# HG changeset patch # User Matthew Ahrens # Date 1341086447 25200 # Node ID 95aba6e49b9f73304fea0f5d4907827c9d9ca6d4 # Parent b6bbdd77139c7ae473a97abac32c69e7386f39f2 2882 implement libzfs_core 2883 changing "canmount" property to "on" should not always remount dataset 2900 "zfs snapshot" should be able to create multiple, arbitrary snapshots at once Reviewed by: George Wilson Reviewed by: Chris Siden Reviewed by: Garrett D'Amore Reviewed by: Bill Pijewski Reviewed by: Dan Kruchinin Approved by: Eric Schrock diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/Targetdirs --- a/usr/src/Targetdirs Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/Targetdirs Sat Jun 30 13:00:47 2012 -0700 @@ -22,6 +22,7 @@ # Copyright (c) 1989, 2010, Oracle and/or its affiliates. All rights reserved. # Copyright 2011, Richard Lowe # Copyright 2011 Nexenta Systems, Inc. All rights reserved. +# Copyright (c) 2012 by Delphix. All rights reserved. # # @@ -1031,6 +1032,8 @@ $(ROOT)/usr/lib/libxnet.so:= REALPATH=../../lib/libxnet.so.1 $(ROOT)/usr/lib/libzfs.so.1:= REALPATH=../../lib/libzfs.so.1 $(ROOT)/usr/lib/libzfs.so:= REALPATH=../../lib/libzfs.so.1 +$(ROOT)/usr/lib/libzfs_core.so.1:= REALPATH=../../lib/libzfs_core.so.1 +$(ROOT)/usr/lib/libzfs_core.so:= REALPATH=../../lib/libzfs_core.so.1 $(ROOT)/usr/lib/llib-ladm.ln:= REALPATH=../../lib/llib-ladm.ln $(ROOT)/usr/lib/llib-ladm:= REALPATH=../../lib/llib-ladm $(ROOT)/usr/lib/llib-laio.ln:= REALPATH=../../lib/llib-laio.ln @@ -1131,6 +1134,8 @@ $(ROOT)/usr/lib/llib-lxnet:= REALPATH=../../lib/llib-lxnet $(ROOT)/usr/lib/llib-lzfs.ln:= REALPATH=../../lib/llib-lzfs.ln $(ROOT)/usr/lib/llib-lzfs:= REALPATH=../../lib/llib-lzfs +$(ROOT)/usr/lib/llib-lzfs_core.ln:= REALPATH=../../lib/llib-lzfs_core.ln +$(ROOT)/usr/lib/llib-lzfs_core:= REALPATH=../../lib/llib-lzfs_core $(ROOT)/usr/lib/nss_compat.so.1:= REALPATH=../../lib/nss_compat.so.1 $(ROOT)/usr/lib/nss_dns.so.1:= REALPATH=../../lib/nss_dns.so.1 $(ROOT)/usr/lib/nss_files.so.1:= REALPATH=../../lib/nss_files.so.1 @@ -1377,6 +1382,10 @@ REALPATH=../../../lib/$(MACH64)/libzfs.so.1 $(ROOT)/usr/lib/$(MACH64)/libzfs.so.1:= \ REALPATH=../../../lib/$(MACH64)/libzfs.so.1 +$(ROOT)/usr/lib/$(MACH64)/libzfs_core.so:= \ + REALPATH=../../../lib/$(MACH64)/libzfs_core.so.1 +$(ROOT)/usr/lib/$(MACH64)/libzfs_core.so.1:= \ + REALPATH=../../../lib/$(MACH64)/libzfs_core.so.1 $(ROOT)/usr/lib/$(MACH64)/libfdisk.so.1:= \ REALPATH=../../../lib/$(MACH64)/libfdisk.so.1 $(ROOT)/usr/lib/$(MACH64)/libfdisk.so:= \ @@ -1473,6 +1482,8 @@ REALPATH=../../../lib/$(MACH64)/llib-lxnet.ln $(ROOT)/usr/lib/$(MACH64)/llib-lzfs.ln:= \ REALPATH=../../../lib/$(MACH64)/llib-lzfs.ln +$(ROOT)/usr/lib/$(MACH64)/llib-lzfs_core.ln:= \ + REALPATH=../../../lib/$(MACH64)/llib-lzfs_core.ln $(ROOT)/usr/lib/$(MACH64)/llib-lfdisk.ln:= \ REALPATH=../../../lib/$(MACH64)/llib-lfdisk.ln $(ROOT)/usr/lib/$(MACH64)/nss_compat.so.1:= \ @@ -1627,6 +1638,8 @@ /usr/lib/libxnet.so.1 \ /usr/lib/libzfs.so \ /usr/lib/libzfs.so.1 \ + /usr/lib/libzfs_core.so \ + /usr/lib/libzfs_core.so.1 \ /usr/lib/llib-ladm \ /usr/lib/llib-ladm.ln \ /usr/lib/llib-laio \ @@ -1725,6 +1738,8 @@ /usr/lib/llib-lxnet.ln \ /usr/lib/llib-lzfs \ /usr/lib/llib-lzfs.ln \ + /usr/lib/llib-lzfs_core \ + /usr/lib/llib-lzfs_core.ln \ /usr/lib/nss_compat.so.1 \ /usr/lib/nss_dns.so.1 \ /usr/lib/nss_files.so.1 \ @@ -1863,6 +1878,8 @@ /usr/lib/$(MACH64)/libxnet.so.1 \ /usr/lib/$(MACH64)/libzfs.so \ /usr/lib/$(MACH64)/libzfs.so.1 \ + /usr/lib/$(MACH64)/libzfs_core.so \ + /usr/lib/$(MACH64)/libzfs_core.so.1 \ /usr/lib/$(MACH64)/llib-ladm.ln \ /usr/lib/$(MACH64)/llib-laio.ln \ /usr/lib/$(MACH64)/llib-lavl.ln \ @@ -1909,6 +1926,7 @@ /usr/lib/$(MACH64)/llib-luuid.ln \ /usr/lib/$(MACH64)/llib-lxnet.ln \ /usr/lib/$(MACH64)/llib-lzfs.ln \ + /usr/lib/$(MACH64)/llib-lzfs_core.ln \ /usr/lib/$(MACH64)/nss_compat.so.1 \ /usr/lib/$(MACH64)/nss_dns.so.1 \ /usr/lib/$(MACH64)/nss_files.so.1 \ diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/cmd/truss/codes.c --- a/usr/src/cmd/truss/codes.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/cmd/truss/codes.c Sat Jun 30 13:00:47 2012 -0700 @@ -1251,7 +1251,7 @@ "zfs_cmd_t" }, { (uint_t)ZFS_IOC_SPACE_WRITTEN, "ZFS_IOC_SPACE_WRITTEN", "zfs_cmd_t" }, - { (uint_t)ZFS_IOC_DESTROY_SNAPS_NVL, "ZFS_IOC_DESTROY_SNAPS_NVL", + { (uint_t)ZFS_IOC_DESTROY_SNAPS, "ZFS_IOC_DESTROY_SNAPS", "zfs_cmd_t" }, { (uint_t)ZFS_IOC_POOL_REGUID, "ZFS_IOC_POOL_REGUID", "zfs_cmd_t" }, @@ -1259,6 +1259,14 @@ "zfs_cmd_t" }, { (uint_t)ZFS_IOC_SEND_PROGRESS, "ZFS_IOC_SEND_PROGRESS", "zfs_cmd_t" }, + { (uint_t)ZFS_IOC_LOG_HISTORY, "ZFS_IOC_LOG_HISTORY", + "zfs_cmd_t" }, + { (uint_t)ZFS_IOC_SEND_NEW, "ZFS_IOC_SEND_NEW", + "zfs_cmd_t" }, + { (uint_t)ZFS_IOC_SEND_SPACE, "ZFS_IOC_SEND_SPACE", + "zfs_cmd_t" }, + { (uint_t)ZFS_IOC_CLONE, "ZFS_IOC_CLONE", + "zfs_cmd_t" }, /* kssl ioctls */ { (uint_t)KSSL_ADD_ENTRY, "KSSL_ADD_ENTRY", diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/cmd/zdb/zdb.c --- a/usr/src/cmd/zdb/zdb.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/cmd/zdb/zdb.c Sat Jun 30 13:00:47 2012 -0700 @@ -57,6 +57,7 @@ #include #include #include +#include #undef ZFS_MAXNAMELEN #undef verify #include @@ -204,6 +205,27 @@ nvlist_free(nv); } +/* ARGSUSED */ +static void +dump_history_offsets(objset_t *os, uint64_t object, void *data, size_t size) +{ + spa_history_phys_t *shp = data; + + if (shp == NULL) + return; + + (void) printf("\t\tpool_create_len = %llu\n", + (u_longlong_t)shp->sh_pool_create_len); + (void) printf("\t\tphys_max_off = %llu\n", + (u_longlong_t)shp->sh_phys_max_off); + (void) printf("\t\tbof = %llu\n", + (u_longlong_t)shp->sh_bof); + (void) printf("\t\teof = %llu\n", + (u_longlong_t)shp->sh_eof); + (void) printf("\t\trecords_lost = %llu\n", + (u_longlong_t)shp->sh_records_lost); +} + static void zdb_nicenum(uint64_t num, char *buf) { @@ -853,21 +875,22 @@ for (int i = 0; i < num; i++) { uint64_t time, txg, ievent; char *cmd, *intstr; + boolean_t printed = B_FALSE; if (nvlist_lookup_uint64(events[i], ZPOOL_HIST_TIME, &time) != 0) - continue; + goto next; if (nvlist_lookup_string(events[i], ZPOOL_HIST_CMD, &cmd) != 0) { if (nvlist_lookup_uint64(events[i], ZPOOL_HIST_INT_EVENT, &ievent) != 0) - continue; + goto next; verify(nvlist_lookup_uint64(events[i], ZPOOL_HIST_TXG, &txg) == 0); verify(nvlist_lookup_string(events[i], ZPOOL_HIST_INT_STR, &intstr) == 0); - if (ievent >= LOG_END) - continue; + if (ievent >= ZFS_NUM_LEGACY_HISTORY_EVENTS) + goto next; (void) snprintf(internalstr, sizeof (internalstr), @@ -880,6 +903,14 @@ (void) localtime_r(&tsec, &t); (void) strftime(tbuf, sizeof (tbuf), "%F.%T", &t); (void) printf("%s %s\n", tbuf, cmd); + printed = B_TRUE; + +next: + if (dump_opt['h'] > 1) { + if (!printed) + (void) printf("unrecognized record:\n"); + dump_nvlist(events[i], 2); + } } } @@ -1456,7 +1487,7 @@ dump_zap, /* other ZAP */ dump_zap, /* persistent error log */ dump_uint8, /* SPA history */ - dump_uint64, /* SPA history offsets */ + dump_history_offsets, /* SPA history offsets */ dump_zap, /* Pool properties */ dump_zap, /* DSL permissions */ dump_acl, /* ZFS ACL */ diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/cmd/zfs/Makefile --- a/usr/src/cmd/zfs/Makefile Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/cmd/zfs/Makefile Sat Jun 30 13:00:47 2012 -0700 @@ -18,10 +18,11 @@ # # CDDL HEADER END # +# Copyright 2010 Sun Microsystems, Inc. All rights reserved. +# Use is subject to license terms. # -# Copyright 2010 Sun Microsystems, Inc. All rights reserved. # Copyright 2010 Nexenta Systems, Inc. All rights reserved. -# Use is subject to license terms. +# Copyright (c) 2012 by Delphix. All rights reserved. # PROG= zfs @@ -38,7 +39,7 @@ ROOTETCFSTYPE= $(ROOTETC)/fs/$(FSTYPE) USRLIBFSTYPE= $(ROOTLIB)/fs/$(FSTYPE) -LDLIBS += -lzfs -luutil -lumem -lnvpair -lsec -lidmap +LDLIBS += -lzfs_core -lzfs -luutil -lumem -lnvpair -lsec -lidmap INCS += -I../../common/zfs diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/cmd/zfs/zfs_main.c --- a/usr/src/cmd/zfs/zfs_main.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/cmd/zfs/zfs_main.c Sat Jun 30 13:00:47 2012 -0700 @@ -56,6 +56,7 @@ #include #include +#include #include #include #include @@ -70,6 +71,7 @@ static FILE *mnttab_file; static char history_str[HIS_MAX_RECORD_LEN]; +static boolean_t log_history = B_TRUE; static int zfs_do_clone(int argc, char **argv); static int zfs_do_create(int argc, char **argv); @@ -259,7 +261,7 @@ return (gettext("\tshare <-a | filesystem>\n")); case HELP_SNAPSHOT: return (gettext("\tsnapshot [-r] [-o property=value] ... " - "\n")); + " ...\n")); case HELP_UNMOUNT: return (gettext("\tunmount [-f] " "<-a | filesystem|mountpoint>\n")); @@ -888,9 +890,9 @@ nvlist_t *cb_nvl; /* first snap in contiguous run */ - zfs_handle_t *cb_firstsnap; + char *cb_firstsnap; /* previous snap in contiguous run */ - zfs_handle_t *cb_prevsnap; + char *cb_prevsnap; int64_t cb_snapused; char *cb_snapspec; } destroy_cbdata_t; @@ -1004,11 +1006,13 @@ if (nvlist_exists(cb->cb_nvl, name)) { if (cb->cb_firstsnap == NULL) - cb->cb_firstsnap = zfs_handle_dup(zhp); + cb->cb_firstsnap = strdup(name); if (cb->cb_prevsnap != NULL) - zfs_close(cb->cb_prevsnap); + free(cb->cb_prevsnap); /* this snap continues the current range */ - cb->cb_prevsnap = zfs_handle_dup(zhp); + cb->cb_prevsnap = strdup(name); + if (cb->cb_firstsnap == NULL || cb->cb_prevsnap == NULL) + nomem(); if (cb->cb_verbose) { if (cb->cb_parsable) { (void) printf("destroy\t%s\n", name); @@ -1023,12 +1027,12 @@ } else if (cb->cb_firstsnap != NULL) { /* end of this range */ uint64_t used = 0; - err = zfs_get_snapused_int(cb->cb_firstsnap, + err = lzc_snaprange_space(cb->cb_firstsnap, cb->cb_prevsnap, &used); cb->cb_snapused += used; - zfs_close(cb->cb_firstsnap); + free(cb->cb_firstsnap); cb->cb_firstsnap = NULL; - zfs_close(cb->cb_prevsnap); + free(cb->cb_prevsnap); cb->cb_prevsnap = NULL; } zfs_close(zhp); @@ -1045,13 +1049,13 @@ if (cb->cb_firstsnap != NULL) { uint64_t used = 0; if (err == 0) { - err = zfs_get_snapused_int(cb->cb_firstsnap, + err = lzc_snaprange_space(cb->cb_firstsnap, cb->cb_prevsnap, &used); } cb->cb_snapused += used; - zfs_close(cb->cb_firstsnap); + free(cb->cb_firstsnap); cb->cb_firstsnap = NULL; - zfs_close(cb->cb_prevsnap); + free(cb->cb_prevsnap); cb->cb_prevsnap = NULL; } return (err); @@ -1904,9 +1908,11 @@ /* * 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. + * to log this history once to each pool, and bypass + * the normal history logging that happens in main(). */ - verify(zpool_stage_history(g_zfs, history_str) == 0); + (void) zpool_log_history(g_zfs, history_str); + log_history = B_FALSE; } if (zfs_prop_set(zhp, "version", verstr) == 0) cb->cb_numupgraded++; @@ -3424,6 +3430,32 @@ return (ret); } +typedef struct snap_cbdata { + nvlist_t *sd_nvl; + boolean_t sd_recursive; + const char *sd_snapname; +} snap_cbdata_t; + +static int +zfs_snapshot_cb(zfs_handle_t *zhp, void *arg) +{ + snap_cbdata_t *sd = arg; + char *name; + int rv = 0; + int error; + + error = asprintf(&name, "%s@%s", zfs_get_name(zhp), sd->sd_snapname); + if (error == -1) + nomem(); + fnvlist_add_boolean(sd->sd_nvl, name); + free(name); + + if (sd->sd_recursive) + rv = zfs_iter_filesystems(zhp, zfs_snapshot_cb, sd); + zfs_close(zhp); + return (rv); +} + /* * zfs snapshot [-r] [-o prop=value] ... * @@ -3433,13 +3465,16 @@ static int zfs_do_snapshot(int argc, char **argv) { - boolean_t recursive = B_FALSE; int ret = 0; char c; nvlist_t *props; + snap_cbdata_t sd = { 0 }; + boolean_t multiple_snaps = B_FALSE; if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) nomem(); + if (nvlist_alloc(&sd.sd_nvl, NV_UNIQUE_NAME, 0) != 0) + nomem(); /* check options */ while ((c = getopt(argc, argv, "ro:")) != -1) { @@ -3449,7 +3484,8 @@ return (1); break; case 'r': - recursive = B_TRUE; + sd.sd_recursive = B_TRUE; + multiple_snaps = B_TRUE; break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), @@ -3466,18 +3502,35 @@ (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); + + if (argc > 1) + multiple_snaps = B_TRUE; + for (; argc > 0; argc--, argv++) { + char *atp; + zfs_handle_t *zhp; + + atp = strchr(argv[0], '@'); + if (atp == NULL) + goto usage; + *atp = '\0'; + sd.sd_snapname = atp + 1; + zhp = zfs_open(g_zfs, argv[0], + ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME); + if (zhp == NULL) + goto usage; + if (zfs_snapshot_cb(zhp, &sd) != 0) + goto usage; + } + + ret = zfs_snapshot_nvl(g_zfs, sd.sd_nvl, props); + nvlist_free(sd.sd_nvl); nvlist_free(props); - if (ret && recursive) + if (ret != 0 && multiple_snaps) (void) fprintf(stderr, gettext("no snapshots were created\n")); return (ret != 0); usage: + nvlist_free(sd.sd_nvl); nvlist_free(props); usage(B_FALSE); return (-1); @@ -6479,8 +6532,7 @@ return (1); } - zpool_set_history_str("zfs", argc, argv, history_str); - verify(zpool_stage_history(g_zfs, history_str) == 0); + zfs_save_arguments(argc, argv, history_str, sizeof (history_str)); libzfs_print_on_error(g_zfs, B_TRUE); @@ -6549,6 +6601,9 @@ (void) fclose(mnttab_file); + if (ret == 0 && log_history) + (void) zpool_log_history(g_zfs, history_str); + libzfs_fini(g_zfs); /* diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/cmd/zhack/zhack.c --- a/usr/src/cmd/zhack/zhack.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/cmd/zhack/zhack.c Sat Jun 30 13:00:47 2012 -0700 @@ -279,6 +279,9 @@ zfeature_info_t *feature = arg2; spa_feature_enable(spa, feature, tx); + spa_history_log_internal(spa, "zhack enable feature", tx, + "name=%s can_readonly=%u", + feature->fi_guid, feature->fi_can_readonly); } static void @@ -356,6 +359,8 @@ zfeature_info_t *feature = arg2; spa_feature_incr(spa, feature, tx); + spa_history_log_internal(spa, "zhack feature incr", tx, + "name=%s", feature->fi_guid); } static void @@ -365,6 +370,8 @@ zfeature_info_t *feature = arg2; spa_feature_decr(spa, feature, tx); + spa_history_log_internal(spa, "zhack feature decr", tx, + "name=%s", feature->fi_guid); } static void diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/cmd/zpool/zpool_main.c --- a/usr/src/cmd/zpool/zpool_main.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/cmd/zpool/zpool_main.c Sat Jun 30 13:00:47 2012 -0700 @@ -185,9 +185,9 @@ #define NCOMMAND (sizeof (command_table) / sizeof (command_table[0])) -zpool_command_t *current_command; +static zpool_command_t *current_command; static char history_str[HIS_MAX_RECORD_LEN]; - +static boolean_t log_history = B_TRUE; static uint_t timestamp_fmt = NODATE; static const char * @@ -935,7 +935,10 @@ return (1); } - ret = (zpool_destroy(zhp) != 0); + /* The history must be logged as part of the export */ + log_history = B_FALSE; + + ret = (zpool_destroy(zhp, history_str) != 0); zpool_close(zhp); @@ -999,10 +1002,13 @@ continue; } + /* The history must be logged as part of the export */ + log_history = B_FALSE; + if (hardforce) { - if (zpool_export_force(zhp) != 0) + if (zpool_export_force(zhp, history_str) != 0) ret = 1; - } else if (zpool_export(zhp, force) != 0) { + } else if (zpool_export(zhp, force, history_str) != 0) { ret = 1; } @@ -4269,6 +4275,14 @@ (void) printf(gettext("Successfully upgraded " "'%s'\n\n"), zpool_get_name(zhp)); } + /* + * If they did "zpool upgrade -a", then we could + * be doing ioctls to different pools. We need + * to log this history once to each pool, and bypass + * the normal history logging that happens in main(). + */ + (void) zpool_log_history(g_zfs, history_str); + log_history = B_FALSE; } } else if (cbp->cb_newer && !SPA_VERSION_IS_SUPPORTED(version)) { assert(!cbp->cb_all); @@ -4491,8 +4505,8 @@ typedef struct hist_cbdata { boolean_t first; - int longfmt; - int internal; + boolean_t longfmt; + boolean_t internal; } hist_cbdata_t; /* @@ -4504,21 +4518,8 @@ 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; @@ -4530,64 +4531,94 @@ 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) + nvlist_t *rec = records[i]; + char tbuf[30] = ""; + + if (nvlist_exists(rec, ZPOOL_HIST_TIME)) { + time_t tsec; + struct tm t; + + tsec = fnvlist_lookup_uint64(records[i], + ZPOOL_HIST_TIME); + (void) localtime_r(&tsec, &t); + (void) strftime(tbuf, sizeof (tbuf), "%F.%T", &t); + } + + if (nvlist_exists(rec, ZPOOL_HIST_CMD)) { + (void) printf("%s %s", tbuf, + fnvlist_lookup_string(rec, ZPOOL_HIST_CMD)); + } else if (nvlist_exists(rec, ZPOOL_HIST_INT_EVENT)) { + int ievent = + fnvlist_lookup_uint64(rec, ZPOOL_HIST_INT_EVENT); + if (!cb->internal) continue; - - if (nvlist_lookup_uint64(records[i], - ZPOOL_HIST_INT_EVENT, &ievent) != 0) + if (ievent >= ZFS_NUM_LEGACY_HISTORY_EVENTS) { + (void) printf("%s unrecognized record:\n", + tbuf); + dump_nvlist(rec, 4); + continue; + } + (void) printf("%s [internal %s txg:%lld] %s", tbuf, + zfs_history_event_names[ievent], + fnvlist_lookup_uint64(rec, ZPOOL_HIST_TXG), + fnvlist_lookup_string(rec, ZPOOL_HIST_INT_STR)); + } else if (nvlist_exists(rec, ZPOOL_HIST_INT_NAME)) { + if (!cb->internal) 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) + (void) printf("%s [txg:%lld] %s", tbuf, + fnvlist_lookup_uint64(rec, ZPOOL_HIST_TXG), + fnvlist_lookup_string(rec, ZPOOL_HIST_INT_NAME)); + if (nvlist_exists(rec, ZPOOL_HIST_DSNAME)) { + (void) printf(" %s (%llu)", + fnvlist_lookup_string(rec, + ZPOOL_HIST_DSNAME), + fnvlist_lookup_uint64(rec, + ZPOOL_HIST_DSID)); + } + (void) printf(" %s", fnvlist_lookup_string(rec, + ZPOOL_HIST_INT_STR)); + } else if (nvlist_exists(rec, ZPOOL_HIST_IOCTL)) { + if (!cb->internal) continue; - (void) snprintf(internalstr, - sizeof (internalstr), - "[internal %s txg:%lld] %s", - zfs_history_event_names[ievent], txg, - pathstr); - cmdstr = internalstr; + (void) printf("%s ioctl %s\n", tbuf, + fnvlist_lookup_string(rec, ZPOOL_HIST_IOCTL)); + if (nvlist_exists(rec, ZPOOL_HIST_INPUT_NVL)) { + (void) printf(" input:\n"); + dump_nvlist(fnvlist_lookup_nvlist(rec, + ZPOOL_HIST_INPUT_NVL), 8); + } + if (nvlist_exists(rec, ZPOOL_HIST_OUTPUT_NVL)) { + (void) printf(" output:\n"); + dump_nvlist(fnvlist_lookup_nvlist(rec, + ZPOOL_HIST_OUTPUT_NVL), 8); + } + } else { + if (!cb->internal) + continue; + (void) printf("%s unrecognized record:\n", tbuf); + dump_nvlist(rec, 4); } - 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_exists(rec, ZPOOL_HIST_WHO)) { + uid_t who = fnvlist_lookup_uint64(rec, ZPOOL_HIST_WHO); + struct passwd *pwd = getpwuid(who); + (void) printf("user %d ", (int)who); + if (pwd != NULL) + (void) printf("(%s) ", pwd->pw_name); } - if (nvlist_lookup_string(records[i], - ZPOOL_HIST_HOST, &hostname) == 0) { - (void) printf(" %s", hostname); + if (nvlist_exists(rec, ZPOOL_HIST_HOST)) { + (void) printf("on %s", + fnvlist_lookup_string(rec, ZPOOL_HIST_HOST)); } - if (nvlist_lookup_string(records[i], - ZPOOL_HIST_ZONE, &zonename) == 0) { - (void) printf(":%s", zonename); + if (nvlist_exists(rec, ZPOOL_HIST_ZONE)) { + (void) printf(":%s", + fnvlist_lookup_string(rec, ZPOOL_HIST_ZONE)); } - (void) printf("]"); (void) printf("\n"); } @@ -4602,8 +4633,6 @@ * * Displays the history of commands that modified pools. */ - - int zpool_do_history(int argc, char **argv) { @@ -4616,10 +4645,10 @@ while ((c = getopt(argc, argv, "li")) != -1) { switch (c) { case 'l': - cbdata.longfmt = 1; + cbdata.longfmt = B_TRUE; break; case 'i': - cbdata.internal = 1; + cbdata.internal = B_TRUE; break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), @@ -4844,8 +4873,7 @@ 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); + zfs_save_arguments(argc, argv, history_str, sizeof (history_str)); /* * Run the appropriate command. @@ -4872,6 +4900,9 @@ usage(B_FALSE); } + if (ret == 0 && log_history) + (void) zpool_log_history(g_zfs, history_str); + libzfs_fini(g_zfs); /* diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/cmd/ztest/ztest.c --- a/usr/src/cmd/ztest/ztest.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/cmd/ztest/ztest.c Sat Jun 30 13:00:47 2012 -0700 @@ -2252,7 +2252,7 @@ */ nvroot = make_vdev_root("/dev/bogus", NULL, 0, 0, 0, 0, 0, 1); VERIFY3U(ENOENT, ==, - spa_create("ztest_bad_file", nvroot, NULL, NULL, NULL)); + spa_create("ztest_bad_file", nvroot, NULL, NULL)); nvlist_free(nvroot); /* @@ -2260,7 +2260,7 @@ */ nvroot = make_vdev_root("/dev/bogus", NULL, 0, 0, 0, 0, 2, 1); VERIFY3U(ENOENT, ==, - spa_create("ztest_bad_mirror", nvroot, NULL, NULL, NULL)); + spa_create("ztest_bad_mirror", nvroot, NULL, NULL)); nvlist_free(nvroot); /* @@ -2269,7 +2269,7 @@ */ (void) rw_rdlock(&ztest_name_lock); nvroot = make_vdev_root("/dev/bogus", NULL, 0, 0, 0, 0, 0, 1); - VERIFY3U(EEXIST, ==, spa_create(zo->zo_pool, nvroot, NULL, NULL, NULL)); + VERIFY3U(EEXIST, ==, spa_create(zo->zo_pool, nvroot, NULL, NULL)); nvlist_free(nvroot); VERIFY3U(0, ==, spa_open(zo->zo_pool, &spa, FTAG)); VERIFY3U(EBUSY, ==, spa_destroy(zo->zo_pool)); @@ -3056,8 +3056,7 @@ (void) snprintf(snapname, MAXNAMELEN, "%s@%llu", osname, (u_longlong_t)id); - error = dmu_objset_snapshot(osname, strchr(snapname, '@') + 1, - NULL, NULL, B_FALSE, B_FALSE, -1); + error = dmu_objset_snapshot_one(osname, strchr(snapname, '@') + 1); if (error == ENOSPC) { ztest_record_enospc(FTAG); return (B_FALSE); @@ -3257,8 +3256,7 @@ (void) snprintf(clone2name, MAXNAMELEN, "%s/c2_%llu", osname, id); (void) snprintf(snap3name, MAXNAMELEN, "%s@s3_%llu", clone1name, id); - error = dmu_objset_snapshot(osname, strchr(snap1name, '@')+1, - NULL, NULL, B_FALSE, B_FALSE, -1); + error = dmu_objset_snapshot_one(osname, strchr(snap1name, '@') + 1); if (error && error != EEXIST) { if (error == ENOSPC) { ztest_record_enospc(FTAG); @@ -3281,8 +3279,7 @@ fatal(0, "dmu_objset_create(%s) = %d", clone1name, error); } - error = dmu_objset_snapshot(clone1name, strchr(snap2name, '@')+1, - NULL, NULL, B_FALSE, B_FALSE, -1); + error = dmu_objset_snapshot_one(clone1name, strchr(snap2name, '@') + 1); if (error && error != EEXIST) { if (error == ENOSPC) { ztest_record_enospc(FTAG); @@ -3291,8 +3288,7 @@ fatal(0, "dmu_open_snapshot(%s) = %d", snap2name, error); } - error = dmu_objset_snapshot(clone1name, strchr(snap3name, '@')+1, - NULL, NULL, B_FALSE, B_FALSE, -1); + error = dmu_objset_snapshot_one(clone1name, strchr(snap3name, '@') + 1); if (error && error != EEXIST) { if (error == ENOSPC) { ztest_record_enospc(FTAG); @@ -4480,8 +4476,7 @@ * Create snapshot, clone it, mark snap for deferred destroy, * destroy clone, verify snap was also destroyed. */ - error = dmu_objset_snapshot(osname, snapname, NULL, NULL, FALSE, - FALSE, -1); + error = dmu_objset_snapshot_one(osname, snapname); if (error) { if (error == ENOSPC) { ztest_record_enospc("dmu_objset_snapshot"); @@ -4523,8 +4518,7 @@ * destroy a held snapshot, mark for deferred destroy, * release hold, verify snapshot was destroyed. */ - error = dmu_objset_snapshot(osname, snapname, NULL, NULL, FALSE, - FALSE, -1); + error = dmu_objset_snapshot_one(osname, snapname); if (error) { if (error == ENOSPC) { ztest_record_enospc("dmu_objset_snapshot"); @@ -5612,8 +5606,7 @@ spa_feature_table[i].fi_uname); VERIFY3U(0, ==, nvlist_add_uint64(props, buf, 0)); } - VERIFY3U(0, ==, spa_create(ztest_opts.zo_pool, nvroot, props, - NULL, NULL)); + VERIFY3U(0, ==, spa_create(ztest_opts.zo_pool, nvroot, props, NULL)); nvlist_free(nvroot); VERIFY3U(0, ==, spa_open(ztest_opts.zo_pool, &spa, FTAG)); diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/common/zfs/zfs_comutil.c --- a/usr/src/common/zfs/zfs_comutil.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/common/zfs/zfs_comutil.c Sat Jun 30 13:00:47 2012 -0700 @@ -20,6 +20,7 @@ */ /* * Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ /* @@ -157,7 +158,11 @@ return (version); } -const char *zfs_history_event_names[LOG_END] = { +/* + * This is the table of legacy internal event names; it should not be modified. + * The internal events are now stored in the history log as strings. + */ +const char *zfs_history_event_names[ZFS_NUM_LEGACY_HISTORY_EVENTS] = { "invalid event", "pool create", "vdev add", diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/common/zfs/zfs_comutil.h --- a/usr/src/common/zfs/zfs_comutil.h Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/common/zfs/zfs_comutil.h Sat Jun 30 13:00:47 2012 -0700 @@ -20,6 +20,7 @@ */ /* * Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ #ifndef _ZFS_COMUTIL_H @@ -37,7 +38,8 @@ extern int zfs_zpl_version_map(int spa_version); extern int zfs_spa_version_map(int zpl_version); -extern const char *zfs_history_event_names[LOG_END]; +#define ZFS_NUM_LEGACY_HISTORY_EVENTS 41 +extern const char *zfs_history_event_names[ZFS_NUM_LEGACY_HISTORY_EVENTS]; #ifdef __cplusplus } diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/common/zfs/zprop_common.c --- a/usr/src/common/zfs/zprop_common.