); + + za[0].za_start = gethrtime(); + za[0].za_stop = za[0].za_start + zopt_passtime * NANOSEC; + za[0].za_stop = MIN(za[0].za_stop, zs->zs_stop_time); + za[0].za_kill = za[0].za_stop; + if (ztest_random(100) < zopt_killrate) + za[0].za_kill -= ztest_random(zopt_passtime * NANOSEC); + + for (t = 0; t < zopt_threads; t++) { + d = t % zopt_datasets; + + (void) strcpy(za[t].za_pool, pool); + za[t].za_os = za[d].za_os; + za[t].za_spa = spa; + za[t].za_zilog = za[d].za_zilog; + za[t].za_instance = t; + za[t].za_random = ztest_random(-1ULL); + za[t].za_start = za[0].za_start; + za[t].za_stop = za[0].za_stop; + za[t].za_kill = za[0].za_kill; + + if (t < zopt_datasets) { + ztest_replay_t zr; + int test_future = FALSE; + (void) pthread_rwlock_rdlock(&ztest_shared->zs_name_lock); + (void) snprintf(name, 100, "%s/%s_%d", pool, pool, d); + error = dmu_objset_create(name, DMU_OST_OTHER, NULL, 0, + ztest_create_cb, NULL); + if (error == EEXIST) { + test_future = TRUE; + } else if (error == ENOSPC) { + zs->zs_enospc_count++; + (void) pthread_rwlock_unlock(&ztest_shared->zs_name_lock); + break; + } else if (error != 0) { + fatal(0, "dmu_objset_create(%s) = %d", + name, error); + } + error = dmu_objset_open(name, DMU_OST_OTHER, + DS_MODE_USER, &za[d].za_os); + if (error) + fatal(0, "dmu_objset_open('%s') = %d", + name, error); + (void) pthread_rwlock_unlock(&ztest_shared->zs_name_lock); + if (test_future) + ztest_dmu_check_future_leak(&za[t]); + zr.zr_os = za[d].za_os; + zil_replay(zr.zr_os, &zr, &zr.zr_assign, + ztest_replay_vector, NULL); + za[d].za_zilog = zil_open(za[d].za_os, NULL); + } + + VERIFY(pthread_create(&za[t].za_thread, NULL, ztest_thread, + &za[t]) == 0); + } + + while (--t >= 0) { + VERIFY(pthread_join(za[t].za_thread, NULL) == 0); + if (t < zopt_datasets) { + zil_close(za[t].za_zilog); + dmu_objset_close(za[t].za_os); + } + } + + if (zopt_verbose >= 3) + show_pool_stats(spa); + + txg_wait_synced(spa_get_dsl(spa), 0); + + zs->zs_alloc = spa_get_alloc(spa); + zs->zs_space = spa_get_space(spa); + + /* + * If we had out-of-space errors, destroy a random objset. + */ + if (zs->zs_enospc_count != 0) { + (void) pthread_rwlock_rdlock(&ztest_shared->zs_name_lock); + d = (int)ztest_random(zopt_datasets); + (void) snprintf(name, 100, "%s/%s_%d", pool, pool, d); + if (zopt_verbose >= 3) + (void) printf("Destroying %s to free up space\n", name); + (void) dmu_objset_find(name, ztest_destroy_cb, &za[d], + DS_FIND_SNAPSHOTS | DS_FIND_CHILDREN); + (void) pthread_rwlock_unlock(&ztest_shared->zs_name_lock); + } + + txg_wait_synced(spa_get_dsl(spa), 0); + + umem_free(za, zopt_threads * sizeof (ztest_args_t)); + + /* Kill the resume thread */ + ztest_exiting = B_TRUE; + VERIFY(pthread_join(resume_tid, NULL) == 0); + + /* + * Right before closing the pool, kick off a bunch of async I/O; + * spa_close() should wait for it to complete. + */ + for (t = 1; t < 50; t++) + dmu_prefetch(spa->spa_meta_objset, t, 0, 1 << 15); + + spa_close(spa, FTAG); + + kernel_fini(); +} + +void +print_time(hrtime_t t, char *timebuf) +{ + hrtime_t s = t / NANOSEC; + hrtime_t m = s / 60; + hrtime_t h = m / 60; + hrtime_t d = h / 24; + + s -= m * 60; + m -= h * 60; + h -= d * 24; + + timebuf[0] = '\0'; + + if (d) + (void) sprintf(timebuf, + "%llud%02lluh%02llum%02llus", d, h, m, s); + else if (h) + (void) sprintf(timebuf, "%lluh%02llum%02llus", h, m, s); + else if (m) + (void) sprintf(timebuf, "%llum%02llus", m, s); + else + (void) sprintf(timebuf, "%llus", s); +} + +/* + * Create a storage pool with the given name and initial vdev size. + * Then create the specified number of datasets in the pool. + */ +static void +ztest_init(char *pool) +{ + spa_t *spa; + int error; + nvlist_t *nvroot; + + kernel_init(FREAD | FWRITE); + + /* + * Create the storage pool. + */ + (void) spa_destroy(pool); + ztest_shared->zs_vdev_primaries = 0; + nvroot = make_vdev_root(NULL, NULL, zopt_vdev_size, 0, + 0, zopt_raidz, zopt_mirrors, 1); + error = spa_create(pool, nvroot, NULL, NULL, NULL); + nvlist_free(nvroot); + + if (error) + fatal(0, "spa_create() = %d", error); + error = spa_open(pool, &spa, FTAG); + if (error) + fatal(0, "spa_open() = %d", error); + + if (zopt_verbose >= 3) + show_pool_stats(spa); + + spa_close(spa, FTAG); + + kernel_fini(); +} + +int +main(int argc, char **argv) +{ + int kills = 0; + int iters = 0; + int i, f; + ztest_shared_t *zs; + ztest_info_t *zi; + char timebuf[100]; + char numbuf[6]; + + (void) setvbuf(stdout, NULL, _IOLBF, 0); + + /* Override location of zpool.cache */ + spa_config_path = "/tmp/zpool.cache"; + + ztest_random_fd = open("/dev/urandom", O_RDONLY); + + process_options(argc, argv); + + argc -= optind; + argv += optind; + + dprintf_setup(&argc, argv); + + /* + * Blow away any existing copy of zpool.cache + */ + if (zopt_init != 0) + (void) remove("/tmp/zpool.cache"); + + zs = ztest_shared = (void *)mmap(0, + P2ROUNDUP(sizeof (ztest_shared_t), getpagesize()), + PROT_READ | PROT_WRITE, MAP_SHARED | MAP_ANON, -1, 0); + + if (zopt_verbose >= 1) { + (void) printf("%llu vdevs, %d datasets, %d threads," + " %llu seconds...\n", + (u_longlong_t)zopt_vdevs, zopt_datasets, zopt_threads, + (u_longlong_t)zopt_time); + } + + /* + * Create and initialize our storage pool. + */ + for (i = 1; i <= zopt_init; i++) { + bzero(zs, sizeof (ztest_shared_t)); + if (zopt_verbose >= 3 && zopt_init != 1) + (void) printf("ztest_init(), pass %d\n", i); + ztest_init(zopt_pool); + } + + /* + * Initialize the call targets for each function. + */ + for (f = 0; f < ZTEST_FUNCS; f++) { + zi = &zs->zs_info[f]; + + *zi = ztest_info[f]; + + if (*zi->zi_interval == 0) + zi->zi_call_target = UINT64_MAX; + else + zi->zi_call_target = zopt_time / *zi->zi_interval; + } + + zs->zs_start_time = gethrtime(); + zs->zs_stop_time = zs->zs_start_time + zopt_time * NANOSEC; + + /* + * Run the tests in a loop. These tests include fault injection + * to verify that self-healing data works, and forced crashes + * to verify that we never lose on-disk consistency. + */ + while (gethrtime() < zs->zs_stop_time) { + int status; + pid_t pid; + char *tmp; + + /* + * Initialize the workload counters for each function. + */ + for (f = 0; f < ZTEST_FUNCS; f++) { + zi = &zs->zs_info[f]; + zi->zi_calls = 0; + zi->zi_call_time = 0; + } + + pid = fork(); + + if (pid == -1) + fatal(1, "fork failed"); + + if (pid == 0) { /* child */ + struct rlimit rl = { 1024, 1024 }; + (void) setrlimit(RLIMIT_NOFILE, &rl); + (void) enable_extended_FILE_stdio(-1, -1); + ztest_run(zopt_pool); + exit(0); + } + + while (waitpid(pid, &status, 0) != pid) + continue; + + if (WIFEXITED(status)) { + if (WEXITSTATUS(status) != 0) { + (void) fprintf(stderr, + "child exited with code %d\n", + WEXITSTATUS(status)); + exit(2); + } + } else if (WIFSIGNALED(status)) { + if (WTERMSIG(status) != SIGKILL) { + (void) fprintf(stderr, + "child died with signal %d\n", + WTERMSIG(status)); + exit(3); + } + kills++; + } else { + (void) fprintf(stderr, "something strange happened " + "to child\n"); + exit(4); + } + + iters++; + + if (zopt_verbose >= 1) { + hrtime_t now = gethrtime(); + + now = MIN(now, zs->zs_stop_time); + print_time(zs->zs_stop_time - now, timebuf); + nicenum(zs->zs_space, numbuf); + + (void) printf("Pass %3d, %8s, %3llu ENOSPC, " + "%4.1f%% of %5s used, %3.0f%% done, %8s to go\n", + iters, + WIFEXITED(status) ? "Complete" : "SIGKILL", + (u_longlong_t)zs->zs_enospc_count, + 100.0 * zs->zs_alloc / zs->zs_space, + numbuf, + 100.0 * (now - zs->zs_start_time) / + (zopt_time * NANOSEC), timebuf); + } + + if (zopt_verbose >= 2) { + (void) printf("\nWorkload summary:\n\n"); + (void) printf("%7s %9s %s\n", + "Calls", "Time", "Function"); + (void) printf("%7s %9s %s\n", + "-----", "----", "--------"); + for (f = 0; f < ZTEST_FUNCS; f++) { + Dl_info dli; + + zi = &zs->zs_info[f]; + print_time(zi->zi_call_time, timebuf); + (void) dladdr((void *)zi->zi_func, &dli); + (void) printf("%7llu %9s %s\n", + (u_longlong_t)zi->zi_calls, timebuf, + dli.dli_sname); + } + (void) printf("\n"); + } + + /* + * It's possible that we killed a child during a rename test, in + * which case we'll have a 'ztest_tmp' pool lying around instead + * of 'ztest'. Do a blind rename in case this happened. + */ + tmp = umem_alloc(strlen(zopt_pool) + 5, UMEM_NOFAIL); + (void) strcpy(tmp, zopt_pool); + (void) strcat(tmp, "_tmp"); + kernel_init(FREAD | FWRITE); + (void) spa_rename(tmp, zopt_pool); + kernel_fini(); + umem_free(tmp, strlen(tmp) + 1); + } + + ztest_verify_blocks(zopt_pool); + + if (zopt_verbose >= 1) { + (void) printf("%d killed, %d completed, %.0f%% kill rate\n", + kills, iters - kills, (100.0 * kills) / MAX(1, iters)); + } + + return (0); +} |