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/common/zfs/zprop_common.c Sat Jun 30 13:00:47 2012 -0700 @@ -22,6 +22,9 @@ * Copyright 2010 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ +/* + * Copyright (c) 2012 by Delphix. All rights reserved. + */ /* * Common routines used by zfs and zpool property management. @@ -129,7 +132,8 @@ zprop_attr_t attr, int objset_types, const char *colname) { zprop_register_impl(prop, name, type, 0, NULL, attr, - objset_types, NULL, colname, B_FALSE, B_FALSE, NULL); + objset_types, NULL, colname, + type == PROP_TYPE_NUMBER, B_FALSE, NULL); } diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/Makefile --- a/usr/src/lib/Makefile Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/lib/Makefile Sat Jun 30 13:00:47 2012 -0700 @@ -21,6 +21,7 @@ # Copyright 2011 Nexenta Systems, Inc. All rights reserved. # Copyright (c) 1989, 2010, Oracle and/or its affiliates. All rights reserved. +# Copyright (c) 2012 by Delphix. All rights reserved. include ../Makefile.master @@ -234,9 +235,10 @@ sasl_plugins \ udapl \ libzpool \ + libzfs_core \ libzfs \ - libbe \ - pylibbe \ + libbe \ + pylibbe \ libzfs_jni \ pyzfs \ pysolaris \ @@ -474,6 +476,7 @@ libwrap \ libxcurses2 \ libzfs \ + libzfs_core \ libzfs_jni \ libzoneinfo \ libzonestat \ @@ -638,7 +641,8 @@ print: libldap5 udapl/udapl_tavor: udapl/libdat libzfs: libdevid libgen libnvpair libuutil \ - libadm libavl libefi libidmap libmd + libadm libavl libefi libidmap libmd libzfs_core +libzfs_core: libnvpair libzfs_jni: libdiskmgt libnvpair libzfs libzpool: libavl libumem libnvpair libcmdutils libsec: libavl libidmap diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs/Makefile.com --- a/usr/src/lib/libzfs/Makefile.com Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/lib/libzfs/Makefile.com Sat Jun 30 13:00:47 2012 -0700 @@ -69,7 +69,7 @@ C99MODE= -xc99=%all C99LMODE= -Xc99=%all LDLIBS += -lc -lm -ldevid -lgen -lnvpair -luutil -lavl -lefi \ - -ladm -lidmap -ltsol -lmd -lumem + -ladm -lidmap -ltsol -lmd -lumem -lzfs_core CPPFLAGS += $(INCS) -D_LARGEFILE64_SOURCE=1 -D_REENTRANT SRCS= $(OBJS_COMMON:%.o=$(SRCDIR)/%.c) \ diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs/common/libzfs.h --- a/usr/src/lib/libzfs/common/libzfs.h Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/lib/libzfs/common/libzfs.h Sat Jun 30 13:00:47 2012 -0700 @@ -54,7 +54,8 @@ /* * libzfs errors */ -enum { +typedef enum zfs_error { + EZFS_SUCCESS = 0, /* no error -- success */ EZFS_NOMEM = 2000, /* out of memory */ EZFS_BADPROP, /* invalid property value */ EZFS_PROPREADONLY, /* cannot set readonly property */ @@ -126,7 +127,7 @@ EZFS_DIFFDATA, /* bad zfs diff data */ EZFS_POOLREADONLY, /* pool is in read-only mode */ EZFS_UNKNOWN -}; +} zfs_error_t; /* * The following data structures are all part @@ -182,6 +183,9 @@ extern void libzfs_print_on_error(libzfs_handle_t *, boolean_t); +extern void zfs_save_arguments(int argc, char **, char *, int); +extern int zpool_log_history(libzfs_handle_t *, const char *); + extern int libzfs_errno(libzfs_handle_t *); extern const char *libzfs_error_action(libzfs_handle_t *); extern const char *libzfs_error_description(libzfs_handle_t *); @@ -216,7 +220,7 @@ */ extern int zpool_create(libzfs_handle_t *, const char *, nvlist_t *, nvlist_t *, nvlist_t *); -extern int zpool_destroy(zpool_handle_t *); +extern int zpool_destroy(zpool_handle_t *, const char *); extern int zpool_add(zpool_handle_t *, nvlist_t *); typedef struct splitflags { @@ -338,8 +342,8 @@ /* * Import and export functions */ -extern int zpool_export(zpool_handle_t *, boolean_t); -extern int zpool_export_force(zpool_handle_t *); +extern int zpool_export(zpool_handle_t *, boolean_t, const char *); +extern int zpool_export_force(zpool_handle_t *, const char *); extern int zpool_import(libzfs_handle_t *, nvlist_t *, const char *, char *altroot); extern int zpool_import_props(libzfs_handle_t *, nvlist_t *, const char *, @@ -373,7 +377,7 @@ */ struct zfs_cmd; -extern const char *zfs_history_event_names[LOG_END]; +extern const char *zfs_history_event_names[]; extern char *zpool_vdev_name(libzfs_handle_t *, zpool_handle_t *, nvlist_t *, boolean_t verbose); @@ -381,9 +385,6 @@ extern int zpool_get_history(zpool_handle_t *, nvlist_t **); extern int zpool_history_unpack(char *, uint64_t, uint64_t *, nvlist_t ***, uint_t *); -extern void zpool_set_history_str(const char *subcommand, int argc, - char **argv, char *history_str); -extern int zpool_stage_history(libzfs_handle_t *, const char *); extern void zpool_obj_to_path(zpool_handle_t *, uint64_t, uint64_t, char *, size_t len); extern int zfs_ioctl(libzfs_handle_t *, int, struct zfs_cmd *); @@ -436,8 +437,6 @@ char *propbuf, int proplen, boolean_t literal); extern int zfs_prop_get_feature(zfs_handle_t *zhp, const char *propname, char *buf, size_t len); -extern int zfs_get_snapused_int(zfs_handle_t *firstsnap, zfs_handle_t *lastsnap, - uint64_t *usedp); extern uint64_t zfs_prop_get_int(zfs_handle_t *, zfs_prop_t); extern int zfs_prop_inherit(zfs_handle_t *, const char *, boolean_t); extern const char *zfs_prop_values(zfs_prop_t); @@ -553,6 +552,8 @@ extern int zfs_destroy_snaps_nvl(zfs_handle_t *, nvlist_t *, boolean_t); extern int zfs_clone(zfs_handle_t *, const char *, nvlist_t *); extern int zfs_snapshot(libzfs_handle_t *, const char *, boolean_t, nvlist_t *); +extern int zfs_snapshot_nvl(libzfs_handle_t *hdl, nvlist_t *snaps, + nvlist_t *props); extern int zfs_rollback(zfs_handle_t *, zfs_handle_t *, boolean_t); extern int zfs_rename(zfs_handle_t *, const char *, boolean_t, boolean_t); diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs/common/libzfs_dataset.c --- a/usr/src/lib/libzfs/common/libzfs_dataset.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/lib/libzfs/common/libzfs_dataset.c Sat Jun 30 13:00:47 2012 -0700 @@ -1407,8 +1407,7 @@ libzfs_handle_t *hdl = zhp->zfs_hdl; nvlist_t *nvl = NULL, *realprops; zfs_prop_t prop; - boolean_t do_prefix; - uint64_t idx; + boolean_t do_prefix = B_TRUE; int added_resv; (void) snprintf(errbuf, sizeof (errbuf), @@ -1447,12 +1446,17 @@ } /* - * If the dataset's canmount property is being set to noauto, - * then we want to prevent unmounting & remounting it. + * We don't want to unmount & remount the dataset when changing + * its canmount property to 'on' or 'noauto'. We only use + * the changelist logic to unmount when setting canmount=off. */ - do_prefix = !((prop == ZFS_PROP_CANMOUNT) && - (zprop_string_to_index(prop, propval, &idx, - ZFS_TYPE_DATASET) == 0) && (idx == ZFS_CANMOUNT_NOAUTO)); + if (prop == ZFS_PROP_CANMOUNT) { + uint64_t idx; + int err = zprop_string_to_index(prop, propval, &idx, + ZFS_TYPE_DATASET); + if (err == 0 && idx != ZFS_CANMOUNT_OFF) + do_prefix = B_FALSE; + } if (do_prefix && (ret = changelist_prefix(cl)) != 0) goto error; @@ -2641,25 +2645,6 @@ return (0); } -int -zfs_get_snapused_int(zfs_handle_t *firstsnap, zfs_handle_t *lastsnap, - uint64_t *usedp) -{ - int err; - zfs_cmd_t zc = { 0 }; - - (void) strlcpy(zc.zc_name, lastsnap->zfs_name, sizeof (zc.zc_name)); - (void) strlcpy(zc.zc_value, firstsnap->zfs_name, sizeof (zc.zc_value)); - - err = ioctl(lastsnap->zfs_hdl->libzfs_fd, ZFS_IOC_SPACE_SNAPS, &zc); - if (err) - return (err); - - *usedp = zc.zc_cookie; - - return (0); -} - /* * Returns the name of the given zfs handle. */ @@ -2860,7 +2845,6 @@ */ for (cp = target + prefixlen + 1; cp = strchr(cp, '/'); *cp = '/', cp++) { - char *logstr; *cp = '\0'; @@ -2871,16 +2855,12 @@ continue; } - logstr = hdl->libzfs_log_str; - hdl->libzfs_log_str = NULL; if (zfs_create(hdl, target, ZFS_TYPE_FILESYSTEM, NULL) != 0) { - hdl->libzfs_log_str = logstr; opname = dgettext(TEXT_DOMAIN, "create"); goto ancestorerr; } - hdl->libzfs_log_str = logstr; h = zfs_open(hdl, target, ZFS_TYPE_FILESYSTEM); if (h == NULL) { opname = dgettext(TEXT_DOMAIN, "open"); @@ -2938,12 +2918,12 @@ zfs_create(libzfs_handle_t *hdl, const char *path, zfs_type_t type, nvlist_t *props) { - zfs_cmd_t zc = { 0 }; int ret; uint64_t size = 0; uint64_t blocksize = zfs_prop_default_numeric(ZFS_PROP_VOLBLOCKSIZE); char errbuf[1024]; uint64_t zoned; + dmu_objset_type_t ost; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot create '%s'"), path); @@ -2963,17 +2943,16 @@ * will return ENOENT, not EEXIST. To prevent this from happening, we * first try to see if the dataset exists. */ - (void) strlcpy(zc.zc_name, path, sizeof (zc.zc_name)); - if (zfs_dataset_exists(hdl, zc.zc_name, ZFS_TYPE_DATASET)) { + if (zfs_dataset_exists(hdl, path, ZFS_TYPE_DATASET)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "dataset already exists")); return (zfs_error(hdl, EZFS_EXISTS, errbuf)); } if (type == ZFS_TYPE_VOLUME) - zc.zc_objset_type = DMU_OST_ZVOL; + ost = DMU_OST_ZVOL; else - zc.zc_objset_type = DMU_OST_ZFS; + ost = DMU_OST_ZFS; if (props && (props = zfs_valid_proplist(hdl, type, props, zoned, NULL, errbuf)) == 0) @@ -3025,15 +3004,10 @@ } } - if (props && zcmd_write_src_nvlist(hdl, &zc, props) != 0) - return (-1); + /* create the dataset */ + ret = lzc_create(path, ost, props); nvlist_free(props); - /* create the dataset */ - ret = zfs_ioctl(hdl, ZFS_IOC_CREATE, &zc); - - zcmd_free_nvlists(&zc); - /* check for failure */ if (ret != 0) { char parent[ZFS_MAXNAMELEN]; @@ -3170,33 +3144,35 @@ zfs_destroy_snaps_nvl(zfs_handle_t *zhp, nvlist_t *snaps, boolean_t defer) { int ret; - zfs_cmd_t zc = { 0 }; - - (void) strlcpy(zc.zc_name, zhp->zfs_name, sizeof (zc.zc_name)); - if (zcmd_write_src_nvlist(zhp->zfs_hdl, &zc, snaps) != 0) - return (-1); - zc.zc_defer_destroy = defer; - - ret = zfs_ioctl(zhp->zfs_hdl, ZFS_IOC_DESTROY_SNAPS_NVL, &zc); + nvlist_t *errlist; + + ret = lzc_destroy_snaps(snaps, defer, &errlist); + if (ret != 0) { - char errbuf[1024]; - - (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, - "cannot destroy snapshots in %s"), zc.zc_name); - - switch (errno) { - case EEXIST: - zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN, - "snapshot is cloned")); - return (zfs_error(zhp->zfs_hdl, EZFS_EXISTS, errbuf)); - - default: - return (zfs_standard_error(zhp->zfs_hdl, errno, - errbuf)); + for (nvpair_t *pair = nvlist_next_nvpair(errlist, NULL); + pair != NULL; pair = nvlist_next_nvpair(errlist, pair)) { + char errbuf[1024]; + (void) snprintf(errbuf, sizeof (errbuf), + dgettext(TEXT_DOMAIN, "cannot destroy snapshot %s"), + nvpair_name(pair)); + + switch (fnvpair_value_int32(pair)) { + case EEXIST: + zfs_error_aux(zhp->zfs_hdl, + dgettext(TEXT_DOMAIN, + "snapshot is cloned")); + ret = zfs_error(zhp->zfs_hdl, EZFS_EXISTS, + errbuf); + break; + default: + ret = zfs_standard_error(zhp->zfs_hdl, errno, + errbuf); + break; + } } } - return (0); + return (ret); } /* @@ -3205,12 +3181,10 @@ int zfs_clone(zfs_handle_t *zhp, const char *target, nvlist_t *props) { - zfs_cmd_t zc = { 0 }; char parent[ZFS_MAXNAMELEN]; int ret; char errbuf[1024]; libzfs_handle_t *hdl = zhp->zfs_hdl; - zfs_type_t type; uint64_t zoned; assert(zhp->zfs_type == ZFS_TYPE_SNAPSHOT); @@ -3229,32 +3203,21 @@ (void) parent_name(target, parent, sizeof (parent)); /* do the clone */ - if (ZFS_IS_VOLUME(zhp)) { - zc.zc_objset_type = DMU_OST_ZVOL; - type = ZFS_TYPE_VOLUME; - } else { - zc.zc_objset_type = DMU_OST_ZFS; - type = ZFS_TYPE_FILESYSTEM; - } if (props) { + zfs_type_t type; + if (ZFS_IS_VOLUME(zhp)) { + type = ZFS_TYPE_VOLUME; + } else { + type = ZFS_TYPE_FILESYSTEM; + } if ((props = zfs_valid_proplist(hdl, type, props, zoned, zhp, errbuf)) == NULL) return (-1); - - if (zcmd_write_src_nvlist(hdl, &zc, props) != 0) { - nvlist_free(props); - return (-1); - } - - nvlist_free(props); } - (void) strlcpy(zc.zc_name, target, sizeof (zc.zc_name)); - (void) strlcpy(zc.zc_value, zhp->zfs_name, sizeof (zc.zc_value)); - ret = zfs_ioctl(zhp->zfs_hdl, ZFS_IOC_CREATE, &zc); - - zcmd_free_nvlists(&zc); + ret = lzc_clone(target, zhp->zfs_name, props); + nvlist_free(props); if (ret != 0) { switch (errno) { @@ -3339,74 +3302,134 @@ return (ret); } +typedef struct snapdata { + nvlist_t *sd_nvl; + const char *sd_snapname; +} snapdata_t; + +static int +zfs_snapshot_cb(zfs_handle_t *zhp, void *arg) +{ + snapdata_t *sd = arg; + char name[ZFS_MAXNAMELEN]; + int rv = 0; + + (void) snprintf(name, sizeof (name), + "%s@%s", zfs_get_name(zhp), sd->sd_snapname); + + fnvlist_add_boolean(sd->sd_nvl, name); + + rv = zfs_iter_filesystems(zhp, zfs_snapshot_cb, sd); + zfs_close(zhp); + return (rv); +} + /* - * Takes a snapshot of the given dataset. + * Creates snapshots. The keys in the snaps nvlist are the snapshots to be + * created. */ int +zfs_snapshot_nvl(libzfs_handle_t *hdl, nvlist_t *snaps, nvlist_t *props) +{ + int ret; + char errbuf[1024]; + nvpair_t *elem; + nvlist_t *errors; + + (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, + "cannot create snapshots ")); + + elem = NULL; + while ((elem = nvlist_next_nvpair(snaps, elem)) != NULL) { + const char *snapname = nvpair_name(elem); + + /* validate the target name */ + if (!zfs_validate_name(hdl, snapname, ZFS_TYPE_SNAPSHOT, + B_TRUE)) { + (void) snprintf(errbuf, sizeof (errbuf), + dgettext(TEXT_DOMAIN, + "cannot create snapshot '%s'"), snapname); + return (zfs_error(hdl, EZFS_INVALIDNAME, errbuf)); + } + } + + if (props != NULL && + (props = zfs_valid_proplist(hdl, ZFS_TYPE_SNAPSHOT, + props, B_FALSE, NULL, errbuf)) == NULL) { + return (-1); + } + + ret = lzc_snapshot(snaps, props, &errors); + + if (ret != 0) { + boolean_t printed = B_FALSE; + for (elem = nvlist_next_nvpair(errors, NULL); + elem != NULL; + elem = nvlist_next_nvpair(errors, elem)) { + (void) snprintf(errbuf, sizeof (errbuf), + dgettext(TEXT_DOMAIN, + "cannot create snapshot '%s'"), nvpair_name(elem)); + (void) zfs_standard_error(hdl, + fnvpair_value_int32(elem), errbuf); + printed = B_TRUE; + } + if (!printed) { + switch (ret) { + case EXDEV: + zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, + "multiple snapshots of same " + "fs not allowed")); + (void) zfs_error(hdl, EZFS_EXISTS, errbuf); + + break; + default: + (void) zfs_standard_error(hdl, ret, errbuf); + } + } + } + + nvlist_free(props); + nvlist_free(errors); + return (ret); +} + +int zfs_snapshot(libzfs_handle_t *hdl, const char *path, boolean_t recursive, nvlist_t *props) { - const char *delim; - char parent[ZFS_MAXNAMELEN]; + int ret; + snapdata_t sd = { 0 }; + char fsname[ZFS_MAXNAMELEN]; + char *cp; zfs_handle_t *zhp; - zfs_cmd_t zc = { 0 }; - int ret; char errbuf[1024]; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, - "cannot snapshot '%s'"), path); - - /* validate the target name */ + "cannot snapshot %s"), path); + if (!zfs_validate_name(hdl, path, ZFS_TYPE_SNAPSHOT, B_TRUE)) return (zfs_error(hdl, EZFS_INVALIDNAME, errbuf)); - if (props) { - if ((props = zfs_valid_proplist(hdl, ZFS_TYPE_SNAPSHOT, - props, B_FALSE, NULL, errbuf)) == NULL) - return (-1); - - if (zcmd_write_src_nvlist(hdl, &zc, props) != 0) { - nvlist_free(props); - return (-1); - } - - nvlist_free(props); - } - - /* make sure the parent exists and is of the appropriate type */ - delim = strchr(path, '@'); - (void) strncpy(parent, path, delim - path); - parent[delim - path] = '\0'; - - if ((zhp = zfs_open(hdl, parent, ZFS_TYPE_FILESYSTEM | + (void) strlcpy(fsname, path, sizeof (fsname)); + cp = strchr(fsname, '@'); + *cp = '\0'; + sd.sd_snapname = cp + 1; + + if ((zhp = zfs_open(hdl, fsname, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME)) == NULL) { - zcmd_free_nvlists(&zc); return (-1); } - (void) strlcpy(zc.zc_name, zhp->zfs_name, sizeof (zc.zc_name)); - (void) strlcpy(zc.zc_value, delim+1, sizeof (zc.zc_value)); - if (ZFS_IS_VOLUME(zhp)) - zc.zc_objset_type = DMU_OST_ZVOL; - else - zc.zc_objset_type = DMU_OST_ZFS; - zc.zc_cookie = recursive; - ret = zfs_ioctl(zhp->zfs_hdl, ZFS_IOC_SNAPSHOT, &zc); - - zcmd_free_nvlists(&zc); - - /* - * if it was recursive, the one that actually failed will be in - * zc.zc_name. - */ - if (ret != 0) { - (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, - "cannot create snapshot '%s@%s'"), zc.zc_name, zc.zc_value); - (void) zfs_standard_error(hdl, errno, errbuf); + verify(nvlist_alloc(&sd.sd_nvl, NV_UNIQUE_NAME, 0) == 0); + if (recursive) { + (void) zfs_snapshot_cb(zfs_handle_dup(zhp), &sd); + } else { + fnvlist_add_boolean(sd.sd_nvl, path); } + ret = zfs_snapshot_nvl(hdl, sd.sd_nvl, props); + nvlist_free(sd.sd_nvl); zfs_close(zhp); - return (ret); } @@ -3434,17 +3457,13 @@ zfs_get_type(zhp) == ZFS_TYPE_SNAPSHOT && zfs_prop_get_int(zhp, ZFS_PROP_CREATETXG) > cbp->cb_create) { - char *logstr; cbp->cb_dependent = B_TRUE; cbp->cb_error |= zfs_iter_dependents(zhp, B_FALSE, rollback_destroy, cbp); cbp->cb_dependent = B_FALSE; - logstr = zhp->zfs_hdl->libzfs_log_str; - zhp->zfs_hdl->libzfs_log_str = NULL; cbp->cb_error |= zfs_destroy(zhp, B_FALSE); - zhp->zfs_hdl->libzfs_log_str = logstr; } } else { /* We must destroy this clone; first unmount it */ diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs/common/libzfs_impl.h --- a/usr/src/lib/libzfs/common/libzfs_impl.h Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/lib/libzfs/common/libzfs_impl.h Sat Jun 30 13:00:47 2012 -0700 @@ -21,11 +21,11 @@ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. - * Copyright (c) 2011 by Delphix. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ -#ifndef _LIBFS_IMPL_H -#define _LIBFS_IMPL_H +#ifndef _LIBZFS_IMPL_H +#define _LIBZFS_IMPL_H #include #include @@ -36,6 +36,7 @@ #include #include #include +#include #include @@ -67,7 +68,6 @@ int libzfs_desc_active; char libzfs_action[1024]; char libzfs_desc[1024]; - char *libzfs_log_str; int libzfs_printerr; int libzfs_storeerr; /* stuff error messages into buffer */ void *libzfs_sharehdl; /* libshare handle */ @@ -213,4 +213,4 @@ } #endif -#endif /* _LIBFS_IMPL_H */ +#endif /* _LIBZFS_IMPL_H */ diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs/common/libzfs_iter.c --- a/usr/src/lib/libzfs/common/libzfs_iter.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/lib/libzfs/common/libzfs_iter.c Sat Jun 30 13:00:47 2012 -0700 @@ -22,7 +22,7 @@ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright 2010 Nexenta Systems, Inc. All rights reserved. - * Copyright (c) 2011 by Delphix. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ #include @@ -301,12 +301,11 @@ zfs_iter_snapspec(zfs_handle_t *fs_zhp, const char *spec_orig, zfs_iter_f func, void *arg) { - char buf[ZFS_MAXNAMELEN]; - char *comma_separated, *cp; + char *buf, *comma_separated, *cp; int err = 0; int ret = 0; - (void) strlcpy(buf, spec_orig, sizeof (buf)); + buf = zfs_strdup(fs_zhp->zfs_hdl, spec_orig); cp = buf; while ((comma_separated = strsep(&cp, ",")) != NULL) { @@ -364,6 +363,7 @@ } } + free(buf); return (ret); } diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs/common/libzfs_pool.c --- a/usr/src/lib/libzfs/common/libzfs_pool.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/lib/libzfs/common/libzfs_pool.c Sat Jun 30 13:00:47 2012 -0700 @@ -34,6 +34,7 @@ #include #include #include +#include #include #include #include @@ -1205,7 +1206,7 @@ * datasets left in the pool. */ int -zpool_destroy(zpool_handle_t *zhp) +zpool_destroy(zpool_handle_t *zhp, const char *log_str) { zfs_cmd_t zc = { 0 }; zfs_handle_t *zfp = NULL; @@ -1217,6 +1218,7 @@ return (-1); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); + zc.zc_history = (uint64_t)(uintptr_t)log_str; if (zfs_ioctl(hdl, ZFS_IOC_POOL_DESTROY, &zc) != 0) { (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, @@ -1371,8 +1373,9 @@ * Exports the pool from the system. The caller must ensure that there are no * mounted datasets in the pool. */ -int -zpool_export_common(zpool_handle_t *zhp, boolean_t force, boolean_t hardforce) +static int +zpool_export_common(zpool_handle_t *zhp, boolean_t force, boolean_t hardforce, + const char *log_str) { zfs_cmd_t zc = { 0 }; char msg[1024]; @@ -1383,6 +1386,7 @@ (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_cookie = force; zc.zc_guid = hardforce; + zc.zc_history = (uint64_t)(uintptr_t)log_str; if (zfs_ioctl(zhp->zpool_hdl, ZFS_IOC_POOL_EXPORT, &zc) != 0) { switch (errno) { @@ -1404,15 +1408,15 @@ } int -zpool_export(zpool_handle_t *zhp, boolean_t force) +zpool_export(zpool_handle_t *zhp, boolean_t force, const char *log_str) { - return (zpool_export_common(zhp, force, B_FALSE)); + return (zpool_export_common(zhp, force, B_FALSE, log_str)); } int -zpool_export_force(zpool_handle_t *zhp) +zpool_export_force(zpool_handle_t *zhp, const char *log_str) { - return (zpool_export_common(zhp, B_TRUE, B_TRUE)); + return (zpool_export_common(zhp, B_TRUE, B_TRUE, log_str)); } static void @@ -3574,40 +3578,30 @@ } void -zpool_set_history_str(const char *subcommand, int argc, char **argv, - char *history_str) +zfs_save_arguments(int argc, char **argv, char *string, int len) { - int i; - - (void) strlcpy(history_str, subcommand, HIS_MAX_RECORD_LEN); - for (i = 1; i < argc; i++) { - if (strlen(history_str) + 1 + strlen(argv[i]) > - HIS_MAX_RECORD_LEN) - break; - (void) strlcat(history_str, " ", HIS_MAX_RECORD_LEN); - (void) strlcat(history_str, argv[i], HIS_MAX_RECORD_LEN); + (void) strlcpy(string, basename(argv[0]), len); + for (int i = 1; i < argc; i++) { + (void) strlcat(string, " ", len); + (void) strlcat(string, argv[i], len); } } -/* - * Stage command history for logging. - */ int -zpool_stage_history(libzfs_handle_t *hdl, const char *history_str) +zpool_log_history(libzfs_handle_t *hdl, const char *message) { - if (history_str == NULL) - return (EINVAL); - - if (strlen(history_str) > HIS_MAX_RECORD_LEN) - return (EINVAL); - - if (hdl->libzfs_log_str != NULL) - free(hdl->libzfs_log_str); - - if ((hdl->libzfs_log_str = strdup(history_str)) == NULL) - return (no_memory(hdl)); - - return (0); + zfs_cmd_t zc = { 0 }; + nvlist_t *args; + int err; + + args = fnvlist_alloc(); + fnvlist_add_string(args, "message", message); + err = zcmd_write_src_nvlist(hdl, &zc, args); + if (err == 0) + err = ioctl(hdl->libzfs_fd, ZFS_IOC_LOG_HISTORY, &zc); + nvlist_free(args); + zcmd_free_nvlists(&zc); + return (err); } /* diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs/common/libzfs_util.c --- a/usr/src/lib/libzfs/common/libzfs_util.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/lib/libzfs/common/libzfs_util.c Sat Jun 30 13:00:47 2012 -0700 @@ -43,6 +43,7 @@ #include #include +#include #include "libzfs_impl.h" #include "zfs_prop.h" @@ -630,6 +631,14 @@ hdl->libzfs_sharetab = fopen("/etc/dfs/sharetab", "r"); + if (libzfs_core_init() != 0) { + (void) close(hdl->libzfs_fd); + (void) fclose(hdl->libzfs_mnttab); + (void) fclose(hdl->libzfs_sharetab); + free(hdl); + return (NULL); + } + zfs_prop_init(); zpool_prop_init(); zpool_feature_init(); @@ -647,12 +656,11 @@ if (hdl->libzfs_sharetab) (void) fclose(hdl->libzfs_sharetab); zfs_uninit_libshare(hdl); - if (hdl->libzfs_log_str) - (void) free(hdl->libzfs_log_str); zpool_free_handles(hdl); libzfs_fru_clear(hdl, B_TRUE); namespace_clear(hdl); libzfs_mnttab_fini(hdl); + libzfs_core_fini(); free(hdl); } @@ -814,17 +822,7 @@ int zfs_ioctl(libzfs_handle_t *hdl, int request, zfs_cmd_t *zc) { - int error; - - zc->zc_history = (uint64_t)(uintptr_t)hdl->libzfs_log_str; - error = ioctl(hdl->libzfs_fd, request, zc); - if (hdl->libzfs_log_str) { - free(hdl->libzfs_log_str); - hdl->libzfs_log_str = NULL; - } - zc->zc_history = 0; - - return (error); + return (ioctl(hdl->libzfs_fd, request, zc)); } /* diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs/common/llib-lzfs --- a/usr/src/lib/libzfs/common/llib-lzfs Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/lib/libzfs/common/llib-lzfs Sat Jun 30 13:00:47 2012 -0700 @@ -32,6 +32,7 @@ /*PROTOLIB1*/ #include +#include #include "../../../common/zfs/zfs_comutil.h" #include "../../../common/zfs/zfs_fletcher.h" #include "../../../common/zfs/zfs_prop.h" diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs/common/mapfile-vers --- a/usr/src/lib/libzfs/common/mapfile-vers Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/lib/libzfs/common/mapfile-vers Sat Jun 30 13:00:47 2012 -0700 @@ -76,7 +76,6 @@ zfs_get_holds; zfs_get_name; zfs_get_pool_handle; - zfs_get_snapused_int; zfs_get_user_props; zfs_get_type; zfs_handle_dup; @@ -133,6 +132,7 @@ zfs_release; zfs_rename; zfs_rollback; + zfs_save_arguments; zfs_send; zfs_share; zfs_shareall; @@ -144,6 +144,7 @@ zfs_smb_acl_remove; zfs_smb_acl_rename; zfs_snapshot; + zfs_snapshot_nvl; zfs_spa_version; zfs_spa_version_map; zfs_type_to_name; @@ -197,6 +198,7 @@ zpool_is_bootable; zpool_iter; zpool_label_disk; + zpool_log_history; zpool_mount_datasets; zpool_name_to_prop; zpool_obj_to_path; @@ -217,9 +219,7 @@ zpool_reopen; zpool_scan; zpool_search_import; - zpool_set_history_str; zpool_set_prop; - zpool_stage_history; zpool_state_to_name; zpool_unmount_datasets; zpool_upgrade; diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs_core/Makefile --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/usr/src/lib/libzfs_core/Makefile Sat Jun 30 13:00:47 2012 -0700 @@ -0,0 +1,64 @@ +# +# 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. +# +# Copyright (c) 2012 by Delphix. All rights reserved. +# + +include ../Makefile.lib + +HDRS= libzfs_core.h + +HDRDIR= common + +SUBDIRS= $(MACH) +$(BUILD64)SUBDIRS += $(MACH64) + +all := TARGET= all +clean := TARGET= clean +clobber := TARGET= clobber +install := TARGET= install +lint := TARGET= lint + +MSGFILES = `$(GREP) -l gettext $(HDRDIR)/*.[ch]` +POFILE = libzfs_core.po + +.KEEP_STATE: + +all clean clobber install lint: $(SUBDIRS) + +$(POFILE): pofile_MSGFILES + +install_h: $(ROOTHDRS) + +check: $(CHECKHDRS) + +_msg: $(MSGDOMAINPOFILE) + +$(SUBDIRS): FRC + @cd $@; pwd; $(MAKE) $(TARGET) + +FRC: + +include ../Makefile.targ +include ../../Makefile.msg.targ diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs_core/Makefile.com --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/usr/src/lib/libzfs_core/Makefile.com Sat Jun 30 13:00:47 2012 -0700 @@ -0,0 +1,69 @@ +# +# CDDL HEADER START +# +# The contents of this file are subject to the terms of the +# Common Development and Distribution License (the "License"). +# You may not use this file except in compliance with the License. +# +# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE +# or http://www.opensolaris.org/os/licensing. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# When distributing Covered Code, include this CDDL HEADER in each +# file and include the License file at usr/src/OPENSOLARIS.LICENSE. +# If applicable, add the following below this CDDL HEADER, with the +# fields enclosed by brackets "[]" replaced with your own identifying +# information: Portions Copyright [yyyy] [name of copyright owner] +# +# CDDL HEADER END +# +# +# Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. +# Copyright (c) 2012 by Delphix. All rights reserved. +# + +LIBRARY= libzfs_core.a +VERS= .1 + +OBJS_SHARED= + +OBJS_COMMON= \ + libzfs_core.o + +OBJECTS= $(OBJS_COMMON) $(OBJS_SHARED) + +include ../../Makefile.lib + +# libzfs_core must be installed in the root filesystem for mount(1M) +include ../../Makefile.rootfs + +LIBS= $(DYNLIB) $(LINTLIB) + +SRCDIR = ../common + +INCS += -I$(SRCDIR) +INCS += -I../../../uts/common/fs/zfs +INCS += -I../../../common/zfs +INCS += -I../../libc/inc + +C99MODE= -xc99=%all +C99LMODE= -Xc99=%all +LDLIBS += -lc -lnvpair +CPPFLAGS += $(INCS) -D_LARGEFILE64_SOURCE=1 -D_REENTRANT + +SRCS= $(OBJS_COMMON:%.o=$(SRCDIR)/%.c) \ + $(OBJS_SHARED:%.o=$(SRC)/common/zfs/%.c) +$(LINTLIB) := SRCS= $(SRCDIR)/$(LINTSRC) + +.KEEP_STATE: + +all: $(LIBS) + +lint: lintcheck + +pics/%.o: ../../../common/zfs/%.c + $(COMPILE.c) -o $@ $< + $(POST_PROCESS_O) + +include ../../Makefile.targ diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs_core/amd64/Makefile --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/usr/src/lib/libzfs_core/amd64/Makefile Sat Jun 30 13:00:47 2012 -0700 @@ -0,0 +1,30 @@ +# +# CDDL HEADER START +# +# The contents of this file are subject to the terms of the +# Common Development and Distribution License, Version 1.0 only +# (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 2005 Sun Microsystems, Inc. All rights reserved. +# Use is subject to license terms. +# + +include ../Makefile.com +include ../../Makefile.lib.64 + +install: all $(ROOTLIBS64) $(ROOTLINKS64) diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs_core/common/libzfs_core.c --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/usr/src/lib/libzfs_core/common/libzfs_core.c Sat Jun 30 13:00:47 2012 -0700 @@ -0,0 +1,477 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ + +/* + * Copyright (c) 2012 by Delphix. All rights reserved. + */ + +/* + * LibZFS_Core (lzc) is intended to replace most functionality in libzfs. + * It has the following characteristics: + * + * - Thread Safe. libzfs_core is accessible concurrently from multiple + * threads. This is accomplished primarily by avoiding global data + * (e.g. caching). Since it's thread-safe, there is no reason for a + * process to have multiple libzfs "instances". Therefore, we store + * our few pieces of data (e.g. the file descriptor) in global + * variables. The fd is reference-counted so that the libzfs_core + * library can be "initialized" multiple times (e.g. by different + * consumers within the same process). + * + * - Committed Interface. The libzfs_core interface will be committed, + * therefore consumers can compile against it and be confident that + * their code will continue to work on future releases of this code. + * Currently, the interface is Evolving (not Committed), but we intend + * to commit to it once it is more complete and we determine that it + * meets the needs of all consumers. + * + * - Programatic Error Handling. libzfs_core communicates errors with + * defined error numbers, and doesn't print anything to stdout/stderr. + * + * - Thin Layer. libzfs_core is a thin layer, marshaling arguments + * to/from the kernel ioctls. There is generally a 1:1 correspondence + * between libzfs_core functions and ioctls to /dev/zfs. + * + * - Clear Atomicity. Because libzfs_core functions are generally 1:1 + * with kernel ioctls, and kernel ioctls are general atomic, each + * libzfs_core function is atomic. For example, creating multiple + * snapshots with a single call to lzc_snapshot() is atomic -- it + * can't fail with only some of the requested snapshots created, even + * in the event of power loss or system crash. + * + * - Continued libzfs Support. Some higher-level operations (e.g. + * support for "zfs send -R") are too complicated to fit the scope of + * libzfs_core. This functionality will continue to live in libzfs. + * Where appropriate, libzfs will use the underlying atomic operations + * of libzfs_core. For example, libzfs may implement "zfs send -R | + * zfs receive" by using individual "send one snapshot", rename, + * destroy, and "receive one snapshot" operations in libzfs_core. + * /sbin/zfs and /zbin/zpool will link with both libzfs and + * libzfs_core. Other consumers should aim to use only libzfs_core, + * since that will be the supported, stable interface going forwards. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +static int g_fd; +static pthread_mutex_t g_lock = PTHREAD_MUTEX_INITIALIZER; +static int g_refcount; + +int +libzfs_core_init(void) +{ + (void) pthread_mutex_lock(&g_lock); + if (g_refcount == 0) { + g_fd = open("/dev/zfs", O_RDWR); + if (g_fd < 0) { + (void) pthread_mutex_unlock(&g_lock); + return (errno); + } + } + g_refcount++; + (void) pthread_mutex_unlock(&g_lock); + return (0); +} + +void +libzfs_core_fini(void) +{ + (void) pthread_mutex_lock(&g_lock); + ASSERT3S(g_refcount, >, 0); + g_refcount--; + if (g_refcount == 0) + (void) close(g_fd); + (void) pthread_mutex_unlock(&g_lock); +} + +static int +lzc_ioctl(zfs_ioc_t ioc, const char *name, + nvlist_t *source, nvlist_t **resultp) +{ + zfs_cmd_t zc = { 0 }; + int error = 0; + char *packed; + size_t size; + + ASSERT3S(g_refcount, >, 0); + + (void) strlcpy(zc.zc_name, name, sizeof (zc.zc_name)); + + packed = fnvlist_pack(source, &size); + zc.zc_nvlist_src = (uint64_t)(uintptr_t)packed; + zc.zc_nvlist_src_size = size; + + if (resultp != NULL) { + zc.zc_nvlist_dst_size = MAX(size * 2, 128 * 1024); + zc.zc_nvlist_dst = (uint64_t)(uintptr_t) + malloc(zc.zc_nvlist_dst_size); + if (zc.zc_nvlist_dst == NULL) { + error = ENOMEM; + goto out; + } + } + + while (ioctl(g_fd, ioc, &zc) != 0) { + if (errno == ENOMEM && resultp != NULL) { + free((void *)(uintptr_t)zc.zc_nvlist_dst); + zc.zc_nvlist_dst_size *= 2; + zc.zc_nvlist_dst = (uint64_t)(uintptr_t) + malloc(zc.zc_nvlist_dst_size); + if (zc.zc_nvlist_dst == NULL) { + error = ENOMEM; + goto out; + } + } else { + error = errno; + break; + } + } + if (zc.zc_nvlist_dst_filled) { + *resultp = fnvlist_unpack((void *)(uintptr_t)zc.zc_nvlist_dst, + zc.zc_nvlist_dst_size); + } else if (resultp != NULL) { + *resultp = NULL; + } + +out: + fnvlist_pack_free(packed, size); + free((void *)(uintptr_t)zc.zc_nvlist_dst); + return (error); +} + +int +lzc_create(const char *fsname, dmu_objset_type_t type, nvlist_t *props) +{ + int error; + nvlist_t *args = fnvlist_alloc(); + fnvlist_add_int32(args, "type", type); + if (props != NULL) + fnvlist_add_nvlist(args, "props", props); + error = lzc_ioctl(ZFS_IOC_CREATE, fsname, args, NULL); + nvlist_free(args); + return (error); +} + +int +lzc_clone(const char *fsname, const char *origin, + nvlist_t *props) +{ + int error; + nvlist_t *args = fnvlist_alloc(); + fnvlist_add_string(args, "origin", origin); + if (props != NULL) + fnvlist_add_nvlist(args, "props", props); + error = lzc_ioctl(ZFS_IOC_CLONE, fsname, args, NULL); + nvlist_free(args); + return (error); +} + +/* + * Creates snapshots. + * + * The keys in the snaps nvlist are the snapshots to be created. + * They must all be in the same pool. + * + * The props nvlist is properties to set. Currently only user properties + * are supported. { user:prop_name -> string value } + * + * The returned results nvlist will have an entry for each snapshot that failed. + * The value will be the (int32) error code. + * + * The return value will be 0 if all snapshots were created, otherwise it will + * be the errno of a (undetermined) snapshot that failed. + */ +int +lzc_snapshot(nvlist_t *snaps, nvlist_t *props, nvlist_t **errlist) +{ + nvpair_t *elem; + nvlist_t *args; + int error; + char pool[MAXNAMELEN]; + + *errlist = NULL; + + /* determine the pool name */ + elem = nvlist_next_nvpair(snaps, NULL); + if (elem == NULL) + return (0); + (void) strlcpy(pool, nvpair_name(elem), sizeof (pool)); + pool[strcspn(pool, "/@")] = '\0'; + + args = fnvlist_alloc(); + fnvlist_add_nvlist(args, "snaps", snaps); + if (props != NULL) + fnvlist_add_nvlist(args, "props", props); + + error = lzc_ioctl(ZFS_IOC_SNAPSHOT, pool, args, errlist); + nvlist_free(args); + + return (error); +} + +/* + * Destroys snapshots. + * + * The keys in the snaps nvlist are the snapshots to be destroyed. + * They must all be in the same pool. + * + * Snapshots that do not exist will be silently ignored. + * + * If 'defer' is not set, and a snapshot has user holds or clones, the + * destroy operation will fail and none of the snapshots will be + * destroyed. + * + * If 'defer' is set, and a snapshot has user holds or clones, it will be + * marked for deferred destruction, and will be destroyed when the last hold + * or clone is removed/destroyed. + * + * The return value will be 0 if all snapshots were destroyed (or marked for + * later destruction if 'defer' is set) or didn't exist to begin with. + * + * Otherwise the return value will be the errno of a (undetermined) snapshot + * that failed, no snapshots will be destroyed, and the errlist will have an + * entry for each snapshot that failed. The value in the errlist will be + * the (int32) error code. + */ +int +lzc_destroy_snaps(nvlist_t *snaps, boolean_t defer, nvlist_t **errlist) +{ + nvpair_t *elem; + nvlist_t *args; + int error; + char pool[MAXNAMELEN]; + + /* determine the pool name */ + elem = nvlist_next_nvpair(snaps, NULL); + if (elem == NULL) + return (0); + (void) strlcpy(pool, nvpair_name(elem), sizeof (pool)); + pool[strcspn(pool, "/@")] = '\0'; + + args = fnvlist_alloc(); + fnvlist_add_nvlist(args, "snaps", snaps); + if (defer) + fnvlist_add_boolean(args, "defer"); + + error = lzc_ioctl(ZFS_IOC_DESTROY_SNAPS, pool, args, errlist); + nvlist_free(args); + + return (error); + +} + +int +lzc_snaprange_space(const char *firstsnap, const char *lastsnap, + uint64_t *usedp) +{ + nvlist_t *args; + nvlist_t *result; + int err; + char fs[MAXNAMELEN]; + char *atp; + + /* determine the fs name */ + (void) strlcpy(fs, firstsnap, sizeof (fs)); + atp = strchr(fs, '@'); + if (atp == NULL) + return (EINVAL); + *atp = '\0'; + + args = fnvlist_alloc(); + fnvlist_add_string(args, "firstsnap", firstsnap); + + err = lzc_ioctl(ZFS_IOC_SPACE_SNAPS, lastsnap, args, &result); + nvlist_free(args); + if (err == 0) + *usedp = fnvlist_lookup_uint64(result, "used"); + fnvlist_free(result); + + return (err); +} + +boolean_t +lzc_exists(const char *dataset) +{ + /* + * The objset_stats ioctl is still legacy, so we need to construct our + * own zfs_cmd_t rather than using zfsc_ioctl(). + */ + zfs_cmd_t zc = { 0 }; + + (void) strlcpy(zc.zc_name, dataset, sizeof (zc.zc_name)); + return (ioctl(g_fd, ZFS_IOC_OBJSET_STATS, &zc) == 0); +} + +/* + * If fromsnap is NULL, a full (non-incremental) stream will be sent. + */ +int +lzc_send(const char *snapname, const char *fromsnap, int fd) +{ + nvlist_t *args; + int err; + + args = fnvlist_alloc(); + fnvlist_add_int32(args, "fd", fd); + if (fromsnap != NULL) + fnvlist_add_string(args, "fromsnap", fromsnap); + err = lzc_ioctl(ZFS_IOC_SEND_NEW, snapname, args, NULL); + nvlist_free(args); + return (err); +} + +/* + * If fromsnap is NULL, a full (non-incremental) stream will be estimated. + */ +int +lzc_send_space(const char *snapname, const char *fromsnap, uint64_t *spacep) +{ + nvlist_t *args; + nvlist_t *result; + int err; + + args = fnvlist_alloc(); + if (fromsnap != NULL) + fnvlist_add_string(args, "fromsnap", fromsnap); + err = lzc_ioctl(ZFS_IOC_SEND_SPACE, snapname, args, &result); + nvlist_free(args); + if (err == 0) + *spacep = fnvlist_lookup_uint64(result, "space"); + nvlist_free(result); + return (err); +} + +static int +recv_read(int fd, void *buf, int ilen) +{ + char *cp = buf; + int rv; + int len = ilen; + + do { + rv = read(fd, cp, len); + cp += rv; + len -= rv; + } while (rv > 0); + + if (rv < 0 || len != 0) + return (EIO); + + return (0); +} + +/* + * The simplest receive case: receive from the specified fd, creating the + * specified snapshot. Apply the specified properties a "received" properties + * (which can be overridden by locally-set properties). If the stream is a + * clone, its origin snapshot must be specified by 'origin'. The 'force' + * flag will cause the target filesystem to be rolled back or destroyed if + * necessary to receive. + * + * Return 0 on success or an errno on failure. + * + * Note: this interface does not work on dedup'd streams + * (those with DMU_BACKUP_FEATURE_DEDUP). + */ +int +lzc_receive(const char *snapname, nvlist_t *props, const char *origin, + boolean_t force, int fd) +{ + /* + * The receive ioctl is still legacy, so we need to construct our own + * zfs_cmd_t rather than using zfsc_ioctl(). + */ + zfs_cmd_t zc = { 0 }; + char *atp; + char *packed = NULL; + size_t size; + dmu_replay_record_t drr; + int error; + + ASSERT3S(g_refcount, >, 0); + + /* zc_name is name of containing filesystem */ + (void) strlcpy(zc.zc_name, snapname, sizeof (zc.zc_name)); + atp = strchr(zc.zc_name, '@'); + if (atp == NULL) + return (EINVAL); + *atp = '\0'; + + /* if the fs does not exist, try its parent. */ + if (!lzc_exists(zc.zc_name)) { + char *slashp = strrchr(zc.zc_name, '/'); + if (slashp == NULL) + return (ENOENT); + *slashp = '\0'; + + } + + /* zc_value is full name of the snapshot to create */ + (void) strlcpy(zc.zc_value, snapname, sizeof (zc.zc_value)); + + if (props != NULL) { + /* zc_nvlist_src is props to set */ + packed = fnvlist_pack(props, &size); + zc.zc_nvlist_src = (uint64_t)(uintptr_t)packed; + zc.zc_nvlist_src_size = size; + } + + /* zc_string is name of clone origin (if DRR_FLAG_CLONE) */ + if (origin != NULL) + (void) strlcpy(zc.zc_string, origin, sizeof (zc.zc_string)); + + /* zc_begin_record is non-byteswapped BEGIN record */ + error = recv_read(fd, &drr, sizeof (drr)); + if (error != 0) + goto out; + zc.zc_begin_record = drr.drr_u.drr_begin; + + /* zc_cookie is fd to read from */ + zc.zc_cookie = fd; + + /* zc guid is force flag */ + zc.zc_guid = force; + + /* zc_cleanup_fd is unused */ + zc.zc_cleanup_fd = -1; + + error = ioctl(g_fd, ZFS_IOC_RECV, &zc); + if (error != 0) + error = errno; + +out: + if (packed != NULL) + fnvlist_pack_free(packed, size); + free((void*)(uintptr_t)zc.zc_nvlist_dst); + return (error); +} diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs_core/common/libzfs_core.h --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/usr/src/lib/libzfs_core/common/libzfs_core.h Sat Jun 30 13:00:47 2012 -0700 @@ -0,0 +1,62 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ + +/* + * Copyright (c) 2012 by Delphix. All rights reserved. + */ + +#ifndef _LIBZFS_CORE_H +#define _LIBZFS_CORE_H + +#include +#include +#include +#include + +#ifdef __cplusplus +extern "C" { +#endif + +int libzfs_core_init(void); +void libzfs_core_fini(void); + +int lzc_snapshot(nvlist_t *snaps, nvlist_t *props, nvlist_t **errlist); +int lzc_create(const char *fsname, dmu_objset_type_t type, nvlist_t *props); +int lzc_clone(const char *fsname, const char *origin, nvlist_t *props); +int lzc_destroy_snaps(nvlist_t *snaps, boolean_t defer, nvlist_t **errlist); + +int lzc_snaprange_space(const char *firstsnap, const char *lastsnap, + uint64_t *usedp); + +int lzc_send(const char *snapname, const char *fromsnap, int fd); +int lzc_receive(const char *snapname, nvlist_t *props, const char *origin, + boolean_t force, int fd); +int lzc_send_space(const char *snapname, const char *fromsnap, + uint64_t *result); + +boolean_t lzc_exists(const char *dataset); + + +#ifdef __cplusplus +} +#endif + +#endif /* _LIBZFS_CORE_H */ diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs_core/common/llib-lzfs_core --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/usr/src/lib/libzfs_core/common/llib-lzfs_core Sat Jun 30 13:00:47 2012 -0700 @@ -0,0 +1,28 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or http://www.opensolaris.org/os/licensing. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ +/* + * Copyright (c) 2012 by Delphix. All rights reserved. + */ + +/*LINTLIBRARY*/ +/*PROTOLIB1*/ + +#include diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs_core/common/mapfile-vers --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/usr/src/lib/libzfs_core/common/mapfile-vers Sat Jun 30 13:00:47 2012 -0700 @@ -0,0 +1,56 @@ +# +# CDDL HEADER START +# +# The contents of this file are subject to the terms of the +# Common Development and Distribution License (the "License"). +# You may not use this file except in compliance with the License. +# +# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE +# or http://www.opensolaris.org/os/licensing. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# When distributing Covered Code, include this CDDL HEADER in each +# file and include the License file at usr/src/OPENSOLARIS.LICENSE. +# If applicable, add the following below this CDDL HEADER, with the +# fields enclosed by brackets "[]" replaced with your own identifying +# information: Portions Copyright [yyyy] [name of copyright owner] +# +# CDDL HEADER END +# +# Copyright (c) 2006, 2010, Oracle and/or its affiliates. All rights reserved. +# Copyright (c) 2012 by Delphix. All rights reserved. +# +# MAPFILE HEADER START +# +# WARNING: STOP NOW. DO NOT MODIFY THIS FILE. +# Object versioning must comply with the rules detailed in +# +# usr/src/lib/README.mapfiles +# +# You should not be making modifications here until you've read the most current +# copy of that file. If you need help, contact a gatekeeper for guidance. +# +# MAPFILE HEADER END +# + +$mapfile_version 2 + +SYMBOL_VERSION ILLUMOS_0.1 { + global: + + libzfs_core_fini; + libzfs_core_init; + lzc_clone; + lzc_create; + lzc_destroy_snaps; + lzc_exists; + lzc_receive; + lzc_send; + lzc_send_space; + lzc_snaprange_space; + lzc_snapshot; + + local: + *; +}; diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs_core/i386/Makefile --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/usr/src/lib/libzfs_core/i386/Makefile Sat Jun 30 13:00:47 2012 -0700 @@ -0,0 +1,29 @@ +# +# CDDL HEADER START +# +# The contents of this file are subject to the terms of the +# Common Development and Distribution License, Version 1.0 only +# (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 2005 Sun Microsystems, Inc. All rights reserved. +# Use is subject to license terms. +# + +include ../Makefile.com + +install: all $(ROOTLIBS) $(ROOTLINKS) $(ROOTLINT) diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs_core/sparc/Makefile --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/usr/src/lib/libzfs_core/sparc/Makefile Sat Jun 30 13:00:47 2012 -0700 @@ -0,0 +1,29 @@ +# +# CDDL HEADER START +# +# The contents of this file are subject to the terms of the +# Common Development and Distribution License, Version 1.0 only +# (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 2005 Sun Microsystems, Inc. All rights reserved. +# Use is subject to license terms. +# + +include ../Makefile.com + +install: all $(ROOTLIBS) $(ROOTLINKS) $(ROOTLINT) diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzfs_core/sparcv9/Makefile --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/usr/src/lib/libzfs_core/sparcv9/Makefile Sat Jun 30 13:00:47 2012 -0700 @@ -0,0 +1,32 @@ +# +# CDDL HEADER START +# +# The contents of this file are subject to the terms of the +# Common Development and Distribution License, Version 1.0 only +# (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. +# + +include ../Makefile.com +include ../../Makefile.lib.64 + +sparcv9_C_PICFLAGS= -K PIC + +install: all $(ROOTLIBS64) $(ROOTLINKS64) diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzpool/common/kernel.c --- a/usr/src/lib/libzpool/common/kernel.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/lib/libzpool/common/kernel.c Sat Jun 30 13:00:47 2012 -0700 @@ -871,6 +871,12 @@ return (0); } +uid_t +crgetruid(cred_t *cr) +{ + return (0); +} + gid_t crgetgid(cred_t *cr) { diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/lib/libzpool/common/sys/zfs_context.h --- a/usr/src/lib/libzpool/common/sys/zfs_context.h Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/lib/libzpool/common/sys/zfs_context.h Sat Jun 30 13:00:47 2012 -0700 @@ -286,6 +286,7 @@ #define rw_downgrade(rwlp) do { } while (0) extern uid_t crgetuid(cred_t *cr); +extern uid_t crgetruid(cred_t *cr); extern gid_t crgetgid(cred_t *cr); extern int crgetngroups(cred_t *cr); extern gid_t *crgetgroups(cred_t *cr); diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/man/man1m/zfs.1m --- a/usr/src/man/man1m/zfs.1m Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/man/man1m/zfs.1m Sat Jun 30 13:00:47 2012 -0700 @@ -39,7 +39,7 @@ .LP .nf \fBzfs\fR \fBsnapshot\fR [\fB-r\fR] [\fB-o\fR \fIproperty\fR=\fIvalue\fR]... - \fIfilesystem@snapname\fR|\fIvolume@snapname\fR + \fIfilesystem@snapname\fR|\fIvolume@snapname\fR... .fi .LP @@ -1837,13 +1837,14 @@ .ne 2 .na \fB\fBzfs snapshot\fR [\fB-r\fR] [\fB-o\fR \fIproperty\fR=\fIvalue\fR] ... -\fIfilesystem@snapname\fR|\fIvolume@snapname\fR\fR +\fIfilesystem@snapname\fR|\fIvolume@snapname\fR\fR... .ad .sp .6 .RS 4n -Creates a snapshot with the given name. All previous modifications by -successful system calls to the file system are part of the snapshot. See the -"Snapshots" section for details. +Creates snapshots with the given names. All previous modifications by +successful system calls to the file system are part of the snapshots. +Snapshots are taken atomically, so that all snapshots correspond to the same +moment in time. See the "Snapshots" section for details. .sp .ne 2 .na @@ -1851,9 +1852,7 @@ .ad .sp .6 .RS 4n -Recursively create snapshots of all descendent datasets. Snapshots are taken -atomically, so that all recursive snapshots correspond to the same moment in -time. +Recursively create snapshots of all descendent datasets .RE .sp diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/pkg/manifests/system-file-system-zfs.mf --- a/usr/src/pkg/manifests/system-file-system-zfs.mf Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/pkg/manifests/system-file-system-zfs.mf Sat Jun 30 13:00:47 2012 -0700 @@ -85,10 +85,15 @@ file path=kernel/kmdb/$(ARCH64)/zfs group=sys mode=0555 $(i386_ONLY)file path=kernel/kmdb/zfs group=sys mode=0555 file path=lib/$(ARCH64)/libzfs.so.1 +file path=lib/$(ARCH64)/libzfs_core.so.1 file path=lib/$(ARCH64)/llib-lzfs.ln +file path=lib/$(ARCH64)/llib-lzfs_core.ln file path=lib/libzfs.so.1 +file path=lib/libzfs_core.so.1 file path=lib/llib-lzfs file path=lib/llib-lzfs.ln +file path=lib/llib-lzfs_core +file path=lib/llib-lzfs_core.ln file path=sbin/zfs mode=0555 file path=sbin/zpool mode=0555 file path=usr/lib/$(ARCH64)/libzfs_jni.so.1 @@ -146,23 +151,35 @@ link path=etc/fs/zfs/mount target=../../../sbin/zfs link path=etc/fs/zfs/umount target=../../../sbin/zfs link path=lib/$(ARCH64)/libzfs.so target=libzfs.so.1 +link path=lib/$(ARCH64)/libzfs_core.so target=libzfs_core.so.1 link path=lib/libzfs.so target=libzfs.so.1 +link path=lib/libzfs_core.so target=libzfs_core.so.1 link path=usr/lib/$(ARCH64)/libzfs.so \ target=../../../lib/$(ARCH64)/libzfs.so.1 link path=usr/lib/$(ARCH64)/libzfs.so.1 \ target=../../../lib/$(ARCH64)/libzfs.so.1 +link path=usr/lib/$(ARCH64)/libzfs_core.so \ + target=../../../lib/$(ARCH64)/libzfs_core.so.1 +link path=usr/lib/$(ARCH64)/libzfs_core.so.1 \ + target=../../../lib/$(ARCH64)/libzfs_core.so.1 link path=usr/lib/$(ARCH64)/libzfs_jni.so target=libzfs_jni.so.1 link path=usr/lib/$(ARCH64)/libzpool.so target=libzpool.so.1 link path=usr/lib/$(ARCH64)/llib-lzfs.ln \ target=../../../lib/$(ARCH64)/llib-lzfs.ln +link path=usr/lib/$(ARCH64)/llib-lzfs_core.ln \ + target=../../../lib/$(ARCH64)/llib-lzfs_core.ln link path=usr/lib/fs/zfs/mount target=../../../../sbin/zfs link path=usr/lib/fs/zfs/umount target=../../../../sbin/zfs link path=usr/lib/libzfs.so target=../../lib/libzfs.so.1 link path=usr/lib/libzfs.so.1 target=../../lib/libzfs.so.1 +link path=usr/lib/libzfs_core.so target=../../lib/libzfs_core.so.1 +link path=usr/lib/libzfs_core.so.1 target=../../lib/libzfs_core.so.1 link path=usr/lib/libzfs_jni.so target=libzfs_jni.so.1 $(i386_ONLY)link path=usr/lib/libzpool.so target=libzpool.so.1 link path=usr/lib/llib-lzfs target=../../lib/llib-lzfs link path=usr/lib/llib-lzfs.ln target=../../lib/llib-lzfs.ln +link path=usr/lib/llib-lzfs_core target=../../lib/llib-lzfs_core +link path=usr/lib/llib-lzfs_core.ln target=../../lib/llib-lzfs_core.ln link path=usr/sbin/zfs target=../../sbin/zfs link path=usr/sbin/zpool target=../../sbin/zpool depend fmri=runtime/python-26 type=require diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/pkg/manifests/system-header.mf --- a/usr/src/pkg/manifests/system-header.mf Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/pkg/manifests/system-header.mf Sat Jun 30 13:00:47 2012 -0700 @@ -22,6 +22,7 @@ # # Copyright 2011 Nexenta Systems, Inc. All rights reserved. # Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved. +# Copyright (c) 2012 by Delphix. All rights reserved. # set name=pkg.fmri value=pkg:/system/header@$(PKGVERS) @@ -505,6 +506,7 @@ $(sparc_ONLY)file path=usr/include/libv12n.h file path=usr/include/libw.h file path=usr/include/libzfs.h +file path=usr/include/libzfs_core.h file path=usr/include/libzoneinfo.h file path=usr/include/limits.h file path=usr/include/linenum.h diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/dmu_objset.c --- a/usr/src/uts/common/fs/zfs/dmu_objset.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/dmu_objset.c Sat Jun 30 13:00:47 2012 -0700 @@ -20,6 +20,7 @@ */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ /* Portions Copyright 2010 Robert Milkowski */ @@ -699,30 +700,33 @@ spa_t *spa = dd->dd_pool->dp_spa; struct oscarg *oa = arg2; uint64_t obj; + dsl_dataset_t *ds; + blkptr_t *bp; ASSERT(dmu_tx_is_syncing(tx)); obj = dsl_dataset_create_sync(dd, oa->lastname, oa->clone_origin, oa->flags, oa->cr, tx); - if (oa->clone_origin == NULL) { - dsl_pool_t *dp = dd->dd_pool; - dsl_dataset_t *ds; - blkptr_t *bp; - objset_t *os; - - VERIFY3U(0, ==, dsl_dataset_hold_obj(dp, obj, FTAG, &ds)); - bp = dsl_dataset_get_blkptr(ds); - ASSERT(BP_IS_HOLE(bp)); - - os = dmu_objset_create_impl(spa, ds, bp, oa->type, tx); + VERIFY3U(0, ==, dsl_dataset_hold_obj(dd->dd_pool, obj, FTAG, &ds)); + bp = dsl_dataset_get_blkptr(ds); + if (BP_IS_HOLE(bp)) { + objset_t *os = + dmu_objset_create_impl(spa, ds, bp, oa->type, tx); if (oa->userfunc) oa->userfunc(os, oa->userarg, oa->cr, tx); - dsl_dataset_rele(ds, FTAG); } - spa_history_log_internal(LOG_DS_CREATE, spa, tx, "dataset = %llu", obj); + if (oa->clone_origin == NULL) { + spa_history_log_internal_ds(ds, "create", tx, ""); + } else { + char namebuf[MAXNAMELEN]; + dsl_dataset_name(oa->clone_origin, namebuf); + spa_history_log_internal_ds(ds, "clone", tx, + "origin=%s (%llu)", namebuf, oa->clone_origin->ds_object); + } + dsl_dataset_rele(ds, FTAG); } int @@ -799,34 +803,40 @@ return (error); } -struct snaparg { - dsl_sync_task_group_t *dstg; - char *snapname; - char *htag; - char failed[MAXPATHLEN]; - boolean_t recursive; - boolean_t needsuspend; - boolean_t temporary; - nvlist_t *props; - struct dsl_ds_holdarg *ha; /* only needed in the temporary case */ - dsl_dataset_t *newds; -}; +typedef struct snapallarg { + dsl_sync_task_group_t *saa_dstg; + boolean_t saa_needsuspend; + nvlist_t *saa_props; + + /* the following are used only if 'temporary' is set: */ + boolean_t saa_temporary; + const char *saa_htag; + struct dsl_ds_holdarg *saa_ha; + dsl_dataset_t *saa_newds; +} snapallarg_t; + +typedef struct snaponearg { + const char *soa_longname; /* long snap name */ + const char *soa_snapname; /* short snap name */ + snapallarg_t *soa_saa; +} snaponearg_t; static int snapshot_check(void *arg1, void *arg2, dmu_tx_t *tx) { objset_t *os = arg1; - struct snaparg *sn = arg2; + snaponearg_t *soa = arg2; + snapallarg_t *saa = soa->soa_saa; int error; /* The props have already been checked by zfs_check_userprops(). */ error = dsl_dataset_snapshot_check(os->os_dsl_dataset, - sn->snapname, tx); + soa->soa_snapname, tx); if (error) return (error); - if (sn->temporary) { + if (saa->saa_temporary) { /* * Ideally we would just call * dsl_dataset_user_hold_check() and @@ -844,12 +854,13 @@ * Not checking number of tags because the tag will be * unique, as it will be the only tag. */ - if (strlen(sn->htag) + MAX_TAG_PREFIX_LEN >= MAXNAMELEN) + if (strlen(saa->saa_htag) + MAX_TAG_PREFIX_LEN >= MAXNAMELEN) return (E2BIG); - sn->ha = kmem_alloc(sizeof (struct dsl_ds_holdarg), KM_SLEEP); - sn->ha->temphold = B_TRUE; - sn->ha->htag = sn->htag; + saa->saa_ha = kmem_alloc(sizeof (struct dsl_ds_holdarg), + KM_SLEEP); + saa->saa_ha->temphold = B_TRUE; + saa->saa_ha->htag = saa->saa_htag; } return (error); } @@ -859,24 +870,25 @@ { objset_t *os = arg1; dsl_dataset_t *ds = os->os_dsl_dataset; - struct snaparg *sn = arg2; - - dsl_dataset_snapshot_sync(ds, sn->snapname, tx); + snaponearg_t *soa = arg2; + snapallarg_t *saa = soa->soa_saa; - if (sn->props) { + dsl_dataset_snapshot_sync(ds, soa->soa_snapname, tx); + + if (saa->saa_props != NULL) { dsl_props_arg_t pa; - pa.pa_props = sn->props; + pa.pa_props = saa->saa_props; pa.pa_source = ZPROP_SRC_LOCAL; dsl_props_set_sync(ds->ds_prev, &pa, tx); } - if (sn->temporary) { + if (saa->saa_temporary) { struct dsl_ds_destroyarg da; - dsl_dataset_user_hold_sync(ds->ds_prev, sn->ha, tx); - kmem_free(sn->ha, sizeof (struct dsl_ds_holdarg)); - sn->ha = NULL; - sn->newds = ds->ds_prev; + dsl_dataset_user_hold_sync(ds->ds_prev, saa->saa_ha, tx); + kmem_free(saa->saa_ha, sizeof (struct dsl_ds_holdarg)); + saa->saa_ha = NULL; + saa->saa_newds = ds->ds_prev; da.ds = ds->ds_prev; da.defer = B_TRUE; @@ -885,131 +897,180 @@ } static int -dmu_objset_snapshot_one(const char *name, void *arg) +snapshot_one_impl(const char *snapname, void *arg) { - struct snaparg *sn = arg; + char fsname[MAXPATHLEN]; + snapallarg_t *saa = arg; + snaponearg_t *soa; objset_t *os; int err; - char *cp; - - /* - * If the objset starts with a '%', then ignore it unless it was - * explicitly named (ie, not recursive). These hidden datasets - * are always inconsistent, and by not opening them here, we can - * avoid a race with dsl_dir_destroy_check(). - */ - cp = strrchr(name, '/'); - if (cp && cp[1] == '%' && sn->recursive) - return (0); - (void) strcpy(sn->failed, name); + (void) strlcpy(fsname, snapname, sizeof (fsname)); + strchr(fsname, '@')[0] = '\0'; - /* - * Check permissions if we are doing a recursive snapshot. The - * permission checks for the starting dataset have already been - * performed in zfs_secpolicy_snapshot() - */ - if (sn->recursive && (err = zfs_secpolicy_snapshot_perms(name, CRED()))) - return (err); - - err = dmu_objset_hold(name, sn, &os); + err = dmu_objset_hold(fsname, saa, &os); if (err != 0) return (err); /* * If the objset is in an inconsistent state (eg, in the process - * of being destroyed), don't snapshot it. As with %hidden - * datasets, we return EBUSY if this name was explicitly - * requested (ie, not recursive), and otherwise ignore it. + * of being destroyed), don't snapshot it. */ if (os->os_dsl_dataset->ds_phys->ds_flags & DS_FLAG_INCONSISTENT) { - dmu_objset_rele(os, sn); - return (sn->recursive ? 0 : EBUSY); + dmu_objset_rele(os, saa); + return (EBUSY); } - if (sn->needsuspend) { + if (saa->saa_needsuspend) { err = zil_suspend(dmu_objset_zil(os)); if (err) { - dmu_objset_rele(os, sn); + dmu_objset_rele(os, saa); return (err); } } - dsl_sync_task_create(sn->dstg, snapshot_check, snapshot_sync, - os, sn, 3); + + soa = kmem_zalloc(sizeof (*soa), KM_SLEEP); + soa->soa_saa = saa; + soa->soa_longname = snapname; + soa->soa_snapname = strchr(snapname, '@') + 1; + + dsl_sync_task_create(saa->saa_dstg, snapshot_check, snapshot_sync, + os, soa, 3); return (0); } +/* + * The snapshots must all be in the same pool. + */ int -dmu_objset_snapshot(char *fsname, char *snapname, char *tag, - nvlist_t *props, boolean_t recursive, boolean_t temporary, int cleanup_fd) +dmu_objset_snapshot(nvlist_t *snaps, nvlist_t *props, nvlist_t *errors) { dsl_sync_task_t *dst; - struct snaparg sn; + snapallarg_t saa = { 0 }; + spa_t *spa; + int rv = 0; + int err; + nvpair_t *pair; + + pair = nvlist_next_nvpair(snaps, NULL); + if (pair == NULL) + return (0); + + err = spa_open(nvpair_name(pair), &spa, FTAG); + if (err) + return (err); + saa.saa_dstg = dsl_sync_task_group_create(spa_get_dsl(spa)); + saa.saa_props = props; + saa.saa_needsuspend = (spa_version(spa) < SPA_VERSION_FAST_SNAP); + + for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; + pair = nvlist_next_nvpair(snaps, pair)) { + err = snapshot_one_impl(nvpair_name(pair), &saa); + if (err != 0) { + if (errors != NULL) { + fnvlist_add_int32(errors, + nvpair_name(pair), err); + } + rv = err; + } + } + + /* + * If any call to snapshot_one_impl() failed, don't execute the + * sync task. The error handling code below will clean up the + * snaponearg_t from any successful calls to + * snapshot_one_impl(). + */ + if (rv == 0) + err = dsl_sync_task_group_wait(saa.saa_dstg); + if (err != 0) + rv = err; + + for (dst = list_head(&saa.saa_dstg->dstg_tasks); dst; + dst = list_next(&saa.saa_dstg->dstg_tasks, dst)) { + objset_t *os = dst->dst_arg1; + snaponearg_t *soa = dst->dst_arg2; + if (dst->dst_err != 0) { + if (errors != NULL) { + fnvlist_add_int32(errors, + soa->soa_longname, dst->dst_err); + } + rv = dst->dst_err; + } + + if (saa.saa_needsuspend) + zil_resume(dmu_objset_zil(os)); + dmu_objset_rele(os, &saa); + kmem_free(soa, sizeof (*soa)); + } + + dsl_sync_task_group_destroy(saa.saa_dstg); + spa_close(spa, FTAG); + return (rv); +} + +int +dmu_objset_snapshot_one(const char *fsname, const char *snapname) +{ + int err; + char *longsnap = kmem_asprintf("%s@%s", fsname, snapname); + nvlist_t *snaps = fnvlist_alloc(); + + fnvlist_add_boolean(snaps, longsnap); + err = dmu_objset_snapshot(snaps, NULL, NULL); + fnvlist_free(snaps); + strfree(longsnap); + return (err); +} + +int +dmu_objset_snapshot_tmp(const char *snapname, const char *tag, int cleanup_fd) +{ + dsl_sync_task_t *dst; + snapallarg_t saa = { 0 }; spa_t *spa; minor_t minor; int err; - (void) strcpy(sn.failed, fsname); - - err = spa_open(fsname, &spa, FTAG); + err = spa_open(snapname, &spa, FTAG); if (err) return (err); + saa.saa_dstg = dsl_sync_task_group_create(spa_get_dsl(spa)); + saa.saa_htag = tag; + saa.saa_needsuspend = (spa_version(spa) < SPA_VERSION_FAST_SNAP); + saa.saa_temporary = B_TRUE; - if (temporary) { - if (cleanup_fd < 0) { - spa_close(spa, FTAG); - return (EINVAL); - } - if ((err = zfs_onexit_fd_hold(cleanup_fd, &minor)) != 0) { - spa_close(spa, FTAG); - return (err); - } + if (cleanup_fd < 0) { + spa_close(spa, FTAG); + return (EINVAL); + } + if ((err = zfs_onexit_fd_hold(cleanup_fd, &minor)) != 0) { + spa_close(spa, FTAG); + return (err); } - sn.dstg = dsl_sync_task_group_create(spa_get_dsl(spa)); - sn.snapname = snapname; - sn.htag = tag; - sn.props = props; - sn.recursive = recursive; - sn.needsuspend = (spa_version(spa) < SPA_VERSION_FAST_SNAP); - sn.temporary = temporary; - sn.ha = NULL; - sn.newds = NULL; - - if (recursive) { - err = dmu_objset_find(fsname, - dmu_objset_snapshot_one, &sn, DS_FIND_CHILDREN); - } else { - err = dmu_objset_snapshot_one(fsname, &sn); - } + err = snapshot_one_impl(snapname, &saa); if (err == 0) - err = dsl_sync_task_group_wait(sn.dstg); + err = dsl_sync_task_group_wait(saa.saa_dstg); - for (dst = list_head(&sn.dstg->dstg_tasks); dst; - dst = list_next(&sn.dstg->dstg_tasks, dst)) { + for (dst = list_head(&saa.saa_dstg->dstg_tasks); dst; + dst = list_next(&saa.saa_dstg->dstg_tasks, dst)) { objset_t *os = dst->dst_arg1; - dsl_dataset_t *ds = os->os_dsl_dataset; - if (dst->dst_err) { - dsl_dataset_name(ds, sn.failed); - } else if (temporary) { - dsl_register_onexit_hold_cleanup(sn.newds, tag, minor); - } - if (sn.needsuspend) + dsl_register_onexit_hold_cleanup(saa.saa_newds, tag, minor); + if (saa.saa_needsuspend) zil_resume(dmu_objset_zil(os)); - dmu_objset_rele(os, &sn); + dmu_objset_rele(os, &saa); } - if (err) - (void) strcpy(fsname, sn.failed); - if (temporary) - zfs_onexit_fd_rele(cleanup_fd); - dsl_sync_task_group_destroy(sn.dstg); + zfs_onexit_fd_rele(cleanup_fd); + dsl_sync_task_group_destroy(saa.saa_dstg); spa_close(spa, FTAG); return (err); } + static void dmu_objset_sync_dnodes(list_t *list, list_t *newlist, dmu_tx_t *tx) { diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/dmu_send.c --- a/usr/src/uts/common/fs/zfs/dmu_send.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/dmu_send.c Sat Jun 30 13:00:47 2012 -0700 @@ -387,9 +387,48 @@ return (err); } +/* + * Return TRUE if 'earlier' is an earlier snapshot in 'later's timeline. + * For example, they could both be snapshots of the same filesystem, and + * 'earlier' is before 'later'. Or 'earlier' could be the origin of + * 'later's filesystem. Or 'earlier' could be an older snapshot in the origin's + * filesystem. Or 'earlier' could be the origin's origin. + */ +static boolean_t +is_before(dsl_dataset_t *later, dsl_dataset_t *earlier) +{ + dsl_pool_t *dp = later->ds_dir->dd_pool; + int error; + boolean_t ret; + dsl_dataset_t *origin; + + if (earlier->ds_phys->ds_creation_txg >= + later->ds_phys->ds_creation_txg) + return (B_FALSE); + + if (later->ds_dir == earlier->ds_dir) + return (B_TRUE); + if (!dsl_dir_is_clone(later->ds_dir)) + return (B_FALSE); + + rw_enter(&dp->dp_config_rwlock, RW_READER); + if (later->ds_dir->dd_phys->dd_origin_obj == earlier->ds_object) { + rw_exit(&dp->dp_config_rwlock); + return (B_TRUE); + } + error = dsl_dataset_hold_obj(dp, + later->ds_dir->dd_phys->dd_origin_obj, FTAG, &origin); + rw_exit(&dp->dp_config_rwlock); + if (error != 0) + return (B_FALSE); + ret = is_before(origin, earlier); + dsl_dataset_rele(origin, FTAG); + return (ret); +} + int -dmu_send(objset_t *tosnap, objset_t *fromsnap, boolean_t fromorigin, - int outfd, vnode_t *vp, offset_t *off) +dmu_send(objset_t *tosnap, objset_t *fromsnap, int outfd, vnode_t *vp, + offset_t *off) { dsl_dataset_t *ds = tosnap->os_dsl_dataset; dsl_dataset_t *fromds = fromsnap ? fromsnap->os_dsl_dataset : NULL; @@ -402,30 +441,13 @@ if (ds->ds_phys->ds_next_snap_obj == 0) return (EINVAL); - /* fromsnap must be an earlier snapshot from the same fs as tosnap */ - if (fromds && (ds->ds_dir != fromds->ds_dir || - fromds->ds_phys->ds_creation_txg >= ds->ds_phys->ds_creation_txg)) + /* + * fromsnap must be an earlier snapshot from the same fs as tosnap, + * or the origin's fs. + */ + if (fromds != NULL && !is_before(ds, fromds)) return (EXDEV); - if (fromorigin) { - dsl_pool_t *dp = ds->ds_dir->dd_pool; - - if (fromsnap) - return (EINVAL); - - if (dsl_dir_is_clone(ds->ds_dir)) { - rw_enter(&dp->dp_config_rwlock, RW_READER); - err = dsl_dataset_hold_obj(dp, - ds->ds_dir->dd_phys->dd_origin_obj, FTAG, &fromds); - rw_exit(&dp->dp_config_rwlock); - if (err) - return (err); - } else { - fromorigin = B_FALSE; - } - } - - drr = kmem_zalloc(sizeof (dmu_replay_record_t), KM_SLEEP); drr->drr_type = DRR_BEGIN; drr->drr_u.drr_begin.drr_magic = DMU_BACKUP_MAGIC; @@ -450,7 +472,7 @@ drr->drr_u.drr_begin.drr_creation_time = ds->ds_phys->ds_creation_time; drr->drr_u.drr_begin.drr_type = tosnap->os_phys->os_type; - if (fromorigin) + if (fromds != NULL && ds->ds_dir != fromds->ds_dir) drr->drr_u.drr_begin.drr_flags |= DRR_FLAG_CLONE; drr->drr_u.drr_begin.drr_toguid = ds->ds_phys->ds_guid; if (ds->ds_phys->ds_flags & DS_FLAG_CI_DATASET) @@ -462,8 +484,6 @@ if (fromds) fromtxg = fromds->ds_phys->ds_creation_txg; - if (fromorigin) - dsl_dataset_rele(fromds, FTAG); dsp = kmem_zalloc(sizeof (dmu_sendarg_t), KM_SLEEP); @@ -521,8 +541,7 @@ } int -dmu_send_estimate(objset_t *tosnap, objset_t *fromsnap, boolean_t fromorigin, - uint64_t *sizep) +dmu_send_estimate(objset_t *tosnap, objset_t *fromsnap, uint64_t *sizep) { dsl_dataset_t *ds = tosnap->os_dsl_dataset; dsl_dataset_t *fromds = fromsnap ? fromsnap->os_dsl_dataset : NULL; @@ -534,27 +553,13 @@ if (ds->ds_phys->ds_next_snap_obj == 0) return (EINVAL); - /* fromsnap must be an earlier snapshot from the same fs as tosnap */ - if (fromds && (ds->ds_dir != fromds->ds_dir || - fromds->ds_phys->ds_creation_txg >= ds->ds_phys->ds_creation_txg)) + /* + * fromsnap must be an earlier snapshot from the same fs as tosnap, + * or the origin's fs. + */ + if (fromds != NULL && !is_before(ds, fromds)) return (EXDEV); - if (fromorigin) { - if (fromsnap) - return (EINVAL); - - if (dsl_dir_is_clone(ds->ds_dir)) { - rw_enter(&dp->dp_config_rwlock, RW_READER); - err = dsl_dataset_hold_obj(dp, - ds->ds_dir->dd_phys->dd_origin_obj, FTAG, &fromds); - rw_exit(&dp->dp_config_rwlock); - if (err) - return (err); - } else { - fromorigin = B_FALSE; - } - } - /* Get uncompressed size estimate of changed data. */ if (fromds == NULL) { size = ds->ds_phys->ds_uncompressed_bytes; @@ -562,8 +567,6 @@ uint64_t used, comp; err = dsl_dataset_space_written(fromds, ds, &used, &comp, &size); - if (fromorigin) - dsl_dataset_rele(fromds, FTAG); if (err) return (err); } @@ -662,8 +665,7 @@ rbsa->ds, &rbsa->ds->ds_phys->ds_bp, rbsa->type, tx); } - spa_history_log_internal(LOG_DS_REPLAY_FULL_SYNC, - dd->dd_pool->dp_spa, tx, "dataset = %lld", dsobj); + spa_history_log_internal_ds(rbsa->ds, "receive new", tx, ""); } /* ARGSUSED */ @@ -764,8 +766,7 @@ rbsa->ds = cds; - spa_history_log_internal(LOG_DS_REPLAY_INC_SYNC, - dp->dp_spa, tx, "dataset = %lld", dsobj); + spa_history_log_internal_ds(cds, "receive over existing", tx, ""); } static boolean_t @@ -1573,6 +1574,7 @@ dmu_buf_will_dirty(ds->ds_dbuf, tx); ds->ds_phys->ds_flags &= ~DS_FLAG_INCONSISTENT; + spa_history_log_internal_ds(ds, "finished receiving", tx, ""); } static int diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/dmu_tx.c --- a/usr/src/uts/common/fs/zfs/dmu_tx.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/dmu_tx.c Sat Jun 30 13:00:47 2012 -0700 @@ -48,7 +48,7 @@ { dmu_tx_t *tx = kmem_zalloc(sizeof (dmu_tx_t), KM_SLEEP); tx->tx_dir = dd; - if (dd) + if (dd != NULL) tx->tx_pool = dd->dd_pool; list_create(&tx->tx_holds, sizeof (dmu_tx_hold_t), offsetof(dmu_tx_hold_t, txh_node)); diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/dsl_dataset.c --- a/usr/src/uts/common/fs/zfs/dsl_dataset.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/dsl_dataset.c Sat Jun 30 13:00:47 2012 -0700 @@ -914,7 +914,8 @@ * The snapshots must all be in the same pool. */ int -dmu_snapshots_destroy_nvl(nvlist_t *snaps, boolean_t defer, char *failed) +dmu_snapshots_destroy_nvl(nvlist_t *snaps, boolean_t defer, + nvlist_t *errlist) { int err; dsl_sync_task_t *dst; @@ -949,7 +950,7 @@ } else if (err == ENOENT) { err = 0; } else { - (void) strcpy(failed, nvpair_name(pair)); + fnvlist_add_int32(errlist, nvpair_name(pair), err); break; } } @@ -963,10 +964,12 @@ dsl_dataset_t *ds = dsda->ds; /* - * Return the file system name that triggered the error + * Return the snapshots that triggered the error. */ - if (dst->dst_err) { - dsl_dataset_name(ds, failed); + if (dst->dst_err != 0) { + char name[ZFS_MAXNAMELEN]; + dsl_dataset_name(ds, name); + fnvlist_add_int32(errlist, name, dst->dst_err); } ASSERT3P(dsda->rm_origin, ==, NULL); dsl_dataset_disown(ds, dstg); @@ -1045,7 +1048,6 @@ dsl_dir_t *dd; uint64_t obj; struct dsl_ds_destroyarg dsda = { 0 }; - dsl_dataset_t dummy_ds = { 0 }; dsda.ds = ds; @@ -1065,8 +1067,6 @@ } dd = ds->ds_dir; - dummy_ds.ds_dir = dd; - dummy_ds.ds_object = ds->ds_object; /* * Check for errors and mark this ds as inconsistent, in @@ -1153,7 +1153,7 @@ dsl_sync_task_create(dstg, dsl_dataset_destroy_check, dsl_dataset_destroy_sync, &dsda, tag, 0); dsl_sync_task_create(dstg, dsl_dir_destroy_check, - dsl_dir_destroy_sync, &dummy_ds, FTAG, 0); + dsl_dir_destroy_sync, dd, FTAG, 0); err = dsl_sync_task_group_wait(dstg); dsl_sync_task_group_destroy(dstg); @@ -1328,14 +1328,12 @@ dsl_dataset_destroy_begin_sync(void *arg1, void *arg2, dmu_tx_t *tx) { dsl_dataset_t *ds = arg1; - dsl_pool_t *dp = ds->ds_dir->dd_pool; /* Mark it as inconsistent on-disk, in case we crash */ dmu_buf_will_dirty(ds->ds_dbuf, tx); ds->ds_phys->ds_flags |= DS_FLAG_INCONSISTENT; - spa_history_log_internal(LOG_DS_DESTROY_BEGIN, dp->dp_spa, tx, - "dataset = %llu", ds->ds_object); + spa_history_log_internal_ds(ds, "destroy begin", tx, ""); } static int @@ -1660,9 +1658,13 @@ ASSERT(spa_version(dp->dp_spa) >= SPA_VERSION_USERREFS); dmu_buf_will_dirty(ds->ds_dbuf, tx); ds->ds_phys->ds_flags |= DS_FLAG_DEFER_DESTROY; + spa_history_log_internal_ds(ds, "defer_destroy", tx, ""); return; } + /* We need to log before removing it from the namespace. */ + spa_history_log_internal_ds(ds, "destroy", tx, ""); + /* signal any waiters that this dataset is going away */ mutex_enter(&ds->ds_lock); ds->ds_owner = dsl_reaper; @@ -1957,8 +1959,6 @@ dsl_dataset_rele(ds_prev, FTAG); spa_prop_clear_bootfs(dp->dp_spa, ds->ds_object, tx); - spa_history_log_internal(LOG_DS_DESTROY, dp->dp_spa, tx, - "dataset = %llu", ds->ds_object); if (ds->ds_phys->ds_next_clones_obj != 0) { uint64_t count; @@ -2006,7 +2006,7 @@ return (ENOSPC); /* - * Propogate any reserved space for this snapshot to other + * Propagate any reserved space for this snapshot to other * snapshot checks in this sync group. */ if (asize > 0) @@ -2016,10 +2016,9 @@ } int -dsl_dataset_snapshot_check(void *arg1, void *arg2, dmu_tx_t *tx) +dsl_dataset_snapshot_check(dsl_dataset_t *ds, const char *snapname, + dmu_tx_t *tx) { - dsl_dataset_t *ds = arg1; - const char *snapname = arg2; int err; uint64_t value; @@ -2031,7 +2030,7 @@ return (EAGAIN); /* - * Check for conflicting name snapshot name. + * Check for conflicting snapshot name. */ err = dsl_dataset_snap_lookup(ds, snapname, &value); if (err == 0) @@ -2055,10 +2054,9 @@ } void -dsl_dataset_snapshot_sync(void *arg1, void *arg2, dmu_tx_t *tx) +dsl_dataset_snapshot_sync(dsl_dataset_t *ds, const char *snapname, + dmu_tx_t *tx) { - dsl_dataset_t *ds = arg1; - const char *snapname = arg2; dsl_pool_t *dp = ds->ds_dir->dd_pool; dmu_buf_t *dbuf; dsl_dataset_phys_t *dsphys; @@ -2164,8 +2162,7 @@ dsl_dir_snap_cmtime_update(ds->ds_dir); - spa_history_log_internal(LOG_DS_SNAPSHOT, dp->dp_spa, tx, - "dataset = %llu", dsobj); + spa_history_log_internal_ds(ds->ds_prev, "snapshot", tx, ""); } void @@ -2252,7 +2249,20 @@ { uint64_t refd, avail, uobjs, aobjs, ratio; - dsl_dir_stats(ds->ds_dir, nv); + ratio = ds->ds_phys->ds_compressed_bytes == 0 ? 100 : + (ds->ds_phys->ds_uncompressed_bytes * 100 / + ds->ds_phys->ds_compressed_bytes); + + dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_REFRATIO, ratio); + + if (dsl_dataset_is_snapshot(ds)) { + dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_COMPRESSRATIO, ratio); + dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_USED, + ds->ds_phys->ds_unique_bytes); + get_clones_stat(ds, nv); + } else { + dsl_dir_stats(ds->ds_dir, nv); + } dsl_dataset_space(ds, &refd, &avail, &uobjs, &aobjs); dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_AVAILABLE, avail); @@ -2297,22 +2307,6 @@ } } - ratio = ds->ds_phys->ds_compressed_bytes == 0 ? 100 : - (ds->ds_phys->ds_uncompressed_bytes * 100 / - ds->ds_phys->ds_compressed_bytes); - dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_REFRATIO, ratio); - - if (ds->ds_phys->ds_next_snap_obj) { - /* - * This is a snapshot; override the dd's space used with - * our unique space and compression ratio. - */ - dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_USED, - ds->ds_phys->ds_unique_bytes); - dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_COMPRESSRATIO, ratio); - - get_clones_stat(ds, nv); - } } void @@ -2321,27 +2315,25 @@ stat->dds_creation_txg = ds->ds_phys->ds_creation_txg; stat->dds_inconsistent = ds->ds_phys->ds_flags & DS_FLAG_INCONSISTENT; stat->dds_guid = ds->ds_phys->ds_guid; - if (ds->ds_phys->ds_next_snap_obj) { + stat->dds_origin[0] = '\0'; + if (dsl_dataset_is_snapshot(ds)) { stat->dds_is_snapshot = B_TRUE; stat->dds_num_clones = ds->ds_phys->ds_num_children - 1; } else { stat->dds_is_snapshot = B_FALSE; stat->dds_num_clones = 0; + + rw_enter(&ds->ds_dir->dd_pool->dp_config_rwlock, RW_READER); + if (dsl_dir_is_clone(ds->ds_dir)) { + dsl_dataset_t *ods; + + VERIFY(0 == dsl_dataset_get_ref(ds->ds_dir->dd_pool, + ds->ds_dir->dd_phys->dd_origin_obj, FTAG, &ods)); + dsl_dataset_name(ods, stat->dds_origin); + dsl_dataset_drop_ref(ods, FTAG); + } + rw_exit(&ds->ds_dir->dd_pool->dp_config_rwlock); } - - /* clone origin is really a dsl_dir thing... */ - rw_enter(&ds->ds_dir->dd_pool->dp_config_rwlock, RW_READER); - if (dsl_dir_is_clone(ds->ds_dir)) { - dsl_dataset_t *ods; - - VERIFY(0 == dsl_dataset_get_ref(ds->ds_dir->dd_pool, - ds->ds_dir->dd_phys->dd_origin_obj, FTAG, &ods)); - dsl_dataset_name(ods, stat->dds_origin); - dsl_dataset_drop_ref(ods, FTAG); - } else { - stat->dds_origin[0] = '\0'; - } - rw_exit(&ds->ds_dir->dd_pool->dp_config_rwlock); } uint64_t @@ -2458,8 +2450,8 @@ ds->ds_snapname, 8, 1, &ds->ds_object, tx); ASSERT3U(err, ==, 0); - spa_history_log_internal(LOG_DS_RENAME, dd->dd_pool->dp_spa, tx, - "dataset = %llu", ds->ds_object); + spa_history_log_internal_ds(ds, "rename", tx, + "-> @%s", newsnapname); dsl_dataset_rele(hds, FTAG); } @@ -2939,8 +2931,7 @@ origin_ds->ds_phys->ds_unique_bytes = pa->unique; /* log history record */ - spa_history_log_internal(LOG_DS_PROMOTE, dd->dd_pool->dp_spa, tx, - "dataset = %llu", hds->ds_object); + spa_history_log_internal_ds(hds, "promote", tx, ""); dsl_dir_close(odd, FTAG); } @@ -3298,6 +3289,9 @@ csa->ohds->ds_phys->ds_deadlist_obj); dsl_scan_ds_clone_swapped(csa->ohds, csa->cds, tx); + + spa_history_log_internal_ds(csa->cds, "clone swap", tx, + "parent=%s", csa->ohds->ds_dir->dd_myname); } /* @@ -3454,9 +3448,8 @@ dmu_buf_will_dirty(ds->ds_dbuf, tx); ds->ds_quota = effective_value; - spa_history_log_internal(LOG_DS_REFQUOTA, - ds->ds_dir->dd_pool->dp_spa, tx, "%lld dataset = %llu ", - (longlong_t)ds->ds_quota, ds->ds_object); + spa_history_log_internal_ds(ds, "set refquota", tx, + "refquota=%lld", (longlong_t)ds->ds_quota); } } @@ -3561,9 +3554,8 @@ dsl_dir_diduse_space(ds->ds_dir, DD_USED_REFRSRV, delta, 0, 0, tx); mutex_exit(&ds->ds_dir->dd_lock); - spa_history_log_internal(LOG_DS_REFRESERV, - ds->ds_dir->dd_pool->dp_spa, tx, "%lld dataset = %llu", - (longlong_t)effective_value, ds->ds_object); + spa_history_log_internal_ds(ds, "set refreservation", tx, + "refreservation=%lld", (longlong_t)effective_value); } int @@ -3629,7 +3621,7 @@ { dsl_dataset_t *ds = arg1; struct dsl_ds_holdarg *ha = arg2; - char *htag = ha->htag; + const char *htag = ha->htag; objset_t *mos = ds->ds_dir->dd_pool->dp_meta_objset; int error = 0; @@ -3663,7 +3655,7 @@ { dsl_dataset_t *ds = arg1; struct dsl_ds_holdarg *ha = arg2; - char *htag = ha->htag; + const char *htag = ha->htag; dsl_pool_t *dp = ds->ds_dir->dd_pool; objset_t *mos = dp->dp_meta_objset; uint64_t now = gethrestime_sec(); @@ -3691,9 +3683,9 @@ htag, &now, tx)); } - spa_history_log_internal(LOG_DS_USER_HOLD, - dp->dp_spa, tx, "<%s> temp = %d dataset = %llu", htag, - (int)ha->temphold, ds->ds_object); + spa_history_log_internal_ds(ds, "hold", tx, + "tag = %s temp = %d holds now = %llu", + htag, (int)ha->temphold, ds->ds_userrefs); } static int @@ -3900,7 +3892,6 @@ dsl_pool_t *dp = ds->ds_dir->dd_pool; objset_t *mos = dp->dp_meta_objset; uint64_t zapobj; - uint64_t dsobj = ds->ds_object; uint64_t refs; int error; @@ -3923,9 +3914,8 @@ dsl_dataset_destroy_sync(&dsda, tag, tx); } - spa_history_log_internal(LOG_DS_USER_RELEASE, - dp->dp_spa, tx, "<%s> %lld dataset = %llu", - ra->htag, (longlong_t)refs, dsobj); + spa_history_log_internal_ds(ds, "release", tx, + "tag = %s refs now = %lld", ra->htag, (longlong_t)refs); } static int diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/dsl_deleg.c --- a/usr/src/uts/common/fs/zfs/dsl_deleg.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/dsl_deleg.c Sat Jun 30 13:00:47 2012 -0700 @@ -181,10 +181,8 @@ VERIFY(zap_update(mos, jumpobj, perm, 8, 1, &n, tx) == 0); - spa_history_log_internal(LOG_DS_PERM_UPDATE, - dd->dd_pool->dp_spa, tx, - "%s %s dataset = %llu", whokey, perm, - dd->dd_phys->dd_head_dataset_obj); + spa_history_log_internal_dd(dd, "permission update", tx, + "%s %s", whokey, perm); } } } @@ -213,10 +211,8 @@ (void) zap_remove(mos, zapobj, whokey, tx); VERIFY(0 == zap_destroy(mos, jumpobj, tx)); } - spa_history_log_internal(LOG_DS_PERM_WHO_REMOVE, - dd->dd_pool->dp_spa, tx, - "%s dataset = %llu", whokey, - dd->dd_phys->dd_head_dataset_obj); + spa_history_log_internal_dd(dd, "permission who remove", + tx, "%s", whokey); continue; } @@ -234,10 +230,8 @@ VERIFY(0 == zap_destroy(mos, jumpobj, tx)); } - spa_history_log_internal(LOG_DS_PERM_REMOVE, - dd->dd_pool->dp_spa, tx, - "%s %s dataset = %llu", whokey, perm, - dd->dd_phys->dd_head_dataset_obj); + spa_history_log_internal_dd(dd, "permission remove", tx, + "%s %s", whokey, perm); } } } @@ -524,12 +518,10 @@ } /* - * Check if user has requested permission. If descendent is set, must have - * descendent perms. + * Check if user has requested permission. */ int -dsl_deleg_access_impl(dsl_dataset_t *ds, boolean_t descendent, const char *perm, - cred_t *cr) +dsl_deleg_access_impl(dsl_dataset_t *ds, const char *perm, cred_t *cr) { dsl_dir_t *dd; dsl_pool_t *dp; @@ -550,7 +542,7 @@ SPA_VERSION_DELEGATED_PERMS) return (EPERM); - if (dsl_dataset_is_snapshot(ds) || descendent) { + if (dsl_dataset_is_snapshot(ds)) { /* * Snapshots are treated as descendents only, * local permissions do not apply. @@ -643,7 +635,7 @@ if (error) return (error); - error = dsl_deleg_access_impl(ds, B_FALSE, perm, cr); + error = dsl_deleg_access_impl(ds, perm, cr); dsl_dataset_rele(ds, FTAG); return (error); diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/dsl_dir.c --- a/usr/src/uts/common/fs/zfs/dsl_dir.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/dsl_dir.c Sat Jun 30 13:00:47 2012 -0700 @@ -20,6 +20,7 @@ */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ #include @@ -39,8 +40,8 @@ #include "zfs_namecheck.h" static uint64_t dsl_dir_space_towrite(dsl_dir_t *dd); -static void dsl_dir_set_reservation_sync(void *arg1, void *arg2, dmu_tx_t *tx); - +static void dsl_dir_set_reservation_sync_impl(dsl_dir_t *dd, + uint64_t value, dmu_tx_t *tx); /* ARGSUSED */ static void @@ -447,8 +448,7 @@ int dsl_dir_destroy_check(void *arg1, void *arg2, dmu_tx_t *tx) { - dsl_dataset_t *ds = arg1; - dsl_dir_t *dd = ds->ds_dir; + dsl_dir_t *dd = arg1; dsl_pool_t *dp = dd->dd_pool; objset_t *mos = dp->dp_meta_objset; int err; @@ -477,24 +477,19 @@ void dsl_dir_destroy_sync(void *arg1, void *tag, dmu_tx_t *tx) { - dsl_dataset_t *ds = arg1; - dsl_dir_t *dd = ds->ds_dir; + dsl_dir_t *dd = arg1; objset_t *mos = dd->dd_pool->dp_meta_objset; - dsl_prop_setarg_t psa; - uint64_t value = 0; uint64_t obj; dd_used_t t; ASSERT(RW_WRITE_HELD(&dd->dd_pool->dp_config_rwlock)); ASSERT(dd->dd_phys->dd_head_dataset_obj == 0); - /* Remove our reservation. */ - dsl_prop_setarg_init_uint64(&psa, "reservation", - (ZPROP_SRC_NONE | ZPROP_SRC_LOCAL | ZPROP_SRC_RECEIVED), - &value); - psa.psa_effective_value = 0; /* predict default value */ - - dsl_dir_set_reservation_sync(ds, &psa, tx); + /* + * Remove our reservation. The impl() routine avoids setting the + * actual property, which would require the (already destroyed) ds. + */ + dsl_dir_set_reservation_sync_impl(dd, 0, tx); ASSERT3U(dd->dd_phys->dd_used_bytes, ==, 0); ASSERT3U(dd->dd_phys->dd_reserved, ==, 0); @@ -1060,9 +1055,8 @@ dd->dd_phys->dd_quota = effective_value; mutex_exit(&dd->dd_lock); - spa_history_log_internal(LOG_DS_QUOTA, dd->dd_pool->dp_spa, - tx, "%lld dataset = %llu ", - (longlong_t)effective_value, dd->dd_phys->dd_head_dataset_obj); + spa_history_log_internal_dd(dd, "set quota", tx, + "quota=%lld", (longlong_t)effective_value); } int @@ -1149,25 +1143,17 @@ } static void -dsl_dir_set_reservation_sync(void *arg1, void *arg2, dmu_tx_t *tx) +dsl_dir_set_reservation_sync_impl(dsl_dir_t *dd, uint64_t value, dmu_tx_t *tx) { - dsl_dataset_t *ds = arg1; - dsl_dir_t *dd = ds->ds_dir; - dsl_prop_setarg_t *psa = arg2; - uint64_t effective_value = psa->psa_effective_value; uint64_t used; int64_t delta; - dsl_prop_set_sync(ds, psa, tx); - DSL_PROP_CHECK_PREDICTION(dd, psa); - dmu_buf_will_dirty(dd->dd_dbuf, tx); mutex_enter(&dd->dd_lock); used = dd->dd_phys->dd_used_bytes; - delta = MAX(used, effective_value) - - MAX(used, dd->dd_phys->dd_reserved); - dd->dd_phys->dd_reserved = effective_value; + delta = MAX(used, value) - MAX(used, dd->dd_phys->dd_reserved); + dd->dd_phys->dd_reserved = value; if (dd->dd_parent != NULL) { /* Roll up this additional usage into our ancestors */ @@ -1175,10 +1161,24 @@ delta, 0, 0, tx); } mutex_exit(&dd->dd_lock); +} - spa_history_log_internal(LOG_DS_RESERVATION, dd->dd_pool->dp_spa, - tx, "%lld dataset = %llu", - (longlong_t)effective_value, dd->dd_phys->dd_head_dataset_obj); + +static void +dsl_dir_set_reservation_sync(void *arg1, void *arg2, dmu_tx_t *tx) +{ + dsl_dataset_t *ds = arg1; + dsl_dir_t *dd = ds->ds_dir; + dsl_prop_setarg_t *psa = arg2; + uint64_t value = psa->psa_effective_value; + + dsl_prop_set_sync(ds, psa, tx); + DSL_PROP_CHECK_PREDICTION(dd, psa); + + dsl_dir_set_reservation_sync_impl(dd, value, tx); + + spa_history_log_internal_dd(dd, "set reservation", tx, + "reservation=%lld", (longlong_t)value); } int @@ -1299,9 +1299,15 @@ dsl_pool_t *dp = dd->dd_pool; objset_t *mos = dp->dp_meta_objset; int err; + char namebuf[MAXNAMELEN]; ASSERT(dmu_buf_refcount(dd->dd_dbuf) <= 2); + /* Log this before we change the name. */ + dsl_dir_name(ra->newparent, namebuf); + spa_history_log_internal_dd(dd, "rename", tx, + "-> %s/%s", namebuf, ra->mynewname); + if (ra->newparent != dd->dd_parent) { dsl_dir_diduse_space(dd->dd_parent, DD_USED_CHILD, -dd->dd_phys->dd_used_bytes, @@ -1341,8 +1347,6 @@ dd->dd_myname, 8, 1, &dd->dd_object, tx); ASSERT3U(err, ==, 0); - spa_history_log_internal(LOG_DS_RENAME, dd->dd_pool->dp_spa, - tx, "dataset = %llu", dd->dd_phys->dd_head_dataset_obj); } int diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/dsl_prop.c --- a/usr/src/uts/common/fs/zfs/dsl_prop.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/dsl_prop.c Sat Jun 30 13:00:47 2012 -0700 @@ -20,6 +20,7 @@ */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ #include @@ -702,11 +703,9 @@ } } - spa_history_log_internal((source == ZPROP_SRC_NONE || - source == ZPROP_SRC_INHERITED) ? LOG_DS_INHERIT : - LOG_DS_PROPSET, ds->ds_dir->dd_pool->dp_spa, tx, - "%s=%s dataset = %llu", propname, - (valstr == NULL ? "" : valstr), ds->ds_object); + spa_history_log_internal_ds(ds, (source == ZPROP_SRC_NONE || + source == ZPROP_SRC_INHERITED) ? "inherit" : "set", tx, + "%s=%s", propname, (valstr == NULL ? "" : valstr)); if (tbuf != NULL) kmem_free(tbuf, ZAP_MAXVALUELEN); @@ -755,24 +754,6 @@ } } -void -dsl_dir_prop_set_uint64_sync(dsl_dir_t *dd, const char *name, uint64_t val, - dmu_tx_t *tx) -{ - objset_t *mos = dd->dd_pool->dp_meta_objset; - uint64_t zapobj = dd->dd_phys->dd_props_zapobj; - - ASSERT(dmu_tx_is_syncing(tx)); - - VERIFY(0 == zap_update(mos, zapobj, name, sizeof (val), 1, &val, tx)); - - dsl_prop_changed_notify(dd->dd_pool, dd->dd_object, name, val, TRUE); - - spa_history_log_internal(LOG_DS_PROPSET, dd->dd_pool->dp_spa, tx, - "%s=%llu dataset = %llu", name, (u_longlong_t)val, - dd->dd_phys->dd_head_dataset_obj); -} - int dsl_prop_set(const char *dsname, const char *propname, zprop_source_t source, int intsz, int numints, const void *buf) diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/dsl_scan.c --- a/usr/src/uts/common/fs/zfs/dsl_scan.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/dsl_scan.c Sat Jun 30 13:00:47 2012 -0700 @@ -228,7 +228,7 @@ dsl_scan_sync_state(scn, tx); - spa_history_log_internal(LOG_POOL_SCAN, spa, tx, + spa_history_log_internal(spa, "scan setup", tx, "func=%u mintxg=%llu maxtxg=%llu", *funcp, scn->scn_phys.scn_min_txg, scn->scn_phys.scn_max_txg); } @@ -277,7 +277,7 @@ else scn->scn_phys.scn_state = DSS_CANCELED; - spa_history_log_internal(LOG_POOL_SCAN_DONE, spa, tx, + spa_history_log_internal(spa, "scan done", tx, "complete=%u", complete); if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) { diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/dsl_synctask.c --- a/usr/src/uts/common/fs/zfs/dsl_synctask.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/dsl_synctask.c Sat Jun 30 13:00:47 2012 -0700 @@ -20,6 +20,7 @@ */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ #include @@ -85,17 +86,17 @@ /* Do a preliminary error check. */ dstg->dstg_err = 0; +#ifdef ZFS_DEBUG + /* + * Only check half the time, otherwise, the sync-context + * check will almost never fail. + */ + if (spa_get_random(2) == 0) + goto skip; +#endif rw_enter(&dstg->dstg_pool->dp_config_rwlock, RW_READER); for (dst = list_head(&dstg->dstg_tasks); dst; dst = list_next(&dstg->dstg_tasks, dst)) { -#ifdef ZFS_DEBUG - /* - * Only check half the time, otherwise, the sync-context - * check will almost never fail. - */ - if (spa_get_random(2) == 0) - continue; -#endif dst->dst_err = dst->dst_checkfunc(dst->dst_arg1, dst->dst_arg2, tx); if (dst->dst_err) @@ -107,6 +108,7 @@ dmu_tx_commit(tx); return (dstg->dstg_err); } +skip: /* * We don't generally have many sync tasks, so pay the price of diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/rrwlock.c --- a/usr/src/uts/common/fs/zfs/rrwlock.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/rrwlock.c Sat Jun 30 13:00:47 2012 -0700 @@ -22,6 +22,9 @@ * Copyright 2009 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ +/* + * Copyright (c) 2012 by Delphix. All rights reserved. + */ #include #include @@ -262,3 +265,13 @@ return (held); } + +void +rrw_tsd_destroy(void *arg) +{ + rrw_node_t *rn = arg; + if (rn != NULL) { + panic("thread %p terminating with rrw lock %p held", + (void *)curthread, (void *)rn->rn_rrl); + } +} diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/spa.c --- a/usr/src/uts/common/fs/zfs/spa.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/spa.c Sat Jun 30 13:00:47 2012 -0700 @@ -2545,6 +2545,12 @@ spa_async_request(spa, SPA_ASYNC_RESILVER); /* + * Log the fact that we booted up (so that we can detect if + * we rebooted in the middle of an operation). + */ + spa_history_log_version(spa, "open"); + + /* * Delete any inconsistent datasets. */ (void) dmu_objset_find(spa_name(spa), @@ -3220,7 +3226,7 @@ */ int spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props, - const char *history_str, nvlist_t *zplprops) + nvlist_t *zplprops) { spa_t *spa; char *altroot = NULL; @@ -3439,9 +3445,7 @@ spa_config_sync(spa, B_FALSE, B_TRUE); - if (version >= SPA_VERSION_ZPOOL_HISTORY && history_str != NULL) - (void) spa_history_log(spa, history_str, LOG_CMD_POOL_CREATE); - spa_history_log_version(spa, LOG_POOL_CREATE); + spa_history_log_version(spa, "create"); spa->spa_minref = refcount_count(&spa->spa_refcount); @@ -3641,7 +3645,6 @@ } error = 0; - spa_history_log_version(spa, LOG_POOL_IMPORT); out: spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); vdev_free(rvd); @@ -3703,7 +3706,7 @@ spa_config_sync(spa, B_FALSE, B_TRUE); mutex_exit(&spa_namespace_lock); - spa_history_log_version(spa, LOG_POOL_IMPORT); + spa_history_log_version(spa, "import"); return (0); } @@ -3834,7 +3837,7 @@ spa_async_request(spa, SPA_ASYNC_AUTOEXPAND); mutex_exit(&spa_namespace_lock); - spa_history_log_version(spa, LOG_POOL_IMPORT); + spa_history_log_version(spa, "import"); return (0); } @@ -4372,7 +4375,7 @@ */ (void) spa_vdev_exit(spa, newrootvd, dtl_max_txg, 0); - spa_history_log_internal(LOG_POOL_VDEV_ATTACH, spa, NULL, + spa_history_log_internal(spa, "vdev attach", NULL, "%s vdev=%s %s vdev=%s", replacing && newvd_isspare ? "spare in" : replacing ? "replace" : "attach", newvdpath, @@ -4589,7 +4592,7 @@ error = spa_vdev_exit(spa, vd, txg, 0); - spa_history_log_internal(LOG_POOL_VDEV_DETACH, spa, NULL, + spa_history_log_internal(spa, "detach", NULL, "vdev=%s", vdpath); spa_strfree(vdpath); @@ -4858,9 +4861,8 @@ if (vml[c] != NULL) { vdev_split(vml[c]); if (error == 0) - spa_history_log_internal(LOG_POOL_VDEV_DETACH, - spa, tx, "vdev=%s", - vml[c]->vdev_path); + spa_history_log_internal(spa, "detach", tx, + "vdev=%s", vml[c]->vdev_path); vdev_free(vml[c]); } } @@ -4875,8 +4877,8 @@ zio_handle_panic_injection(spa, FTAG, 3); /* split is complete; log a history record */ - spa_history_log_internal(LOG_POOL_SPLIT, newspa, NULL, - "split new pool %s from pool %s", newname, spa_name(spa)); + spa_history_log_internal(newspa, "split", NULL, + "from pool %s", spa_name(spa)); kmem_free(vml, children * sizeof (vdev_t *)); @@ -5462,8 +5464,7 @@ * then log an internal history event. */ if (new_space != old_space) { - spa_history_log_internal(LOG_POOL_VDEV_ONLINE, - spa, NULL, + spa_history_log_internal(spa, "vdev online", NULL, "pool '%s' size: %llu(+%llu)", spa_name(spa), new_space, new_space - old_space); } @@ -5699,6 +5700,7 @@ spa->spa_uberblock.ub_version = version; vdev_config_dirty(spa->spa_root_vdev); + spa_history_log_internal(spa, "set", tx, "version=%lld", version); } /* @@ -5733,6 +5735,8 @@ VERIFY3U(0, ==, zfeature_lookup_name(fname, &feature)); spa_feature_enable(spa, feature, tx); + spa_history_log_internal(spa, "set", tx, + "%s=enabled", nvpair_name(elem)); break; case ZPOOL_PROP_VERSION: @@ -5772,6 +5776,8 @@ */ if (tx->tx_txg != TXG_INITIAL) vdev_config_dirty(spa->spa_root_vdev); + spa_history_log_internal(spa, "set", tx, + "%s=%s", nvpair_name(elem), strval); break; default: /* @@ -5794,7 +5800,8 @@ VERIFY(zap_update(mos, spa->spa_pool_props_object, propname, 1, strlen(strval) + 1, strval, tx) == 0); - + spa_history_log_internal(spa, "set", tx, + "%s=%s", nvpair_name(elem), strval); } else if (nvpair_type(elem) == DATA_TYPE_UINT64) { VERIFY(nvpair_value_uint64(elem, &intval) == 0); @@ -5806,6 +5813,8 @@ VERIFY(zap_update(mos, spa->spa_pool_props_object, propname, 8, 1, &intval, tx) == 0); + spa_history_log_internal(spa, "set", tx, + "%s=%lld", nvpair_name(elem), intval); } else { ASSERT(0); /* not allowed */ } @@ -5834,13 +5843,6 @@ } } - /* log internal history if this is not a zpool create */ - if (spa_version(spa) >= SPA_VERSION_ZPOOL_HISTORY && - tx->tx_txg != TXG_INITIAL) { - spa_history_log_internal(LOG_POOL_PROPSET, - spa, tx, "%s %lld %s", - nvpair_name(elem), intval, spa_name(spa)); - } } mutex_exit(&spa->spa_props_lock); diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/spa_history.c --- a/usr/src/uts/common/fs/zfs/spa_history.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/spa_history.c Sat Jun 30 13:00:47 2012 -0700 @@ -21,7 +21,7 @@ /* * Copyright (c) 2006, 2010, Oracle and/or its affiliates. All rights reserved. - * Copyright (c) 2011 by Delphix. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ #include @@ -30,9 +30,12 @@ #include #include #include +#include +#include #include #include #include +#include #include "zfs_comutil.h" #ifdef _KERNEL #include @@ -176,12 +179,14 @@ } static char * -spa_history_zone() +spa_history_zone(void) { #ifdef _KERNEL + if (INGLOBALZONE(curproc)) + return (NULL); return (curproc->p_zone->zone_name); #else - return ("global"); + return (NULL); #endif } @@ -193,14 +198,12 @@ spa_history_log_sync(void *arg1, void *arg2, dmu_tx_t *tx) { spa_t *spa = arg1; - history_arg_t *hap = arg2; - const char *history_str = hap->ha_history_str; + nvlist_t *nvl = arg2; objset_t *mos = spa->spa_meta_objset; dmu_buf_t *dbp; spa_history_phys_t *shpp; size_t reclen; uint64_t le_len; - nvlist_t *nvrecord; char *record_packed = NULL; int ret; @@ -230,46 +233,35 @@ } #endif - VERIFY(nvlist_alloc(&nvrecord, NV_UNIQUE_NAME, KM_SLEEP) == 0); - VERIFY(nvlist_add_uint64(nvrecord, ZPOOL_HIST_TIME, - gethrestime_sec()) == 0); - VERIFY(nvlist_add_uint64(nvrecord, ZPOOL_HIST_WHO, hap->ha_uid) == 0); - if (hap->ha_zone != NULL) - VERIFY(nvlist_add_string(nvrecord, ZPOOL_HIST_ZONE, - hap->ha_zone) == 0); + fnvlist_add_uint64(nvl, ZPOOL_HIST_TIME, gethrestime_sec()); #ifdef _KERNEL - VERIFY(nvlist_add_string(nvrecord, ZPOOL_HIST_HOST, - utsname.nodename) == 0); + fnvlist_add_string(nvl, ZPOOL_HIST_HOST, utsname.nodename); #endif - if (hap->ha_log_type == LOG_CMD_POOL_CREATE || - hap->ha_log_type == LOG_CMD_NORMAL) { - VERIFY(nvlist_add_string(nvrecord, ZPOOL_HIST_CMD, - history_str) == 0); - - zfs_dbgmsg("command: %s", history_str); - } else { - VERIFY(nvlist_add_uint64(nvrecord, ZPOOL_HIST_INT_EVENT, - hap->ha_event) == 0); - VERIFY(nvlist_add_uint64(nvrecord, ZPOOL_HIST_TXG, - tx->tx_txg) == 0); - VERIFY(nvlist_add_string(nvrecord, ZPOOL_HIST_INT_STR, - history_str) == 0); - - zfs_dbgmsg("internal %s pool:%s txg:%llu %s", - zfs_history_event_names[hap->ha_event], spa_name(spa), - (longlong_t)tx->tx_txg, history_str); - + if (nvlist_exists(nvl, ZPOOL_HIST_CMD)) { + zfs_dbgmsg("command: %s", + fnvlist_lookup_string(nvl, ZPOOL_HIST_CMD)); + } else if (nvlist_exists(nvl, ZPOOL_HIST_INT_NAME)) { + if (nvlist_exists(nvl, ZPOOL_HIST_DSNAME)) { + zfs_dbgmsg("txg %lld %s %s (id %llu) %s", + fnvlist_lookup_uint64(nvl, ZPOOL_HIST_TXG), + fnvlist_lookup_string(nvl, ZPOOL_HIST_INT_NAME), + fnvlist_lookup_string(nvl, ZPOOL_HIST_DSNAME), + fnvlist_lookup_uint64(nvl, ZPOOL_HIST_DSID), + fnvlist_lookup_string(nvl, ZPOOL_HIST_INT_STR)); + } else { + zfs_dbgmsg("txg %lld %s %s", + fnvlist_lookup_uint64(nvl, ZPOOL_HIST_TXG), + fnvlist_lookup_string(nvl, ZPOOL_HIST_INT_NAME), + fnvlist_lookup_string(nvl, ZPOOL_HIST_INT_STR)); + } + } else if (nvlist_exists(nvl, ZPOOL_HIST_IOCTL)) { + zfs_dbgmsg("ioctl %s", + fnvlist_lookup_string(nvl, ZPOOL_HIST_IOCTL)); } - VERIFY(nvlist_size(nvrecord, &reclen, NV_ENCODE_XDR) == 0); - record_packed = kmem_alloc(reclen, KM_SLEEP); - - VERIFY(nvlist_pack(nvrecord, &record_packed, &reclen, - NV_ENCODE_XDR, KM_SLEEP) == 0); + record_packed = fnvlist_pack(nvl, &reclen); mutex_enter(&spa->spa_history_lock); - if (hap->ha_log_type == LOG_CMD_POOL_CREATE) - VERIFY(shpp->sh_eof == shpp->sh_pool_create_len); /* write out the packed length as little endian */ le_len = LE_64((uint64_t)reclen); @@ -277,33 +269,42 @@ if (!ret) ret = spa_history_write(spa, record_packed, reclen, shpp, tx); - if (!ret && hap->ha_log_type == LOG_CMD_POOL_CREATE) { - shpp->sh_pool_create_len += sizeof (le_len) + reclen; - shpp->sh_bof = shpp->sh_pool_create_len; + /* The first command is the create, which we keep forever */ + if (ret == 0 && shpp->sh_pool_create_len == 0 && + nvlist_exists(nvl, ZPOOL_HIST_CMD)) { + shpp->sh_pool_create_len = shpp->sh_bof = shpp->sh_eof; } mutex_exit(&spa->spa_history_lock); - nvlist_free(nvrecord); - kmem_free(record_packed, reclen); + fnvlist_pack_free(record_packed, reclen); dmu_buf_rele(dbp, FTAG); - - strfree(hap->ha_history_str); - if (hap->ha_zone != NULL) - strfree(hap->ha_zone); - kmem_free(hap, sizeof (history_arg_t)); + fnvlist_free(nvl); } /* * Write out a history event. */ int -spa_history_log(spa_t *spa, const char *history_str, history_log_type_t what) +spa_history_log(spa_t *spa, const char *msg) { - history_arg_t *ha; + int err; + nvlist_t *nvl = fnvlist_alloc(); + + fnvlist_add_string(nvl, ZPOOL_HIST_CMD, msg); + err = spa_history_log_nvl(spa, nvl); + fnvlist_free(nvl); + return (err); +} + +int +spa_history_log_nvl(spa_t *spa, nvlist_t *nvl) +{ int err = 0; dmu_tx_t *tx; + nvlist_t *nvarg; - ASSERT(what != LOG_INTERNAL); + if (spa_version(spa) < SPA_VERSION_ZPOOL_HISTORY) + return (EINVAL); tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir); err = dmu_tx_assign(tx, TXG_WAIT); @@ -312,19 +313,21 @@ return (err); } - ha = kmem_alloc(sizeof (history_arg_t), KM_SLEEP); - ha->ha_history_str = strdup(history_str); - ha->ha_zone = strdup(spa_history_zone()); - ha->ha_log_type = what; - ha->ha_uid = crgetuid(CRED()); + nvarg = fnvlist_dup(nvl); + if (spa_history_zone() != NULL) { + fnvlist_add_string(nvarg, ZPOOL_HIST_ZONE, + spa_history_zone()); + } + fnvlist_add_uint64(nvarg, ZPOOL_HIST_WHO, crgetruid(CRED())); /* Kick this off asynchronously; errors are ignored. */ dsl_sync_task_do_nowait(spa_get_dsl(spa), NULL, - spa_history_log_sync, spa, ha, 0, tx); + spa_history_log_sync, spa, nvarg, 0, tx); dmu_tx_commit(tx); - /* spa_history_log_sync will free ha and strings */ + /* spa_history_log_sync will free nvl */ return (err); + } /* @@ -341,7 +344,7 @@ int err; /* - * If the command history doesn't exist (older pool), + * If the command history doesn't exist (older pool), * that's ok, just return ENOENT. */ if (!spa->spa_history) @@ -424,11 +427,14 @@ return (err); } +/* + * The nvlist will be consumed by this call. + */ static void -log_internal(history_internal_events_t event, spa_t *spa, +log_internal(nvlist_t *nvl, const char *operation, spa_t *spa, dmu_tx_t *tx, const char *fmt, va_list adx) { - history_arg_t *ha; + char *msg; /* * If this is part of creating a pool, not everything is @@ -437,28 +443,25 @@ if (tx->tx_txg == TXG_INITIAL) return; - ha = kmem_alloc(sizeof (history_arg_t), KM_SLEEP); - ha->ha_history_str = kmem_alloc(vsnprintf(NULL, 0, fmt, adx) + 1, - KM_SLEEP); + msg = kmem_alloc(vsnprintf(NULL, 0, fmt, adx) + 1, KM_SLEEP); + (void) vsprintf(msg, fmt, adx); + fnvlist_add_string(nvl, ZPOOL_HIST_INT_STR, msg); + strfree(msg); - (void) vsprintf(ha->ha_history_str, fmt, adx); - - ha->ha_log_type = LOG_INTERNAL; - ha->ha_event = event; - ha->ha_zone = NULL; - ha->ha_uid = 0; + fnvlist_add_string(nvl, ZPOOL_HIST_INT_NAME, operation); + fnvlist_add_uint64(nvl, ZPOOL_HIST_TXG, tx->tx_txg); if (dmu_tx_is_syncing(tx)) { - spa_history_log_sync(spa, ha, tx); + spa_history_log_sync(spa, nvl, tx); } else { dsl_sync_task_do_nowait(spa_get_dsl(spa), NULL, - spa_history_log_sync, spa, ha, 0, tx); + spa_history_log_sync, spa, nvl, 0, tx); } - /* spa_history_log_sync() will free ha and strings */ + /* spa_history_log_sync() will free nvl */ } void -spa_history_log_internal(history_internal_events_t event, spa_t *spa, +spa_history_log_internal(spa_t *spa, const char *operation, dmu_tx_t *tx, const char *fmt, ...) { dmu_tx_t *htx = tx; @@ -474,7 +477,7 @@ } va_start(adx, fmt); - log_internal(event, spa, htx, fmt, adx); + log_internal(fnvlist_alloc(), operation, spa, htx, fmt, adx); va_end(adx); /* if we didn't get a tx from the caller, commit the one we made */ @@ -483,21 +486,56 @@ } void -spa_history_log_version(spa_t *spa, history_internal_events_t event) +spa_history_log_internal_ds(dsl_dataset_t *ds, const char *operation, + dmu_tx_t *tx, const char *fmt, ...) +{ + va_list adx; + char namebuf[MAXNAMELEN]; + nvlist_t *nvl = fnvlist_alloc(); + + ASSERT(tx != NULL); + + dsl_dataset_name(ds, namebuf); + fnvlist_add_string(nvl, ZPOOL_HIST_DSNAME, namebuf); + fnvlist_add_uint64(nvl, ZPOOL_HIST_DSID, ds->ds_object); + + va_start(adx, fmt); + log_internal(nvl, operation, dsl_dataset_get_spa(ds), tx, fmt, adx); + va_end(adx); +} + +void +spa_history_log_internal_dd(dsl_dir_t *dd, const char *operation, + dmu_tx_t *tx, const char *fmt, ...) +{ + va_list adx; + char namebuf[MAXNAMELEN]; + nvlist_t *nvl = fnvlist_alloc(); + + ASSERT(tx != NULL); + + dsl_dir_name(dd, namebuf); + fnvlist_add_string(nvl, ZPOOL_HIST_DSNAME, namebuf); + fnvlist_add_uint64(nvl, ZPOOL_HIST_DSID, + dd->dd_phys->dd_head_dataset_obj); + + va_start(adx, fmt); + log_internal(nvl, operation, dd->dd_pool->dp_spa, tx, fmt, adx); + va_end(adx); +} + +void +spa_history_log_version(spa_t *spa, const char *operation) { #ifdef _KERNEL uint64_t current_vers = spa_version(spa); - if (current_vers >= SPA_VERSION_ZPOOL_HISTORY) { - spa_history_log_internal(event, spa, NULL, - "pool spa %llu; zfs spa %llu; zpl %d; uts %s %s %s %s", - (u_longlong_t)current_vers, SPA_VERSION, ZPL_VERSION, - utsname.nodename, utsname.release, utsname.version, - utsname.machine); - } - cmn_err(CE_CONT, "!%s version %llu pool %s using %llu", - event == LOG_POOL_IMPORT ? "imported" : - event == LOG_POOL_CREATE ? "created" : "accessed", + spa_history_log_internal(spa, operation, NULL, + "pool version %llu; software version %llu/%d; uts %s %s %s %s", + (u_longlong_t)current_vers, SPA_VERSION, ZPL_VERSION, + utsname.nodename, utsname.release, utsname.version, + utsname.machine); + cmn_err(CE_CONT, "!%s version %llu pool %s using %llu", operation, (u_longlong_t)current_vers, spa_name(spa), SPA_VERSION); #endif } diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/sys/dmu.h --- a/usr/src/uts/common/fs/zfs/sys/dmu.h Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/sys/dmu.h Sat Jun 30 13:00:47 2012 -0700 @@ -44,6 +44,7 @@ #include #include #include +#include #ifdef __cplusplus extern "C" { @@ -216,16 +217,6 @@ DMU_OTN_ZAP_METADATA = DMU_OT(DMU_BSWAP_ZAP, B_TRUE), } dmu_object_type_t; -typedef enum dmu_objset_type { - DMU_OST_NONE, - DMU_OST_META, - DMU_OST_ZFS, - DMU_OST_ZVOL, - DMU_OST_OTHER, /* For testing only! */ - DMU_OST_ANY, /* Be careful! */ - DMU_OST_NUMTYPES -} dmu_objset_type_t; - void byteswap_uint64_array(void *buf, size_t size); void byteswap_uint32_array(void *buf, size_t size); void byteswap_uint16_array(void *buf, size_t size); @@ -270,9 +261,11 @@ int dmu_objset_clone(const char *name, struct dsl_dataset *clone_origin, uint64_t flags); int dmu_objset_destroy(const char *name, boolean_t defer); -int dmu_snapshots_destroy_nvl(struct nvlist *snaps, boolean_t defer, char *); -int dmu_objset_snapshot(char *fsname, char *snapname, char *tag, - struct nvlist *props, boolean_t recursive, boolean_t temporary, int fd); +int dmu_snapshots_destroy_nvl(struct nvlist *snaps, boolean_t defer, + struct nvlist *errlist); +int dmu_objset_snapshot(struct nvlist *snaps, struct nvlist *, struct nvlist *); +int dmu_objset_snapshot_one(const char *fsname, const char *snapname); +int dmu_objset_snapshot_tmp(const char *, const char *, int); int dmu_objset_rename(const char *name, const char *newname, boolean_t recursive); int dmu_objset_find(char *name, int func(const char *, void *), void *arg, @@ -789,10 +782,9 @@ void dmu_traverse_objset(objset_t *os, uint64_t txg_start, dmu_traverse_cb_t cb, void *arg); -int dmu_send(objset_t *tosnap, objset_t *fromsnap, boolean_t fromorigin, +int dmu_send(objset_t *tosnap, objset_t *fromsnap, int outfd, struct vnode *vp, offset_t *off); -int dmu_send_estimate(objset_t *tosnap, objset_t *fromsnap, boolean_t fromorign, - uint64_t *sizep); +int dmu_send_estimate(objset_t *tosnap, objset_t *fromsnap, uint64_t *sizep); typedef struct dmu_recv_cookie { /* diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/sys/dmu_objset.h --- a/usr/src/uts/common/fs/zfs/sys/dmu_objset.h Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/sys/dmu_objset.h Sat Jun 30 13:00:47 2012 -0700 @@ -20,6 +20,7 @@ */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ /* Portions Copyright 2010 Robert Milkowski */ @@ -137,24 +138,14 @@ void dmu_objset_disown(objset_t *os, void *tag); int dmu_objset_from_ds(struct dsl_dataset *ds, objset_t **osp); -int dmu_objset_create(const char *name, dmu_objset_type_t type, uint64_t flags, - void (*func)(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx), void *arg); -int dmu_objset_clone(const char *name, struct dsl_dataset *clone_origin, - uint64_t flags); -int dmu_objset_destroy(const char *name, boolean_t defer); -int dmu_objset_snapshot(char *fsname, char *snapname, char *tag, - struct nvlist *props, boolean_t recursive, boolean_t temporary, int fd); void dmu_objset_stats(objset_t *os, nvlist_t *nv); void dmu_objset_fast_stat(objset_t *os, dmu_objset_stats_t *stat); void dmu_objset_space(objset_t *os, uint64_t *refdbytesp, uint64_t *availbytesp, uint64_t *usedobjsp, uint64_t *availobjsp); uint64_t dmu_objset_fsid_guid(objset_t *os); -int dmu_objset_find(char *name, int func(const char *, void *), void *arg, - int flags); int dmu_objset_find_spa(spa_t *spa, const char *name, int func(spa_t *, uint64_t, const char *, void *), void *arg, int flags); int dmu_objset_prefetch(const char *name, void *arg); -void dmu_objset_byteswap(void *buf, size_t size); int dmu_objset_evict_dbufs(objset_t *os); timestruc_t dmu_objset_snap_cmtime(objset_t *os); diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/sys/dsl_dataset.h --- a/usr/src/uts/common/fs/zfs/sys/dsl_dataset.h Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/sys/dsl_dataset.h Sat Jun 30 13:00:47 2012 -0700 @@ -180,7 +180,7 @@ struct dsl_ds_holdarg { dsl_sync_task_group_t *dstg; - char *htag; + const char *htag; char *snapname; boolean_t recursive; boolean_t gotone; @@ -215,12 +215,11 @@ uint64_t dsl_dataset_create_sync_dd(dsl_dir_t *dd, dsl_dataset_t *origin, uint64_t flags, dmu_tx_t *tx); int dsl_dataset_destroy(dsl_dataset_t *ds, void *tag, boolean_t defer); -int dsl_snapshots_destroy(char *fsname, char *snapname, boolean_t defer); dsl_checkfunc_t dsl_dataset_destroy_check; dsl_syncfunc_t dsl_dataset_destroy_sync; -dsl_checkfunc_t dsl_dataset_snapshot_check; -dsl_syncfunc_t dsl_dataset_snapshot_sync; dsl_syncfunc_t dsl_dataset_user_hold_sync; +int dsl_dataset_snapshot_check(dsl_dataset_t *ds, const char *, dmu_tx_t *tx); +void dsl_dataset_snapshot_sync(dsl_dataset_t *ds, const char *, dmu_tx_t *tx); int dsl_dataset_rename(char *name, const char *newname, boolean_t recursive); int dsl_dataset_promote(const char *name, char *conflsnap); int dsl_dataset_clone_swap(dsl_dataset_t *clone, dsl_dataset_t *origin_head, diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/sys/dsl_deleg.h --- a/usr/src/uts/common/fs/zfs/sys/dsl_deleg.h Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/sys/dsl_deleg.h Sat Jun 30 13:00:47 2012 -0700 @@ -20,7 +20,7 @@ */ /* * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved. - * Copyright (c) 2011 by Delphix. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ #ifndef _SYS_DSL_DELEG_H @@ -65,8 +65,7 @@ int dsl_deleg_get(const char *ddname, nvlist_t **nvp); int dsl_deleg_set(const char *ddname, nvlist_t *nvp, boolean_t unset); int dsl_deleg_access(const char *ddname, const char *perm, cred_t *cr); -int dsl_deleg_access_impl(struct dsl_dataset *ds, boolean_t descendent, - const char *perm, cred_t *cr); +int dsl_deleg_access_impl(struct dsl_dataset *ds, const char *perm, cred_t *cr); void dsl_deleg_set_create_perms(dsl_dir_t *dd, dmu_tx_t *tx, cred_t *cr); int dsl_deleg_can_allow(char *ddname, nvlist_t *nvp, cred_t *cr); int dsl_deleg_can_unallow(char *ddname, nvlist_t *nvp, cred_t *cr); diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/sys/dsl_prop.h --- a/usr/src/uts/common/fs/zfs/sys/dsl_prop.h Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/sys/dsl_prop.h Sat Jun 30 13:00:47 2012 -0700 @@ -20,6 +20,7 @@ */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ #ifndef _SYS_DSL_PROP_H @@ -89,8 +90,6 @@ int dsl_prop_set(const char *ddname, const char *propname, zprop_source_t source, int intsz, int numints, const void *buf); int dsl_props_set(const char *dsname, zprop_source_t source, nvlist_t *nvl); -void dsl_dir_prop_set_uint64_sync(dsl_dir_t *dd, const char *name, uint64_t val, - dmu_tx_t *tx); void dsl_prop_setarg_init_uint64(dsl_prop_setarg_t *psa, const char *propname, zprop_source_t source, uint64_t *value); diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/sys/rrwlock.h --- a/usr/src/uts/common/fs/zfs/sys/rrwlock.h Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/sys/rrwlock.h Sat Jun 30 13:00:47 2012 -0700 @@ -22,12 +22,13 @@ * Copyright 2007 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ +/* + * Copyright (c) 2012 by Delphix. All rights reserved. + */ #ifndef _SYS_RR_RW_LOCK_H #define _SYS_RR_RW_LOCK_H -#pragma ident "%Z%%M% %I% %E% SMI" - #ifdef __cplusplus extern "C" { #endif @@ -69,6 +70,7 @@ void rrw_enter(rrwlock_t *rrl, krw_t rw, void *tag); void rrw_exit(rrwlock_t *rrl, void *tag); boolean_t rrw_held(rrwlock_t *rrl, krw_t rw); +void rrw_tsd_destroy(void *arg); #define RRW_READ_HELD(x) rrw_held(x, RW_READER) #define RRW_WRITE_HELD(x) rrw_held(x, RW_WRITER) diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/sys/spa.h --- a/usr/src/uts/common/fs/zfs/sys/spa.h Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/sys/spa.h Sat Jun 30 13:00:47 2012 -0700 @@ -52,6 +52,7 @@ typedef struct ddt ddt_t; typedef struct ddt_entry ddt_entry_t; struct dsl_pool; +struct dsl_dataset; /* * General-purpose 32-bit and 64-bit bitfield encodings. @@ -418,7 +419,7 @@ extern int spa_get_stats(const char *pool, nvlist_t **config, char *altroot, size_t buflen); extern int spa_create(const char *pool, nvlist_t *config, nvlist_t *props, - const char *history_str, nvlist_t *zplprops); + nvlist_t *zplprops); extern int spa_import_rootpool(char *devpath, char *devid); extern int spa_import(const char *pool, nvlist_t *config, nvlist_t *props, uint64_t flags); @@ -632,31 +633,20 @@ extern int spa_mode(spa_t *spa); extern uint64_t strtonum(const char *str, char **nptr); -/* history logging */ -typedef enum history_log_type { - LOG_CMD_POOL_CREATE, - LOG_CMD_NORMAL, - LOG_INTERNAL -} history_log_type_t; - -typedef struct history_arg { - char *ha_history_str; - history_log_type_t ha_log_type; - history_internal_events_t ha_event; - char *ha_zone; - uid_t ha_uid; -} history_arg_t; - extern char *spa_his_ievent_table[]; extern void spa_history_create_obj(spa_t *spa, dmu_tx_t *tx); extern int spa_history_get(spa_t *spa, uint64_t *offset, uint64_t *len_read, char *his_buf); -extern int spa_history_log(spa_t *spa, const char *his_buf, - history_log_type_t what); -extern void spa_history_log_internal(history_internal_events_t event, - spa_t *spa, dmu_tx_t *tx, const char *fmt, ...); -extern void spa_history_log_version(spa_t *spa, history_internal_events_t evt); +extern int spa_history_log(spa_t *spa, const char *his_buf); +extern int spa_history_log_nvl(spa_t *spa, nvlist_t *nvl); +extern void spa_history_log_version(spa_t *spa, const char *operation); +extern void spa_history_log_internal(spa_t *spa, const char *operation, + dmu_tx_t *tx, const char *fmt, ...); +extern void spa_history_log_internal_ds(struct dsl_dataset *ds, const char *op, + dmu_tx_t *tx, const char *fmt, ...); +extern void spa_history_log_internal_dd(dsl_dir_t *dd, const char *operation, + dmu_tx_t *tx, const char *fmt, ...); /* error handling */ struct zbookmark; diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/sys/zfs_ioctl.h --- a/usr/src/uts/common/fs/zfs/sys/zfs_ioctl.h Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/sys/zfs_ioctl.h Sat Jun 30 13:00:47 2012 -0700 @@ -20,6 +20,7 @@ */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ #ifndef _SYS_ZFS_IOCTL_H @@ -41,6 +42,15 @@ #endif /* + * The structures in this file are passed between userland and the + * kernel. Userland may be running a 32-bit process, while the kernel + * is 64-bit. Therefore, these structures need to compile the same in + * 32-bit and 64-bit. This means not using type "long", and adding + * explicit padding so that the 32-bit structure will not be packed more + * tightly than the 64-bit structure (which requires 64-bit alignment). + */ + +/* * Property values for snapdir */ #define ZFS_SNAPDIR_HIDDEN 0 @@ -256,22 +266,29 @@ } zfs_case_t; typedef struct zfs_cmd { - char zc_name[MAXPATHLEN]; + char zc_name[MAXPATHLEN]; /* name of pool or dataset */ + uint64_t zc_nvlist_src; /* really (char *) */ + uint64_t zc_nvlist_src_size; + uint64_t zc_nvlist_dst; /* really (char *) */ + uint64_t zc_nvlist_dst_size; + boolean_t zc_nvlist_dst_filled; /* put an nvlist in dst? */ + int zc_pad2; + + /* + * The following members are for legacy ioctls which haven't been + * converted to the new method. + */ + uint64_t zc_history; /* really (char *) */ char zc_value[MAXPATHLEN * 2]; char zc_string[MAXNAMELEN]; char zc_top_ds[MAXPATHLEN]; uint64_t zc_guid; uint64_t zc_nvlist_conf; /* really (char *) */ uint64_t zc_nvlist_conf_size; - uint64_t zc_nvlist_src; /* really (char *) */ - uint64_t zc_nvlist_src_size; - uint64_t zc_nvlist_dst; /* really (char *) */ - uint64_t zc_nvlist_dst_size; uint64_t zc_cookie; uint64_t zc_objset_type; uint64_t zc_perm_action; - uint64_t zc_history; /* really (char *) */ - uint64_t zc_history_len; + uint64_t zc_history_len; uint64_t zc_history_offset; uint64_t zc_obj; uint64_t zc_iflags; /* internal to zfs(7fs) */ diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/zfs_ctldir.c --- a/usr/src/uts/common/fs/zfs/zfs_ctldir.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/zfs_ctldir.c Sat Jun 30 13:00:47 2012 -0700 @@ -20,6 +20,7 @@ */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ /* @@ -749,8 +750,7 @@ return (err); if (err == 0) { - err = dmu_objset_snapshot(name, dirname, NULL, NULL, - B_FALSE, B_FALSE, -1); + err = dmu_objset_snapshot_one(name, dirname); if (err) return (err); err = lookupnameat(dirname, seg, follow, NULL, vpp, dvp); diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/zfs_ioctl.c --- a/usr/src/uts/common/fs/zfs/zfs_ioctl.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/zfs_ioctl.c Sat Jun 30 13:00:47 2012 -0700 @@ -23,8 +23,109 @@ * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Portions Copyright 2011 Martin Matuska * Copyright 2011 Nexenta Systems, Inc. All rights reserved. + * Copyright (c) 2012, Joyent, Inc. All rights reserved. * Copyright (c) 2012 by Delphix. All rights reserved. - * Copyright (c) 2012, Joyent, Inc. All rights reserved. + */ + +/* + * ZFS ioctls. + * + * This file handles the ioctls to /dev/zfs, used for configuring ZFS storage + * pools and filesystems, e.g. with /sbin/zfs and /sbin/zpool. + * + * There are two ways that we handle ioctls: the legacy way where almost + * all of the logic is in the ioctl callback, and the new way where most + * of the marshalling is handled in the common entry point, zfsdev_ioctl(). + * + * Non-legacy ioctls should be registered by calling + * zfs_ioctl_register() from zfs_ioctl_init(). The ioctl is invoked + * from userland by lzc_ioctl(). + * + * The registration arguments are as follows: + * + * const char *name + * The name of the ioctl. This is used for history logging. If the + * ioctl returns successfully (the callback returns 0), and allow_log + * is true, then a history log entry will be recorded with the input & + * output nvlists. The log entry can be printed with "zpool history -i". + * + * zfs_ioc_t ioc + * The ioctl request number, which userland will pass to ioctl(2). + * The ioctl numbers can change from release to release, because + * the caller (libzfs) must be matched to the kernel. + * + * zfs_secpolicy_func_t *secpolicy + * This function will be called before the zfs_ioc_func_t, to + * determine if this operation is permitted. It should return EPERM + * on failure, and 0 on success. Checks include determining if the + * dataset is visible in this zone, and if the user has either all + * zfs privileges in the zone (SYS_MOUNT), or has been granted permission + * to do this operation on this dataset with "zfs allow". + * + * zfs_ioc_namecheck_t namecheck + * This specifies what to expect in the zfs_cmd_t:zc_name -- a pool + * name, a dataset name, or nothing. If the name is not well-formed, + * the ioctl will fail and the callback will not be called. + * Therefore, the callback can assume that the name is well-formed + * (e.g. is null-terminated, doesn't have more than one '@' character, + * doesn't have invalid characters). + * + * zfs_ioc_poolcheck_t pool_check + * This specifies requirements on the pool state. If the pool does + * not meet them (is suspended or is readonly), the ioctl will fail + * and the callback will not be called. If any checks are specified + * (i.e. it is not POOL_CHECK_NONE), namecheck must not be NO_NAME. + * Multiple checks can be or-ed together (e.g. POOL_CHECK_SUSPENDED | + * POOL_CHECK_READONLY). + * + * boolean_t smush_outnvlist + * If smush_outnvlist is true, then the output is presumed to be a + * list of errors, and it will be "smushed" down to fit into the + * caller's buffer, by removing some entries and replacing them with a + * single "N_MORE_ERRORS" entry indicating how many were removed. See + * nvlist_smush() for details. If smush_outnvlist is false, and the + * outnvlist does not fit into the userland-provided buffer, then the + * ioctl will fail with ENOMEM. + * + * zfs_ioc_func_t *func + * The callback function that will perform the operation. + * + * The callback should return 0 on success, or an error number on + * failure. If the function fails, the userland ioctl will return -1, + * and errno will be set to the callback's return value. The callback + * will be called with the following arguments: + * + * const char *name + * The name of the pool or dataset to operate on, from + * zfs_cmd_t:zc_name. The 'namecheck' argument specifies the + * expected type (pool, dataset, or none). + * + * nvlist_t *innvl + * The input nvlist, deserialized from zfs_cmd_t:zc_nvlist_src. Or + * NULL if no input nvlist was provided. Changes to this nvlist are + * ignored. If the input nvlist could not be deserialized, the + * ioctl will fail and the callback will not be called. + * + * nvlist_t *outnvl + * The output nvlist, initially empty. The callback can fill it in, + * and it will be returned to userland by serializing it into + * zfs_cmd_t:zc_nvlist_dst. If it is non-empty, and serialization + * fails (e.g. because the caller didn't supply a large enough + * buffer), then the overall ioctl will fail. See the + * 'smush_nvlist' argument above for additional behaviors. + * + * There are two typical uses of the output nvlist: + * - To return state, e.g. property values. In this case, + * smush_outnvlist should be false. If the buffer was not large + * enough, the caller will reallocate a larger buffer and try + * the ioctl again. + * + * - To return multiple errors from an ioctl which makes on-disk + * changes. In this case, smush_outnvlist should be true. + * Ioctls which make on-disk modifications should generally not + * use the outnvl if they succeed, because the caller can not + * distinguish between the operation failing, and + * deserialization failing. */ #include @@ -85,8 +186,13 @@ ldi_ident_t zfs_li = NULL; dev_info_t *zfs_dip; -typedef int zfs_ioc_func_t(zfs_cmd_t *); -typedef int zfs_secpolicy_func_t(zfs_cmd_t *, cred_t *); +uint_t zfs_fsyncer_key; +extern uint_t rrw_tsd_key; +static uint_t zfs_allow_log_key; + +typedef int zfs_ioc_legacy_func_t(zfs_cmd_t *); +typedef int zfs_ioc_func_t(const char *, nvlist_t *, nvlist_t *); +typedef int zfs_secpolicy_func_t(zfs_cmd_t *, nvlist_t *, cred_t *); typedef enum { NO_NAME, @@ -97,15 +203,18 @@ typedef enum { POOL_CHECK_NONE = 1 << 0, POOL_CHECK_SUSPENDED = 1 << 1, - POOL_CHECK_READONLY = 1 << 2 + POOL_CHECK_READONLY = 1 << 2, } zfs_ioc_poolcheck_t; typedef struct zfs_ioc_vec { + zfs_ioc_legacy_func_t *zvec_legacy_func; zfs_ioc_func_t *zvec_func; zfs_secpolicy_func_t *zvec_secpolicy; zfs_ioc_namecheck_t zvec_namecheck; - boolean_t zvec_his_log; + boolean_t zvec_allow_log; zfs_ioc_poolcheck_t zvec_pool_check; + boolean_t zvec_smush_outnvlist; + const char *zvec_name; } zfs_ioc_vec_t; /* This array is indexed by zfs_userquota_prop_t */ @@ -123,7 +232,8 @@ nvlist_t **errors); static int zfs_fill_zplprops_root(uint64_t, nvlist_t *, nvlist_t *, boolean_t *); -int zfs_set_prop_nvlist(const char *, zprop_source_t, nvlist_t *, nvlist_t **); +int zfs_set_prop_nvlist(const char *, zprop_source_t, nvlist_t *, nvlist_t *); +static int get_nvlist(uint64_t nvl, uint64_t size, int iflag, nvlist_t **nvp); /* _NOTE(PRINTFLIKE(4)) - this is printf-like, but lint is too whiney */ void @@ -261,7 +371,7 @@ if (spa_open(zc->zc_name, &spa, FTAG) == 0) { if (spa_version(spa) >= SPA_VERSION_ZPOOL_HISTORY) - (void) spa_history_log(spa, buf, LOG_CMD_NORMAL); + (void) spa_history_log(spa, buf); spa_close(spa, FTAG); } history_str_free(buf); @@ -273,7 +383,7 @@ */ /* ARGSUSED */ static int -zfs_secpolicy_none(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_none(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (0); } @@ -284,7 +394,7 @@ */ /* ARGSUSED */ static int -zfs_secpolicy_read(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_read(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { if (INGLOBALZONE(curproc) || zone_dataset_visible(zc->zc_name, NULL)) @@ -353,26 +463,13 @@ return (zfs_dozonecheck_impl(dataset, zoned, cr)); } -/* - * If name ends in a '@', then require recursive permissions. - */ -int +static int zfs_secpolicy_write_perms(const char *name, const char *perm, cred_t *cr) { int error; - boolean_t descendent = B_FALSE; dsl_dataset_t *ds; - char *at; - - at = strchr(name, '@'); - if (at != NULL && at[1] == '\0') { - *at = '\0'; - descendent = B_TRUE; - } error = dsl_dataset_hold(name, FTAG, &ds); - if (at != NULL) - *at = '@'; if (error != 0) return (error); @@ -380,14 +477,14 @@ if (error == 0) { error = secpolicy_zfs(cr); if (error) - error = dsl_deleg_access_impl(ds, descendent, perm, cr); + error = dsl_deleg_access_impl(ds, perm, cr); } dsl_dataset_rele(ds, FTAG); return (error); } -int +static int zfs_secpolicy_write_perms_ds(const char *name, dsl_dataset_t *ds, const char *perm, cred_t *cr) { @@ -397,7 +494,7 @@ if (error == 0) { error = secpolicy_zfs(cr); if (error) - error = dsl_deleg_access_impl(ds, B_FALSE, perm, cr); + error = dsl_deleg_access_impl(ds, perm, cr); } return (error); } @@ -551,8 +648,9 @@ return (zfs_secpolicy_write_perms(dsname, zfs_prop_to_name(prop), cr)); } -int -zfs_secpolicy_fsacl(zfs_cmd_t *zc, cred_t *cr) +/* ARGSUSED */ +static int +zfs_secpolicy_set_fsacl(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int error; @@ -567,15 +665,17 @@ return (0); } -int -zfs_secpolicy_rollback(zfs_cmd_t *zc, cred_t *cr) +/* ARGSUSED */ +static int +zfs_secpolicy_rollback(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_ROLLBACK, cr)); } -int -zfs_secpolicy_send(zfs_cmd_t *zc, cred_t *cr) +/* ARGSUSED */ +static int +zfs_secpolicy_send(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { spa_t *spa; dsl_pool_t *dp; @@ -611,8 +711,17 @@ return (error); } +/* ARGSUSED */ static int -zfs_secpolicy_deleg_share(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_send_new(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) +{ + return (zfs_secpolicy_write_perms(zc->zc_name, + ZFS_DELEG_PERM_SEND, cr)); +} + +/* ARGSUSED */ +static int +zfs_secpolicy_deleg_share(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { vnode_t *vp; int error; @@ -636,7 +745,7 @@ } int -zfs_secpolicy_share(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_share(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { if (!INGLOBALZONE(curproc)) return (EPERM); @@ -644,12 +753,12 @@ if (secpolicy_nfs(cr) == 0) { return (0); } else { - return (zfs_secpolicy_deleg_share(zc, cr)); + return (zfs_secpolicy_deleg_share(zc, innvl, cr)); } } int -zfs_secpolicy_smb_acl(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_smb_acl(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { if (!INGLOBALZONE(curproc)) return (EPERM); @@ -657,7 +766,7 @@ if (secpolicy_smb(cr) == 0) { return (0); } else { - return (zfs_secpolicy_deleg_share(zc, cr)); + return (zfs_secpolicy_deleg_share(zc, innvl, cr)); } } @@ -695,27 +804,55 @@ return (zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_DESTROY, cr)); } +/* ARGSUSED */ static int -zfs_secpolicy_destroy(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_destroy(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_destroy_perms(zc->zc_name, cr)); } /* * Destroying snapshots with delegated permissions requires - * descendent mount and destroy permissions. + * descendant mount and destroy permissions. */ +/* ARGSUSED */ static int -zfs_secpolicy_destroy_recursive(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_destroy_snaps(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { - int error; - char *dsname; - - dsname = kmem_asprintf("%s@", zc->zc_name); - - error = zfs_secpolicy_destroy_perms(dsname, cr); - - strfree(dsname); + nvlist_t *snaps; + nvpair_t *pair, *nextpair; + int error = 0; + + if (nvlist_lookup_nvlist(innvl, "snaps", &snaps) != 0) + return (EINVAL); + for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; + pair = nextpair) { + dsl_dataset_t *ds; + + nextpair = nvlist_next_nvpair(snaps, pair); + error = dsl_dataset_hold(nvpair_name(pair), FTAG, &ds); + if (error == 0) { + dsl_dataset_rele(ds, FTAG); + } else if (error == ENOENT) { + /* + * Ignore any snapshots that don't exist (we consider + * them "already destroyed"). Remove the name from the + * nvl here in case the snapshot is created between + * now and when we try to destroy it (in which case + * we don't want to destroy it since we haven't + * checked for permission). + */ + fnvlist_remove_nvpair(snaps, pair); + error = 0; + continue; + } else { + break; + } + error = zfs_secpolicy_destroy_perms(nvpair_name(pair), cr); + if (error != 0) + break; + } + return (error); } @@ -748,14 +885,16 @@ return (error); } +/* ARGSUSED */ static int -zfs_secpolicy_rename(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_rename(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_rename_perms(zc->zc_name, zc->zc_value, cr)); } +/* ARGSUSED */ static int -zfs_secpolicy_promote(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_promote(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { char parentname[MAXNAMELEN]; objset_t *clone; @@ -795,8 +934,9 @@ return (error); } +/* ARGSUSED */ static int -zfs_secpolicy_receive(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_recv(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int error; @@ -819,49 +959,72 @@ ZFS_DELEG_PERM_SNAPSHOT, cr)); } +/* + * Check for permission to create each snapshot in the nvlist. + */ +/* ARGSUSED */ static int -zfs_secpolicy_snapshot(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_snapshot(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { - - return (zfs_secpolicy_snapshot_perms(zc->zc_name, cr)); + nvlist_t *snaps; + int error; + nvpair_t *pair; + + if (nvlist_lookup_nvlist(innvl, "snaps", &snaps) != 0) + return (EINVAL); + for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; + pair = nvlist_next_nvpair(snaps, pair)) { + char *name = nvpair_name(pair); + char *atp = strchr(name, '@'); + + if (atp == NULL) { + error = EINVAL; + break; + } + *atp = '\0'; + error = zfs_secpolicy_snapshot_perms(name, cr); + *atp = '@'; + if (error != 0) + break; + } + return (error); +} + +/* ARGSUSED */ +static int +zfs_secpolicy_log_history(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) +{ + /* + * Even root must have a proper TSD so that we know what pool + * to log to. + */ + if (tsd_get(zfs_allow_log_key) == NULL) + return (EPERM); + return (0); } static int -zfs_secpolicy_create(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_create_clone(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { char parentname[MAXNAMELEN]; int error; + char *origin; if ((error = zfs_get_parent(zc->zc_name, parentname, sizeof (parentname))) != 0) return (error); - if (zc->zc_value[0] != '\0') { - if ((error = zfs_secpolicy_write_perms(zc->zc_value, - ZFS_DELEG_PERM_CLONE, cr)) != 0) - return (error); - } + if (nvlist_lookup_string(innvl, "origin", &origin) == 0 && + (error = zfs_secpolicy_write_perms(origin, + ZFS_DELEG_PERM_CLONE, cr)) != 0) + return (error); if ((error = zfs_secpolicy_write_perms(parentname, ZFS_DELEG_PERM_CREATE, cr)) != 0) return (error); - error = zfs_secpolicy_write_perms(parentname, - ZFS_DELEG_PERM_MOUNT, cr); - - return (error); -} - -static int -zfs_secpolicy_umount(zfs_cmd_t *zc, cred_t *cr) -{ - int error; - - error = secpolicy_fs_unmount(cr, NULL); - if (error) { - error = dsl_deleg_access(zc->zc_name, ZFS_DELEG_PERM_MOUNT, cr); - } - return (error); + return (zfs_secpolicy_write_perms(parentname, + ZFS_DELEG_PERM_MOUNT, cr)); } /* @@ -870,7 +1033,7 @@ */ /* ARGSUSED */ static int -zfs_secpolicy_config(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_config(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { if (secpolicy_sys_config(cr, B_FALSE) != 0) return (EPERM); @@ -883,7 +1046,7 @@ */ /* ARGSUSED */ static int -zfs_secpolicy_diff(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_diff(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int error; @@ -899,13 +1062,14 @@ */ /* ARGSUSED */ static int -zfs_secpolicy_inject(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_inject(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (secpolicy_zinject(cr)); } +/* ARGSUSED */ static int -zfs_secpolicy_inherit(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_inherit_prop(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { zfs_prop_t prop = zfs_name_to_prop(zc->zc_value); @@ -921,9 +1085,9 @@ } static int -zfs_secpolicy_userspace_one(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_userspace_one(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { - int err = zfs_secpolicy_read(zc, cr); + int err = zfs_secpolicy_read(zc, innvl, cr); if (err) return (err); @@ -950,9 +1114,9 @@ } static int -zfs_secpolicy_userspace_many(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_userspace_many(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { - int err = zfs_secpolicy_read(zc, cr); + int err = zfs_secpolicy_read(zc, innvl, cr); if (err) return (err); @@ -963,22 +1127,25 @@ userquota_perms[zc->zc_objset_type], cr)); } +/* ARGSUSED */ static int -zfs_secpolicy_userspace_upgrade(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_userspace_upgrade(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_setprop(zc->zc_name, ZFS_PROP_VERSION, NULL, cr)); } +/* ARGSUSED */ static int -zfs_secpolicy_hold(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_hold(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_HOLD, cr)); } +/* ARGSUSED */ static int -zfs_secpolicy_release(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_release(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_RELEASE, cr)); @@ -988,7 +1155,7 @@ * Policy for allowing temporary snapshots to be taken or released */ static int -zfs_secpolicy_tmp_snapshot(zfs_cmd_t *zc, cred_t *cr) +zfs_secpolicy_tmp_snapshot(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { /* * A temporary snapshot is the same as a snapshot, @@ -1001,13 +1168,13 @@ ZFS_DELEG_PERM_DIFF, cr)) == 0) return (0); - error = zfs_secpolicy_snapshot(zc, cr); + error = zfs_secpolicy_snapshot_perms(zc->zc_name, cr); if (!error) - error = zfs_secpolicy_hold(zc, cr); + error = zfs_secpolicy_hold(zc, innvl, cr); if (!error) - error = zfs_secpolicy_release(zc, cr); + error = zfs_secpolicy_release(zc, innvl, cr); if (!error) - error = zfs_secpolicy_destroy(zc, cr); + error = zfs_secpolicy_destroy(zc, innvl, cr); return (error); } @@ -1046,36 +1213,40 @@ return (0); } +/* + * Reduce the size of this nvlist until it can be serialized in 'max' bytes. + * Entries will be removed from the end of the nvlist, and one int32 entry + * named "N_MORE_ERRORS" will be added indicating how many entries were + * removed. + */ static int -fit_error_list(zfs_cmd_t *zc, nvlist_t **errors) +nvlist_smush(nvlist_t *errors, size_t max) { size_t size; - VERIFY(nvlist_size(*errors, &size, NV_ENCODE_NATIVE) == 0); - - if (size > zc->zc_nvlist_dst_size) { + size = fnvlist_size(errors); + + if (size > max) { nvpair_t *more_errors; int n = 0; - if (zc->zc_nvlist_dst_size < 1024) + if (max < 1024) return (ENOMEM); - VERIFY(nvlist_add_int32(*errors, ZPROP_N_MORE_ERRORS, 0) == 0); - more_errors = nvlist_prev_nvpair(*errors, NULL); + fnvlist_add_int32(errors, ZPROP_N_MORE_ERRORS, 0); + more_errors = nvlist_prev_nvpair(errors, NULL); do { - nvpair_t *pair = nvlist_prev_nvpair(*errors, + nvpair_t *pair = nvlist_prev_nvpair(errors, more_errors); - VERIFY(nvlist_remove_nvpair(*errors, pair) == 0); + fnvlist_remove_nvpair(errors, pair); n++; - VERIFY(nvlist_size(*errors, &size, - NV_ENCODE_NATIVE) == 0); - } while (size > zc->zc_nvlist_dst_size); - - VERIFY(nvlist_remove_nvpair(*errors, more_errors) == 0); - VERIFY(nvlist_add_int32(*errors, ZPROP_N_MORE_ERRORS, n) == 0); - ASSERT(nvlist_size(*errors, &size, NV_ENCODE_NATIVE) == 0); - ASSERT(size <= zc->zc_nvlist_dst_size); + size = fnvlist_size(errors); + } while (size > max); + + fnvlist_remove_nvpair(errors, more_errors); + fnvlist_add_int32(errors, ZPROP_N_MORE_ERRORS, n); + ASSERT3U(fnvlist_size(errors), <=, max); } return (0); @@ -1088,21 +1259,20 @@ int error = 0; size_t size; - VERIFY(nvlist_size(nvl, &size, NV_ENCODE_NATIVE) == 0); + size = fnvlist_size(nvl); if (size > zc->zc_nvlist_dst_size) { error = ENOMEM; } else { - packed = kmem_alloc(size, KM_SLEEP); - VERIFY(nvlist_pack(nvl, &packed, &size, NV_ENCODE_NATIVE, - KM_SLEEP) == 0); + packed = fnvlist_pack(nvl, &size); if (ddi_copyout(packed, (void *)(uintptr_t)zc->zc_nvlist_dst, size, zc->zc_iflags) != 0) error = EFAULT; - kmem_free(packed, size); + fnvlist_pack_free(packed, size); } zc->zc_nvlist_dst_size = size; + zc->zc_nvlist_dst_filled = B_TRUE; return (error); } @@ -1181,7 +1351,6 @@ nvlist_t *config, *props = NULL; nvlist_t *rootprops = NULL; nvlist_t *zplprops = NULL; - char *buf; if (error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config)) @@ -1221,9 +1390,7 @@ goto pool_props_bad; } - buf = history_str_get(zc); - - error = spa_create(zc->zc_name, config, props, buf, zplprops); + error = spa_create(zc->zc_name, config, props, zplprops); /* * Set the remaining root properties @@ -1232,9 +1399,6 @@ ZPROP_SRC_LOCAL, rootprops, NULL)) != 0) (void) spa_destroy(zc->zc_name); - if (buf != NULL) - history_str_free(buf); - pool_props_bad: nvlist_free(rootprops); nvlist_free(zplprops); @@ -2229,31 +2393,25 @@ /* * This function is best effort. If it fails to set any of the given properties, - * it continues to set as many as it can and returns the first error - * encountered. If the caller provides a non-NULL errlist, it also gives the - * complete list of names of all the properties it failed to set along with the - * corresponding error numbers. The caller is responsible for freeing the - * returned errlist. + * it continues to set as many as it can and returns the last error + * encountered. If the caller provides a non-NULL errlist, it will be filled in + * with the list of names of all the properties that failed along with the + * corresponding error numbers. * - * If every property is set successfully, zero is returned and the list pointed - * at by errlist is NULL. + * If every property is set successfully, zero is returned and errlist is not + * modified. */ int zfs_set_prop_nvlist(const char *dsname, zprop_source_t source, nvlist_t *nvl, - nvlist_t **errlist) + nvlist_t *errlist) { nvpair_t *pair; nvpair_t *propval; int rv = 0; uint64_t intval; char *strval; - nvlist_t *genericnvl; - nvlist_t *errors; - nvlist_t *retrynvl; - - VERIFY(nvlist_alloc(&genericnvl, NV_UNIQUE_NAME, KM_SLEEP) == 0); - VERIFY(nvlist_alloc(&errors, NV_UNIQUE_NAME, KM_SLEEP) == 0); - VERIFY(nvlist_alloc(&retrynvl, NV_UNIQUE_NAME, KM_SLEEP) == 0); + nvlist_t *genericnvl = fnvlist_alloc(); + nvlist_t *retrynvl = fnvlist_alloc(); retry: pair = NULL; @@ -2266,7 +2424,7 @@ propval = pair; if (nvpair_type(pair) == DATA_TYPE_NVLIST) { nvlist_t *attrs; - VERIFY(nvpair_value_nvlist(pair, &attrs) == 0); + attrs = fnvpair_value_nvlist(pair); if (nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &propval) != 0) err = EINVAL; @@ -2291,8 +2449,7 @@ } else if (nvpair_type(propval) == DATA_TYPE_UINT64) { const char *unused; - VERIFY(nvpair_value_uint64(propval, - &intval) == 0); + intval = fnvpair_value_uint64(propval); switch (zfs_prop_get_type(prop)) { case PROP_TYPE_NUMBER: @@ -2336,8 +2493,11 @@ } } - if (err != 0) - VERIFY(nvlist_add_int32(errors, propname, err) == 0); + if (err != 0) { + if (errlist != NULL) + fnvlist_add_int32(errlist, propname, err); + rv = err; + } } if (nvl != retrynvl && !nvlist_empty(retrynvl)) { @@ -2359,44 +2519,33 @@ propval = pair; if (nvpair_type(pair) == DATA_TYPE_NVLIST) { nvlist_t *attrs; - VERIFY(nvpair_value_nvlist(pair, &attrs) == 0); - VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE, - &propval) == 0); + attrs = fnvpair_value_nvlist(pair); + propval = fnvlist_lookup_nvpair(attrs, + ZPROP_VALUE); } if (nvpair_type(propval) == DATA_TYPE_STRING) { - VERIFY(nvpair_value_string(propval, - &strval) == 0); + strval = fnvpair_value_string(propval); err = dsl_prop_set(dsname, propname, source, 1, strlen(strval) + 1, strval); } else { - VERIFY(nvpair_value_uint64(propval, - &intval) == 0); + intval = fnvpair_value_uint64(propval); err = dsl_prop_set(dsname, propname, source, 8, 1, &intval); } if (err != 0) { - VERIFY(nvlist_add_int32(errors, propname, - err) == 0); + if (errlist != NULL) { + fnvlist_add_int32(errlist, propname, + err); + } + rv = err; } } } nvlist_free(genericnvl); nvlist_free(retrynvl); - if ((pair = nvlist_next_nvpair(errors, NULL)) == NULL) { - nvlist_free(errors); - errors = NULL; - } else { - VERIFY(nvpair_value_int32(pair, &rv) == 0); - } - - if (errlist == NULL) - nvlist_free(errors); - else - *errlist = errors; - return (rv); } @@ -2404,7 +2553,7 @@ * Check that all the properties are valid user properties. */ static int -zfs_check_userprops(char *fsname, nvlist_t *nvl) +zfs_check_userprops(const char *fsname, nvlist_t *nvl) { nvpair_t *pair = NULL; int error = 0; @@ -2484,7 +2633,7 @@ boolean_t received = zc->zc_cookie; zprop_source_t source = (received ? ZPROP_SRC_RECEIVED : ZPROP_SRC_LOCAL); - nvlist_t *errors = NULL; + nvlist_t *errors; int error; if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, @@ -2507,7 +2656,8 @@ } } - error = zfs_set_prop_nvlist(zc->zc_name, source, nvl, &errors); + errors = fnvlist_alloc(); + error = zfs_set_prop_nvlist(zc->zc_name, source, nvl, errors); if (zc->zc_nvlist_dst != NULL && errors != NULL) { (void) put_nvlist(zc, errors); @@ -2589,7 +2739,7 @@ return (EINVAL); } - /* the property name has been validated by zfs_secpolicy_inherit() */ + /* property name has been validated by zfs_secpolicy_inherit_prop() */ return (dsl_prop_set(zc->zc_name, zc->zc_value, source, 0, 0, NULL)); } @@ -2932,26 +3082,30 @@ } /* - * inputs: - * zc_objset_type type of objset to create (fs vs zvol) - * zc_name name of new objset - * zc_value name of snapshot to clone from (may be empty) - * zc_nvlist_src{_size} nvlist of properties to apply + * innvl: { + * "type" -> dmu_objset_type_t (int32) + * (optional) "props" -> { prop -> value } + * } * - * outputs: none + * outnvl: propname -> error code (int32) */ static int -zfs_ioc_create(zfs_cmd_t *zc) +zfs_ioc_create(const char *fsname, nvlist_t *innvl, nvlist_t *outnvl) { - objset_t *clone; int error = 0; - zfs_creat_t zct; + zfs_creat_t zct = { 0 }; nvlist_t *nvprops = NULL; void (*cbfunc)(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx); - dmu_objset_type_t type = zc->zc_objset_type; + int32_t type32; + dmu_objset_type_t type; + boolean_t is_insensitive = B_FALSE; + + if (nvlist_lookup_int32(innvl, "type", &type32) != 0) + return (EINVAL); + type = type32; + (void) nvlist_lookup_nvlist(innvl, "props", &nvprops); switch (type) { - case DMU_OST_ZFS: cbfunc = zfs_create_cb; break; @@ -2964,230 +3118,290 @@ cbfunc = NULL; break; } - if (strchr(zc->zc_name, '@') || - strchr(zc->zc_name, '%')) + if (strchr(fsname, '@') || + strchr(fsname, '%')) + return (EINVAL); + + zct.zct_props = nvprops; + + if (cbfunc == NULL) return (EINVAL); - if (zc->zc_nvlist_src != NULL && - (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, - zc->zc_iflags, &nvprops)) != 0) - return (error); - - zct.zct_zplprops = NULL; - zct.zct_props = nvprops; - - if (zc->zc_value[0] != '\0') { + if (type == DMU_OST_ZVOL) { + uint64_t volsize, volblocksize; + + if (nvprops == NULL) + return (EINVAL); + if (nvlist_lookup_uint64(nvprops, + zfs_prop_to_name(ZFS_PROP_VOLSIZE), &volsize) != 0) + return (EINVAL); + + if ((error = nvlist_lookup_uint64(nvprops, + zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE), + &volblocksize)) != 0 && error != ENOENT) + return (EINVAL); + + if (error != 0) + volblocksize = zfs_prop_default_numeric( + ZFS_PROP_VOLBLOCKSIZE); + + if ((error = zvol_check_volblocksize( + volblocksize)) != 0 || + (error = zvol_check_volsize(volsize, + volblocksize)) != 0) + return (error); + } else if (type == DMU_OST_ZFS) { + int error; + /* - * We're creating a clone of an existing snapshot. + * We have to have normalization and + * case-folding flags correct when we do the + * file system creation, so go figure them out + * now. */ - zc->zc_value[sizeof (zc->zc_value) - 1] = '\0'; - if (dataset_namecheck(zc->zc_value, NULL, NULL) != 0) { - nvlist_free(nvprops); - return (EINVAL); - } - - error = dmu_objset_hold(zc->zc_value, FTAG, &clone); - if (error) { - nvlist_free(nvprops); - return (error); - } - - error = dmu_objset_clone(zc->zc_name, dmu_objset_ds(clone), 0); - dmu_objset_rele(clone, FTAG); - if (error) { - nvlist_free(nvprops); + VERIFY(nvlist_alloc(&zct.zct_zplprops, + NV_UNIQUE_NAME, KM_SLEEP) == 0); + error = zfs_fill_zplprops(fsname, nvprops, + zct.zct_zplprops, &is_insensitive); + if (error != 0) { + nvlist_free(zct.zct_zplprops); return (error); } - } else { - boolean_t is_insensitive = B_FALSE; - - if (cbfunc == NULL) { - nvlist_free(nvprops); - return (EINVAL); - } - - if (type == DMU_OST_ZVOL) { - uint64_t volsize, volblocksize; - - if (nvprops == NULL || - nvlist_lookup_uint64(nvprops, - zfs_prop_to_name(ZFS_PROP_VOLSIZE), - &volsize) != 0) { - nvlist_free(nvprops); - return (EINVAL); - } - - if ((error = nvlist_lookup_uint64(nvprops, - zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE), - &volblocksize)) != 0 && error != ENOENT) { - nvlist_free(nvprops); - return (EINVAL); - } - - if (error != 0) - volblocksize = zfs_prop_default_numeric( - ZFS_PROP_VOLBLOCKSIZE); - - if ((error = zvol_check_volblocksize( - volblocksize)) != 0 || - (error = zvol_check_volsize(volsize, - volblocksize)) != 0) { - nvlist_free(nvprops); - return (error); - } - } else if (type == DMU_OST_ZFS) { - int error; - - /* - * We have to have normalization and - * case-folding flags correct when we do the - * file system creation, so go figure them out - * now. - */ - VERIFY(nvlist_alloc(&zct.zct_zplprops, - NV_UNIQUE_NAME, KM_SLEEP) == 0); - error = zfs_fill_zplprops(zc->zc_name, nvprops, - zct.zct_zplprops, &is_insensitive); - if (error != 0) { - nvlist_free(nvprops); - nvlist_free(zct.zct_zplprops); - return (error); - } - } - error = dmu_objset_create(zc->zc_name, type, - is_insensitive ? DS_FLAG_CI_DATASET : 0, cbfunc, &zct); - nvlist_free(zct.zct_zplprops); + } + + error = dmu_objset_create(fsname, type, + is_insensitive ? DS_FLAG_CI_DATASET : 0, cbfunc, &zct); + nvlist_free(zct.zct_zplprops); + + /* + * It would be nice to do this atomically. + */ + if (error == 0) { + error = zfs_set_prop_nvlist(fsname, ZPROP_SRC_LOCAL, + nvprops, outnvl); + if (error != 0) + (void) dmu_objset_destroy(fsname, B_FALSE); } + return (error); +} + +/* + * innvl: { + * "origin" -> name of origin snapshot + * (optional) "props" -> { prop -> value } + * } + * + * outnvl: propname -> error code (int32) + */ +static int +zfs_ioc_clone(const char *fsname, nvlist_t *innvl, nvlist_t *outnvl) +{ + int error = 0; + nvlist_t *nvprops = NULL; + char *origin_name; + dsl_dataset_t *origin; + + if (nvlist_lookup_string(innvl, "origin", &origin_name) != 0) + return (EINVAL); + (void) nvlist_lookup_nvlist(innvl, "props", &nvprops); + + if (strchr(fsname, '@') || + strchr(fsname, '%')) + return (EINVAL); + + if (dataset_namecheck(origin_name, NULL, NULL) != 0) + return (EINVAL); + + error = dsl_dataset_hold(origin_name, FTAG, &origin); + if (error) + return (error); + + error = dmu_objset_clone(fsname, origin, 0); + dsl_dataset_rele(origin, FTAG); + if (error) + return (error); /* * It would be nice to do this atomically. */ if (error == 0) { - error = zfs_set_prop_nvlist(zc->zc_name, ZPROP_SRC_LOCAL, - nvprops, NULL); + error = zfs_set_prop_nvlist(fsname, ZPROP_SRC_LOCAL, + nvprops, outnvl); if (error != 0) - (void) dmu_objset_destroy(zc->zc_name, B_FALSE); + (void) dmu_objset_destroy(fsname, B_FALSE); } - nvlist_free(nvprops); + return (error); +} + +/* + * innvl: { + * "snaps" -> { snapshot1, snapshot2 } + * (optional) "props" -> { prop -> value (string) } + * } + * + * outnvl: snapshot -> error code (int32) + * + */ +static int +zfs_ioc_snapshot(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) +{ + nvlist_t *snaps; + nvlist_t *props = NULL; + int error, poollen; + nvpair_t *pair; + + (void) nvlist_lookup_nvlist(innvl, "props", &props); + if ((error = zfs_check_userprops(poolname, props)) != 0) + return (error); + + if (!nvlist_empty(props) && + zfs_earlier_version(poolname, SPA_VERSION_SNAP_PROPS)) + return (ENOTSUP); + + if (nvlist_lookup_nvlist(innvl, "snaps", &snaps) != 0) + return (EINVAL); + poollen = strlen(poolname); + for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; + pair = nvlist_next_nvpair(snaps, pair)) { + const char *name = nvpair_name(pair); + const char *cp = strchr(name, '@'); + + /* + * The snap name must contain an @, and the part after it must + * contain only valid characters. + */ + if (cp == NULL || snapshot_namecheck(cp + 1, NULL, NULL) != 0) + return (EINVAL); + + /* + * The snap must be in the specified pool. + */ + if (strncmp(name, poolname, poollen) != 0 || + (name[poollen] != '/' && name[poollen] != '@')) + return (EXDEV); + + /* This must be the only snap of this fs. */ + for (nvpair_t *pair2 = nvlist_next_nvpair(snaps, pair); + pair2 != NULL; pair2 = nvlist_next_nvpair(snaps, pair2)) { + if (strncmp(name, nvpair_name(pair2), cp - name + 1) + == 0) { + return (EXDEV); + } + } + } + + error = dmu_objset_snapshot(snaps, props, outnvl); return (error); } /* - * inputs: - * zc_name name of filesystem - * zc_value short name of snapshot - * zc_cookie recursive flag - * zc_nvlist_src[_size] property list - * - * outputs: - * zc_value short snapname (i.e. part after the '@') + * innvl: "message" -> string */ +/* ARGSUSED */ static int -zfs_ioc_snapshot(zfs_cmd_t *zc) +zfs_ioc_log_history(const char *unused, nvlist_t *innvl, nvlist_t *outnvl) { - nvlist_t *nvprops = NULL; + char *message; + spa_t *spa; int error; - boolean_t recursive = zc->zc_cookie; - - if (snapshot_namecheck(zc->zc_value, NULL, NULL) != 0) + char *poolname; + + /* + * The poolname in the ioctl is not set, we get it from the TSD, + * which was set at the end of the last successful ioctl that allows + * logging. The secpolicy func already checked that it is set. + * Only one log ioctl is allowed after each successful ioctl, so + * we clear the TSD here. + */ + poolname = tsd_get(zfs_allow_log_key); + (void) tsd_set(zfs_allow_log_key, NULL); + error = spa_open(poolname, &spa, FTAG); + strfree(poolname); + if (error != 0) + return (error); + + if (nvlist_lookup_string(innvl, "message", &message) != 0) { + spa_close(spa, FTAG); return (EINVAL); - - if (zc->zc_nvlist_src != NULL && - (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, - zc->zc_iflags, &nvprops)) != 0) - return (error); - - error = zfs_check_userprops(zc->zc_name, nvprops); - if (error) - goto out; - - if (!nvlist_empty(nvprops) && - zfs_earlier_version(zc->zc_name, SPA_VERSION_SNAP_PROPS)) { - error = ENOTSUP; - goto out; } - error = dmu_objset_snapshot(zc->zc_name, zc->zc_value, NULL, - nvprops, recursive, B_FALSE, -1); - -out: - nvlist_free(nvprops); + if (spa_version(spa) < SPA_VERSION_ZPOOL_HISTORY) { + spa_close(spa, FTAG); + return (ENOTSUP); + } + + error = spa_history_log(spa, message); + spa_close(spa, FTAG); return (error); } +/* ARGSUSED */ int zfs_unmount_snap(const char *name, void *arg) { - vfs_t *vfsp = NULL; - - if (arg) { - char *snapname = arg; - char *fullname = kmem_asprintf("%s@%s", name, snapname); - vfsp = zfs_get_vfs(fullname); - strfree(fullname); - } else if (strchr(name, '@')) { - vfsp = zfs_get_vfs(name); + vfs_t *vfsp; + int err; + + if (strchr(name, '@') == NULL) + return (0); + + vfsp = zfs_get_vfs(name); + if (vfsp == NULL) + return (0); + + if ((err = vn_vfswlock(vfsp->vfs_vnodecovered)) != 0) { + VFS_RELE(vfsp); + return (err); } - - if (vfsp) { - /* - * Always force the unmount for snapshots. - */ - int flag = MS_FORCE; - int err; - - if ((err = vn_vfswlock(vfsp->vfs_vnodecovered)) != 0) { - VFS_RELE(vfsp); - return (err); - } - VFS_RELE(vfsp); - if ((err = dounmount(vfsp, flag, kcred)) != 0) - return (err); - } - return (0); + VFS_RELE(vfsp); + + /* + * Always force the unmount for snapshots. + */ + return (dounmount(vfsp, MS_FORCE, kcred)); } /* - * inputs: - * zc_name name of filesystem, snaps must be under it - * zc_nvlist_src[_size] full names of snapshots to destroy - * zc_defer_destroy mark for deferred destroy + * innvl: { + * "snaps" -> { snapshot1, snapshot2 } + * (optional boolean) "defer" + * } * - * outputs: - * zc_name on failure, name of failed snapshot + * outnvl: snapshot -> error code (int32) + * */ static int -zfs_ioc_destroy_snaps_nvl(zfs_cmd_t *zc) +zfs_ioc_destroy_snaps(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { - int err, len; - nvlist_t *nvl; + int poollen; + nvlist_t *snaps; nvpair_t *pair; - - if ((err = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, - zc->zc_iflags, &nvl)) != 0) - return (err); - - len = strlen(zc->zc_name); - for (pair = nvlist_next_nvpair(nvl, NULL); pair != NULL; - pair = nvlist_next_nvpair(nvl, pair)) { + boolean_t defer; + + if (nvlist_lookup_nvlist(innvl, "snaps", &snaps) != 0) + return (EINVAL); + defer = nvlist_exists(innvl, "defer"); + + poollen = strlen(poolname); + for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; + pair = nvlist_next_nvpair(snaps, pair)) { const char *name = nvpair_name(pair); + /* - * The snap name must be underneath the zc_name. This ensures - * that our permission checks were legitimate. + * The snap must be in the specified pool. */ - if (strncmp(zc->zc_name, name, len) != 0 || - (name[len] != '@' && name[len] != '/')) { - nvlist_free(nvl); - return (EINVAL); - } - + if (strncmp(name, poolname, poollen) != 0 || + (name[poollen] != '/' && name[poollen] != '@')) + return (EXDEV); + + /* + * Ignore failures to unmount; dmu_snapshots_destroy_nvl() + * will deal with this gracefully (by filling in outnvl). + */ (void) zfs_unmount_snap(name, NULL); } - err = dmu_snapshots_destroy_nvl(nvl, zc->zc_defer_destroy, - zc->zc_name); - nvlist_free(nvl); - return (err); + return (dmu_snapshots_destroy_nvl(snaps, defer, outnvl)); } /* @@ -3491,7 +3705,7 @@ (void) strcpy(zc->zc_value, nvpair_name(pair)); if ((err = zfs_check_settable(dataset, pair, CRED())) != 0 || - (err = zfs_secpolicy_inherit(zc, CRED())) != 0) { + (err = zfs_secpolicy_inherit_prop(zc, NULL, CRED())) != 0) { VERIFY(nvlist_remove_nvpair(props, pair) == 0); VERIFY(nvlist_add_int32(errors, zc->zc_value, err) == 0); @@ -3699,8 +3913,6 @@ * dmu_recv_begin() succeeds. */ if (props) { - nvlist_t *errlist; - if (dmu_objset_from_ds(drc.drc_logical_ds, &os) == 0) { if (drc.drc_newfs) { if (spa_version(os->os_spa) >= @@ -3719,12 +3931,12 @@ } (void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_RECEIVED, - props, &errlist); - (void) nvlist_merge(errors, errlist, 0); - nvlist_free(errlist); + props, errors); } - if (fit_error_list(zc, &errors) != 0 || put_nvlist(zc, errors) != 0) { + if (zc->zc_nvlist_dst_size != 0 && + (nvlist_smush(errors, zc->zc_nvlist_dst_size) != 0 || + put_nvlist(zc, errors) != 0)) { /* * Caller made zc->zc_nvlist_dst less than the minimum expected * size or supplied an invalid address. @@ -3856,15 +4068,13 @@ rw_enter(&dp->dp_config_rwlock, RW_READER); error = dsl_dataset_hold_obj(dp, zc->zc_sendobj, FTAG, &ds); rw_exit(&dp->dp_config_rwlock); - if (error) { - spa_close(spa, FTAG); + spa_close(spa, FTAG); + if (error) return (error); - } error = dmu_objset_from_ds(ds, &tosnap); if (error) { dsl_dataset_rele(ds, FTAG); - spa_close(spa, FTAG); return (error); } @@ -3872,7 +4082,6 @@ rw_enter(&dp->dp_config_rwlock, RW_READER); error = dsl_dataset_hold_obj(dp, zc->zc_fromobj, FTAG, &dsfrom); rw_exit(&dp->dp_config_rwlock); - spa_close(spa, FTAG); if (error) { dsl_dataset_rele(ds, FTAG); return (error); @@ -3883,12 +4092,37 @@ dsl_dataset_rele(ds, FTAG); return (error); } - } else { - spa_close(spa, FTAG); + } + + if (zc->zc_obj) { + dsl_pool_t *dp = ds->ds_dir->dd_pool; + + if (fromsnap != NULL) { + dsl_dataset_rele(dsfrom, FTAG); + dsl_dataset_rele(ds, FTAG); + return (EINVAL); + } + + if (dsl_dir_is_clone(ds->ds_dir)) { + rw_enter(&dp->dp_config_rwlock, RW_READER); + error = dsl_dataset_hold_obj(dp, + ds->ds_dir->dd_phys->dd_origin_obj, FTAG, &dsfrom); + rw_exit(&dp->dp_config_rwlock); + if (error) { + dsl_dataset_rele(ds, FTAG); + return (error); + } + error = dmu_objset_from_ds(dsfrom, &fromsnap); + if (error) { + dsl_dataset_rele(dsfrom, FTAG); + dsl_dataset_rele(ds, FTAG); + return (error); + } + } } if (estimate) { - error = dmu_send_estimate(tosnap, fromsnap, zc->zc_obj, + error = dmu_send_estimate(tosnap, fromsnap, &zc->zc_objset_type); } else { file_t *fp = getf(zc->zc_cookie); @@ -3900,7 +4134,7 @@ } off = fp->f_offset; - error = dmu_send(tosnap, fromsnap, zc->zc_obj, + error = dmu_send(tosnap, fromsnap, zc->zc_cookie, fp->f_vnode, &off); if (VOP_SEEK(fp->f_vnode, fp->f_offset, &off, NULL) == 0) @@ -4414,6 +4648,7 @@ * zc_cleanup_fd cleanup-on-exit file descriptor for calling process * * outputs: + * zc_value short name of new snapshot */ static int zfs_ioc_tmp_snapshot(zfs_cmd_t *zc) @@ -4421,22 +4656,21 @@ char *snap_name; int error; - snap_name = kmem_asprintf("%s-%016llx", zc->zc_value, + snap_name = kmem_asprintf("%s@%s-%016llx", zc->zc_name, zc->zc_value, (u_longlong_t)ddi_get_lbolt64()); - if (strlen(snap_name) >= MAXNAMELEN) { + if (strlen(snap_name) >= MAXPATHLEN) { strfree(snap_name); return (E2BIG); } - error = dmu_objset_snapshot(zc->zc_name, snap_name, snap_name, - NULL, B_FALSE, B_TRUE, zc->zc_cleanup_fd); + error = dmu_objset_snapshot_tmp(snap_name, "%temp", zc->zc_cleanup_fd); if (error != 0) { strfree(snap_name); return (error); } - (void) strcpy(zc->zc_value, snap_name); + (void) strcpy(zc->zc_value, strchr(snap_name, '@') + 1); strfree(snap_name); return (0); } @@ -4790,172 +5024,427 @@ dsl_dataset_rele(new, FTAG); return (error); } - /* - * inputs: - * zc_name full name of last snapshot - * zc_value full name of first snapshot + * innvl: { + * "firstsnap" -> snapshot name + * } * - * outputs: - * zc_cookie space in bytes - * zc_objset_type compressed space in bytes - * zc_perm_action uncompressed space in bytes + * outnvl: { + * "used" -> space in bytes + * "compressed" -> compressed space in bytes + * "uncompressed" -> uncompressed space in bytes + * } */ static int -zfs_ioc_space_snaps(zfs_cmd_t *zc) +zfs_ioc_space_snaps(const char *lastsnap, nvlist_t *innvl, nvlist_t *outnvl) { int error; dsl_dataset_t *new, *old; - - error = dsl_dataset_hold(zc->zc_name, FTAG, &new); + char *firstsnap; + uint64_t used, comp, uncomp; + + if (nvlist_lookup_string(innvl, "firstsnap", &firstsnap) != 0) + return (EINVAL); + + error = dsl_dataset_hold(lastsnap, FTAG, &new); if (error != 0) return (error); - error = dsl_dataset_hold(zc->zc_value, FTAG, &old); + error = dsl_dataset_hold(firstsnap, FTAG, &old); if (error != 0) { dsl_dataset_rele(new, FTAG); return (error); } - error = dsl_dataset_space_wouldfree(old, new, &zc->zc_cookie, - &zc->zc_objset_type, &zc->zc_perm_action); + error = dsl_dataset_space_wouldfree(old, new, &used, &comp, &uncomp); dsl_dataset_rele(old, FTAG); dsl_dataset_rele(new, FTAG); + fnvlist_add_uint64(outnvl, "used", used); + fnvlist_add_uint64(outnvl, "compressed", comp); + fnvlist_add_uint64(outnvl, "uncompressed", uncomp); + return (error); +} + +/* + * innvl: { + * "fd" -> file descriptor to write stream to (int32) + * (optional) "fromsnap" -> full snap name to send an incremental from + * } + * + * outnvl is unused + */ +/* ARGSUSED */ +static int +zfs_ioc_send_new(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl) +{ + objset_t *fromsnap = NULL; + objset_t *tosnap; + int error; + offset_t off; + char *fromname; + int fd; + + error = nvlist_lookup_int32(innvl, "fd", &fd); + if (error != 0) + return (EINVAL); + + error = dmu_objset_hold(snapname, FTAG, &tosnap); + if (error) + return (error); + + error = nvlist_lookup_string(innvl, "fromsnap", &fromname); + if (error == 0) { + error = dmu_objset_hold(fromname, FTAG, &fromsnap); + if (error) { + dmu_objset_rele(tosnap, FTAG); + return (error); + } + } + + file_t *fp = getf(fd); + if (fp == NULL) { + dmu_objset_rele(tosnap, FTAG); + if (fromsnap != NULL) + dmu_objset_rele(fromsnap, FTAG); + return (EBADF); + } + + off = fp->f_offset; + error = dmu_send(tosnap, fromsnap, fd, fp->f_vnode, &off); + + if (VOP_SEEK(fp->f_vnode, fp->f_offset, &off, NULL) == 0) + fp->f_offset = off; + releasef(fd); + if (fromsnap != NULL) + dmu_objset_rele(fromsnap, FTAG); + dmu_objset_rele(tosnap, FTAG); return (error); } /* - * pool create, destroy, and export don't log the history as part of - * zfsdev_ioctl, but rather zfs_ioc_pool_create, and zfs_ioc_pool_export - * do the logging of those commands. + * Determine approximately how large a zfs send stream will be -- the number + * of bytes that will be written to the fd supplied to zfs_ioc_send_new(). + * + * innvl: { + * (optional) "fromsnap" -> full snap name to send an incremental from + * } + * + * outnvl: { + * "space" -> bytes of space (uint64) + * } + */ +static int +zfs_ioc_send_space(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl) +{ + objset_t *fromsnap = NULL; + objset_t *tosnap; + int error; + char *fromname; + uint64_t space; + + error = dmu_objset_hold(snapname, FTAG, &tosnap); + if (error) + return (error); + + error = nvlist_lookup_string(innvl, "fromsnap", &fromname); + if (error == 0) { + error = dmu_objset_hold(fromname, FTAG, &fromsnap); + if (error) { + dmu_objset_rele(tosnap, FTAG); + return (error); + } + } + + error = dmu_send_estimate(tosnap, fromsnap, &space); + fnvlist_add_uint64(outnvl, "space", space); + + if (fromsnap != NULL) + dmu_objset_rele(fromsnap, FTAG); + dmu_objset_rele(tosnap, FTAG); + return (error); +} + + +static zfs_ioc_vec_t zfs_ioc_vec[ZFS_IOC_LAST - ZFS_IOC_FIRST]; + +static void +zfs_ioctl_register_legacy(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, + zfs_secpolicy_func_t *secpolicy, zfs_ioc_namecheck_t namecheck, + boolean_t log_history, zfs_ioc_poolcheck_t pool_check) +{ + zfs_ioc_vec_t *vec = &zfs_ioc_vec[ioc - ZFS_IOC_FIRST]; + + ASSERT3U(ioc, >=, ZFS_IOC_FIRST); + ASSERT3U(ioc, <, ZFS_IOC_LAST); + ASSERT3P(vec->zvec_legacy_func, ==, NULL); + ASSERT3P(vec->zvec_func, ==, NULL); + + vec->zvec_legacy_func = func; + vec->zvec_secpolicy = secpolicy; + vec->zvec_namecheck = namecheck; + vec->zvec_allow_log = log_history; + vec->zvec_pool_check = pool_check; +} + +/* + * See the block comment at the beginning of this file for details on + * each argument to this function. */ -static zfs_ioc_vec_t zfs_ioc_vec[] = { - { zfs_ioc_pool_create, zfs_secpolicy_config, POOL_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_pool_destroy, zfs_secpolicy_config, POOL_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_pool_import, zfs_secpolicy_config, POOL_NAME, B_TRUE, - POOL_CHECK_NONE }, - { zfs_ioc_pool_export, zfs_secpolicy_config, POOL_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_pool_configs, zfs_secpolicy_none, NO_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_pool_stats, zfs_secpolicy_read, POOL_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_pool_tryimport, zfs_secpolicy_config, NO_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_pool_scan, zfs_secpolicy_config, POOL_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_pool_freeze, zfs_secpolicy_config, NO_NAME, B_FALSE, - POOL_CHECK_READONLY }, - { zfs_ioc_pool_upgrade, zfs_secpolicy_config, POOL_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_pool_get_history, zfs_secpolicy_config, POOL_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_vdev_add, zfs_secpolicy_config, POOL_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_vdev_remove, zfs_secpolicy_config, POOL_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_vdev_set_state, zfs_secpolicy_config, POOL_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_vdev_attach, zfs_secpolicy_config, POOL_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_vdev_detach, zfs_secpolicy_config, POOL_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_vdev_setpath, zfs_secpolicy_config, POOL_NAME, B_FALSE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_vdev_setfru, zfs_secpolicy_config, POOL_NAME, B_FALSE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_objset_stats, zfs_secpolicy_read, DATASET_NAME, B_FALSE, - POOL_CHECK_SUSPENDED }, - { zfs_ioc_objset_zplprops, zfs_secpolicy_read, DATASET_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_dataset_list_next, zfs_secpolicy_read, DATASET_NAME, B_FALSE, - POOL_CHECK_SUSPENDED }, - { zfs_ioc_snapshot_list_next, zfs_secpolicy_read, DATASET_NAME, B_FALSE, - POOL_CHECK_SUSPENDED }, - { zfs_ioc_set_prop, zfs_secpolicy_none, DATASET_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_create, zfs_secpolicy_create, DATASET_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_destroy, zfs_secpolicy_destroy, DATASET_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_rollback, zfs_secpolicy_rollback, DATASET_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_rename, zfs_secpolicy_rename, DATASET_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_recv, zfs_secpolicy_receive, DATASET_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_send, zfs_secpolicy_send, DATASET_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_inject_fault, zfs_secpolicy_inject, NO_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_clear_fault, zfs_secpolicy_inject, NO_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_inject_list_next, zfs_secpolicy_inject, NO_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_error_log, zfs_secpolicy_inject, POOL_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_clear, zfs_secpolicy_config, POOL_NAME, B_TRUE, - POOL_CHECK_NONE }, - { zfs_ioc_promote, zfs_secpolicy_promote, DATASET_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_snapshot, zfs_secpolicy_snapshot, DATASET_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_dsobj_to_dsname, zfs_secpolicy_diff, POOL_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_obj_to_path, zfs_secpolicy_diff, DATASET_NAME, B_FALSE, - POOL_CHECK_SUSPENDED }, - { zfs_ioc_pool_set_props, zfs_secpolicy_config, POOL_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_pool_get_props, zfs_secpolicy_read, POOL_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_set_fsacl, zfs_secpolicy_fsacl, DATASET_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_get_fsacl, zfs_secpolicy_read, DATASET_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_share, zfs_secpolicy_share, DATASET_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_inherit_prop, zfs_secpolicy_inherit, DATASET_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_smb_acl, zfs_secpolicy_smb_acl, DATASET_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_userspace_one, zfs_secpolicy_userspace_one, DATASET_NAME, - B_FALSE, POOL_CHECK_NONE }, - { zfs_ioc_userspace_many, zfs_secpolicy_userspace_many, DATASET_NAME, - B_FALSE, POOL_CHECK_NONE }, - { zfs_ioc_userspace_upgrade, zfs_secpolicy_userspace_upgrade, - DATASET_NAME, B_FALSE, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_hold, zfs_secpolicy_hold, DATASET_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_release, zfs_secpolicy_release, DATASET_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_get_holds, zfs_secpolicy_read, DATASET_NAME, B_FALSE, - POOL_CHECK_SUSPENDED }, - { zfs_ioc_objset_recvd_props, zfs_secpolicy_read, DATASET_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_vdev_split, zfs_secpolicy_config, POOL_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_next_obj, zfs_secpolicy_read, DATASET_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_diff, zfs_secpolicy_diff, DATASET_NAME, B_FALSE, - POOL_CHECK_NONE }, - { zfs_ioc_tmp_snapshot, zfs_secpolicy_tmp_snapshot, DATASET_NAME, - B_FALSE, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_obj_to_stats, zfs_secpolicy_diff, DATASET_NAME, B_FALSE, - POOL_CHECK_SUSPENDED }, - { zfs_ioc_space_written, zfs_secpolicy_read, DATASET_NAME, B_FALSE, - POOL_CHECK_SUSPENDED }, - { zfs_ioc_space_snaps, zfs_secpolicy_read, DATASET_NAME, B_FALSE, - POOL_CHECK_SUSPENDED }, - { zfs_ioc_destroy_snaps_nvl, zfs_secpolicy_destroy_recursive, - DATASET_NAME, B_TRUE, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_pool_reguid, zfs_secpolicy_config, POOL_NAME, B_TRUE, - POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY }, - { zfs_ioc_pool_reopen, zfs_secpolicy_config, POOL_NAME, B_TRUE, - POOL_CHECK_SUSPENDED }, - { zfs_ioc_send_progress, zfs_secpolicy_read, DATASET_NAME, B_FALSE, - POOL_CHECK_NONE } -}; +static void +zfs_ioctl_register(const char *name, zfs_ioc_t ioc, zfs_ioc_func_t *func, + zfs_secpolicy_func_t *secpolicy, zfs_ioc_namecheck_t namecheck, + zfs_ioc_poolcheck_t pool_check, boolean_t smush_outnvlist, + boolean_t allow_log) +{ + zfs_ioc_vec_t *vec = &zfs_ioc_vec[ioc - ZFS_IOC_FIRST]; + + ASSERT3U(ioc, >=, ZFS_IOC_FIRST); + ASSERT3U(ioc, <, ZFS_IOC_LAST); + ASSERT3P(vec->zvec_legacy_func, ==, NULL); + ASSERT3P(vec->zvec_func, ==, NULL); + + /* if we are logging, the name must be valid */ + ASSERT(!allow_log || namecheck != NO_NAME); + + vec->zvec_name = name; + vec->zvec_func = func; + vec->zvec_secpolicy = secpolicy; + vec->zvec_namecheck = namecheck; + vec->zvec_pool_check = pool_check; + vec->zvec_smush_outnvlist = smush_outnvlist; + vec->zvec_allow_log = allow_log; +} + +static void +zfs_ioctl_register_pool(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, + zfs_secpolicy_func_t *secpolicy, boolean_t log_history, + zfs_ioc_poolcheck_t pool_check) +{ + zfs_ioctl_register_legacy(ioc, func, secpolicy, + POOL_NAME, log_history, pool_check); +} + +static void +zfs_ioctl_register_dataset_nolog(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, + zfs_secpolicy_func_t *secpolicy, zfs_ioc_poolcheck_t pool_check) +{ + zfs_ioctl_register_legacy(ioc, func, secpolicy, + DATASET_NAME, B_FALSE, pool_check); +} + +static void +zfs_ioctl_register_pool_modify(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func) +{ + zfs_ioctl_register_legacy(ioc, func, zfs_secpolicy_config, + POOL_NAME, B_TRUE, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); +} + +static void +zfs_ioctl_register_pool_meta(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, + zfs_secpolicy_func_t *secpolicy) +{ + zfs_ioctl_register_legacy(ioc, func, secpolicy, + NO_NAME, B_FALSE, POOL_CHECK_NONE); +} + +static void +zfs_ioctl_register_dataset_read_secpolicy(zfs_ioc_t ioc, + zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy) +{ + zfs_ioctl_register_legacy(ioc, func, secpolicy, + DATASET_NAME, B_FALSE, POOL_CHECK_SUSPENDED); +} + +static void +zfs_ioctl_register_dataset_read(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func) +{ + zfs_ioctl_register_dataset_read_secpolicy(ioc, func, + zfs_secpolicy_read); +} + +static void +zfs_ioctl_register_dataset_modify(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, + zfs_secpolicy_func_t *secpolicy) +{ + zfs_ioctl_register_legacy(ioc, func, secpolicy, + DATASET_NAME, B_TRUE, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); +} + +static void +zfs_ioctl_init(void) +{ + zfs_ioctl_register("snapshot", ZFS_IOC_SNAPSHOT, + zfs_ioc_snapshot, zfs_secpolicy_snapshot, POOL_NAME, + POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); + + zfs_ioctl_register("log_history", ZFS_IOC_LOG_HISTORY, + zfs_ioc_log_history, zfs_secpolicy_log_history, NO_NAME, + POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_FALSE); + + zfs_ioctl_register("space_snaps", ZFS_IOC_SPACE_SNAPS, + zfs_ioc_space_snaps, zfs_secpolicy_read, DATASET_NAME, + POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE); + + zfs_ioctl_register("send", ZFS_IOC_SEND_NEW, + zfs_ioc_send_new, zfs_secpolicy_send_new, DATASET_NAME, + POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE); + + zfs_ioctl_register("send_space", ZFS_IOC_SEND_SPACE, + zfs_ioc_send_space, zfs_secpolicy_read, DATASET_NAME, + POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE); + + zfs_ioctl_register("create", ZFS_IOC_CREATE, + zfs_ioc_create, zfs_secpolicy_create_clone, DATASET_NAME, + POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); + + zfs_ioctl_register("clone", ZFS_IOC_CLONE, + zfs_ioc_clone, zfs_secpolicy_create_clone, DATASET_NAME, + POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); + + zfs_ioctl_register("destroy_snaps", ZFS_IOC_DESTROY_SNAPS, + zfs_ioc_destroy_snaps, zfs_secpolicy_destroy_snaps, POOL_NAME, + POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); + + /* IOCTLS that use the legacy function signature */ + + zfs_ioctl_register_legacy(ZFS_IOC_POOL_FREEZE, zfs_ioc_pool_freeze, + zfs_secpolicy_config, NO_NAME, B_FALSE, POOL_CHECK_READONLY); + + zfs_ioctl_register_pool(ZFS_IOC_POOL_CREATE, zfs_ioc_pool_create, + zfs_secpolicy_config, B_TRUE, POOL_CHECK_NONE); + zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_SCAN, + zfs_ioc_pool_scan); + zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_UPGRADE, + zfs_ioc_pool_upgrade); + zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_ADD, + zfs_ioc_vdev_add); + zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_REMOVE, + zfs_ioc_vdev_remove); + zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SET_STATE, + zfs_ioc_vdev_set_state); + zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_ATTACH, + zfs_ioc_vdev_attach); + zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_DETACH, + zfs_ioc_vdev_detach); + zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SETPATH, + zfs_ioc_vdev_setpath); + zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SETFRU, + zfs_ioc_vdev_setfru); + zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_SET_PROPS, + zfs_ioc_pool_set_props); + zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SPLIT, + zfs_ioc_vdev_split); + zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_REGUID, + zfs_ioc_pool_reguid); + + zfs_ioctl_register_pool_meta(ZFS_IOC_POOL_CONFIGS, + zfs_ioc_pool_configs, zfs_secpolicy_none); + zfs_ioctl_register_pool_meta(ZFS_IOC_POOL_TRYIMPORT, + zfs_ioc_pool_tryimport, zfs_secpolicy_config); + zfs_ioctl_register_pool_meta(ZFS_IOC_INJECT_FAULT, + zfs_ioc_inject_fault, zfs_secpolicy_inject); + zfs_ioctl_register_pool_meta(ZFS_IOC_CLEAR_FAULT, + zfs_ioc_clear_fault, zfs_secpolicy_inject); + zfs_ioctl_register_pool_meta(ZFS_IOC_INJECT_LIST_NEXT, + zfs_ioc_inject_list_next, zfs_secpolicy_inject); + + /* + * pool destroy, and export don't log the history as part of + * zfsdev_ioctl, but rather zfs_ioc_pool_export + * does the logging of those commands. + */ + zfs_ioctl_register_pool(ZFS_IOC_POOL_DESTROY, zfs_ioc_pool_destroy, + zfs_secpolicy_config, B_FALSE, POOL_CHECK_NONE); + zfs_ioctl_register_pool(ZFS_IOC_POOL_EXPORT, zfs_ioc_pool_export, + zfs_secpolicy_config, B_FALSE, POOL_CHECK_NONE); + + zfs_ioctl_register_pool(ZFS_IOC_POOL_STATS, zfs_ioc_pool_stats, + zfs_secpolicy_read, B_FALSE, POOL_CHECK_NONE); + zfs_ioctl_register_pool(ZFS_IOC_POOL_GET_PROPS, zfs_ioc_pool_get_props, + zfs_secpolicy_read, B_FALSE, POOL_CHECK_NONE); + + zfs_ioctl_register_pool(ZFS_IOC_ERROR_LOG, zfs_ioc_error_log, + zfs_secpolicy_inject, B_FALSE, POOL_CHECK_SUSPENDED); + zfs_ioctl_register_pool(ZFS_IOC_DSOBJ_TO_DSNAME, + zfs_ioc_dsobj_to_dsname, + zfs_secpolicy_diff, B_FALSE, POOL_CHECK_SUSPENDED); + zfs_ioctl_register_pool(ZFS_IOC_POOL_GET_HISTORY, + zfs_ioc_pool_get_history, + zfs_secpolicy_config, B_FALSE, POOL_CHECK_SUSPENDED); + + zfs_ioctl_register_pool(ZFS_IOC_POOL_IMPORT, zfs_ioc_pool_import, + zfs_secpolicy_config, B_TRUE, POOL_CHECK_NONE); + + zfs_ioctl_register_pool(ZFS_IOC_CLEAR, zfs_ioc_clear, + zfs_secpolicy_config, B_TRUE, POOL_CHECK_SUSPENDED); + zfs_ioctl_register_pool(ZFS_IOC_POOL_REOPEN, zfs_ioc_pool_reopen, + zfs_secpolicy_config, B_TRUE, POOL_CHECK_SUSPENDED); + + zfs_ioctl_register_dataset_read(ZFS_IOC_SPACE_WRITTEN, + zfs_ioc_space_written); + zfs_ioctl_register_dataset_read(ZFS_IOC_GET_HOLDS, + zfs_ioc_get_holds); + zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_RECVD_PROPS, + zfs_ioc_objset_recvd_props); + zfs_ioctl_register_dataset_read(ZFS_IOC_NEXT_OBJ, + zfs_ioc_next_obj); + zfs_ioctl_register_dataset_read(ZFS_IOC_GET_FSACL, + zfs_ioc_get_fsacl); + zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_STATS, + zfs_ioc_objset_stats); + zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_ZPLPROPS, + zfs_ioc_objset_zplprops); + zfs_ioctl_register_dataset_read(ZFS_IOC_DATASET_LIST_NEXT, + zfs_ioc_dataset_list_next); + zfs_ioctl_register_dataset_read(ZFS_IOC_SNAPSHOT_LIST_NEXT, + zfs_ioc_snapshot_list_next); + zfs_ioctl_register_dataset_read(ZFS_IOC_SEND_PROGRESS, + zfs_ioc_send_progress); + + zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_DIFF, + zfs_ioc_diff, zfs_secpolicy_diff); + zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_OBJ_TO_STATS, + zfs_ioc_obj_to_stats, zfs_secpolicy_diff); + zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_OBJ_TO_PATH, + zfs_ioc_obj_to_path, zfs_secpolicy_diff); + zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_USERSPACE_ONE, + zfs_ioc_userspace_one, zfs_secpolicy_userspace_one); + zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_USERSPACE_MANY, + zfs_ioc_userspace_many, zfs_secpolicy_userspace_many); + zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_SEND, + zfs_ioc_send, zfs_secpolicy_send); + + zfs_ioctl_register_dataset_modify(ZFS_IOC_SET_PROP, zfs_ioc_set_prop, + zfs_secpolicy_none); + zfs_ioctl_register_dataset_modify(ZFS_IOC_DESTROY, zfs_ioc_destroy, + zfs_secpolicy_destroy); + zfs_ioctl_register_dataset_modify(ZFS_IOC_ROLLBACK, zfs_ioc_rollback, + zfs_secpolicy_rollback); + zfs_ioctl_register_dataset_modify(ZFS_IOC_RENAME, zfs_ioc_rename, + zfs_secpolicy_rename); + zfs_ioctl_register_dataset_modify(ZFS_IOC_RECV, zfs_ioc_recv, + zfs_secpolicy_recv); + zfs_ioctl_register_dataset_modify(ZFS_IOC_PROMOTE, zfs_ioc_promote, + zfs_secpolicy_promote); + zfs_ioctl_register_dataset_modify(ZFS_IOC_HOLD, zfs_ioc_hold, + zfs_secpolicy_hold); + zfs_ioctl_register_dataset_modify(ZFS_IOC_RELEASE, zfs_ioc_release, + zfs_secpolicy_release); + zfs_ioctl_register_dataset_modify(ZFS_IOC_INHERIT_PROP, + zfs_ioc_inherit_prop, zfs_secpolicy_inherit_prop); + zfs_ioctl_register_dataset_modify(ZFS_IOC_SET_FSACL, zfs_ioc_set_fsacl, + zfs_secpolicy_set_fsacl); + + zfs_ioctl_register_dataset_nolog(ZFS_IOC_SHARE, zfs_ioc_share, + zfs_secpolicy_share, POOL_CHECK_NONE); + zfs_ioctl_register_dataset_nolog(ZFS_IOC_SMB_ACL, zfs_ioc_smb_acl, + zfs_secpolicy_smb_acl, POOL_CHECK_NONE); + zfs_ioctl_register_dataset_nolog(ZFS_IOC_USERSPACE_UPGRADE, + zfs_ioc_userspace_upgrade, zfs_secpolicy_userspace_upgrade, + POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); + zfs_ioctl_register_dataset_nolog(ZFS_IOC_TMP_SNAPSHOT, + zfs_ioc_tmp_snapshot, zfs_secpolicy_tmp_snapshot, + POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); +} int pool_status_check(const char *name, zfs_ioc_namecheck_t type, @@ -5092,67 +5581,145 @@ zfsdev_ioctl(dev_t dev, int cmd, intptr_t arg, int flag, cred_t *cr, int *rvalp) { zfs_cmd_t *zc; - uint_t vec; - int error, rc; + uint_t vecnum; + int error, rc, len; minor_t minor = getminor(dev); + const zfs_ioc_vec_t *vec; + char *saved_poolname = NULL; + nvlist_t *innvl = NULL; if (minor != 0 && zfsdev_get_soft_state(minor, ZSST_CTLDEV) == NULL) return (zvol_ioctl(dev, cmd, arg, flag, cr, rvalp)); - vec = cmd - ZFS_IOC; + vecnum = cmd - ZFS_IOC_FIRST; ASSERT3U(getmajor(dev), ==, ddi_driver_major(zfs_dip)); - if (vec >= sizeof (zfs_ioc_vec) / sizeof (zfs_ioc_vec[0])) + if (vecnum >= sizeof (zfs_ioc_vec) / sizeof (zfs_ioc_vec[0])) return (EINVAL); + vec = &zfs_ioc_vec[vecnum]; zc = kmem_zalloc(sizeof (zfs_cmd_t), KM_SLEEP); error = ddi_copyin((void *)arg, zc, sizeof (zfs_cmd_t), flag); - if (error != 0) + if (error != 0) { error = EFAULT; - - if ((error == 0) && !(flag & FKIOCTL)) - error = zfs_ioc_vec[vec].zvec_secpolicy(zc, cr); + goto out; + } + + zc->zc_iflags = flag & FKIOCTL; + if (zc->zc_nvlist_src_size != 0) { + error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, + zc->zc_iflags, &innvl); + if (error != 0) + goto out; + } /* * Ensure that all pool/dataset names are valid before we pass down to * the lower layers. */ - if (error == 0) { - zc->zc_name[sizeof (zc->zc_name) - 1] = '\0'; - zc->zc_iflags = flag & FKIOCTL; - switch (zfs_ioc_vec[vec].zvec_namecheck) { - case POOL_NAME: - if (pool_namecheck(zc->zc_name, NULL, NULL) != 0) - error = EINVAL; + zc->zc_name[sizeof (zc->zc_name) - 1] = '\0'; + switch (vec->zvec_namecheck) { + case POOL_NAME: + if (pool_namecheck(zc->zc_name, NULL, NULL) != 0) + error = EINVAL; + else error = pool_status_check(zc->zc_name, - zfs_ioc_vec[vec].zvec_namecheck, - zfs_ioc_vec[vec].zvec_pool_check); - break; - - case DATASET_NAME: - if (dataset_namecheck(zc->zc_name, NULL, NULL) != 0) - error = EINVAL; + vec->zvec_namecheck, vec->zvec_pool_check); + break; + + case DATASET_NAME: + if (dataset_namecheck(zc->zc_name, NULL, NULL) != 0) + error = EINVAL; + else error = pool_status_check(zc->zc_name, - zfs_ioc_vec[vec].zvec_namecheck, - zfs_ioc_vec[vec].zvec_pool_check); - break; - - case NO_NAME: - break; - } + vec->zvec_namecheck, vec->zvec_pool_check); + break; + + case NO_NAME: + break; } - if (error == 0) - error = zfs_ioc_vec[vec].zvec_func(zc); - + + if (error == 0 && !(flag & FKIOCTL)) + error = vec->zvec_secpolicy(zc, innvl, cr); + + if (error != 0) + goto out; + + /* legacy ioctls can modify zc_name */ + len = strcspn(zc->zc_name, "/@") + 1; + saved_poolname = kmem_alloc(len, KM_SLEEP); + (void) strlcpy(saved_poolname, zc->zc_name, len); + + if (vec->zvec_func != NULL) { + nvlist_t *outnvl; + int puterror = 0; + spa_t *spa; + nvlist_t *lognv = NULL; + + ASSERT(vec->zvec_legacy_func == NULL); + + /* + * Add the innvl to the lognv before calling the func, + * in case the func changes the innvl. + */ + if (vec->zvec_allow_log) { + lognv = fnvlist_alloc(); + fnvlist_add_string(lognv, ZPOOL_HIST_IOCTL, + vec->zvec_name); + if (!nvlist_empty(innvl)) { + fnvlist_add_nvlist(lognv, ZPOOL_HIST_INPUT_NVL, + innvl); + } + } + + outnvl = fnvlist_alloc(); + error = vec->zvec_func(zc->zc_name, innvl, outnvl); + + if (error == 0 && vec->zvec_allow_log && + spa_open(zc->zc_name, &spa, FTAG) == 0) { + if (!nvlist_empty(outnvl)) { + fnvlist_add_nvlist(lognv, ZPOOL_HIST_OUTPUT_NVL, + outnvl); + } + (void) spa_history_log_nvl(spa, lognv); + spa_close(spa, FTAG); + } + fnvlist_free(lognv); + + if (!nvlist_empty(outnvl) || zc->zc_nvlist_dst_size != 0) { + int smusherror = 0; + if (vec->zvec_smush_outnvlist) { + smusherror = nvlist_smush(outnvl, + zc->zc_nvlist_dst_size); + } + if (smusherror == 0) + puterror = put_nvlist(zc, outnvl); + } + + if (puterror != 0) + error = puterror; + + nvlist_free(outnvl); + } else { + error = vec->zvec_legacy_func(zc); + } + +out: + nvlist_free(innvl); rc = ddi_copyout(zc, (void *)arg, sizeof (zfs_cmd_t), flag); - if (error == 0) { - if (rc != 0) - error = EFAULT; - if (zfs_ioc_vec[vec].zvec_his_log) - zfs_log_history(zc); + if (error == 0 && rc != 0) + error = EFAULT; + if (error == 0 && vec->zvec_allow_log) { + char *s = tsd_get(zfs_allow_log_key); + if (s != NULL) + strfree(s); + (void) tsd_set(zfs_allow_log_key, saved_poolname); + } else { + if (saved_poolname != NULL) + strfree(saved_poolname); } kmem_free(zc, sizeof (zfs_cmd_t)); @@ -5268,9 +5835,12 @@ NULL }; - -uint_t zfs_fsyncer_key; -extern uint_t rrw_tsd_key; +static void +zfs_allow_log_destroy(void *arg) +{ + char *poolname = arg; + strfree(poolname); +} int _init(void) @@ -5280,6 +5850,7 @@ spa_init(FREAD | FWRITE); zfs_init(); zvol_init(); + zfs_ioctl_init(); if ((error = mod_install(&modlinkage)) != 0) { zvol_fini(); @@ -5289,7 +5860,8 @@ } tsd_create(&zfs_fsyncer_key, NULL); - tsd_create(&rrw_tsd_key, NULL); + tsd_create(&rrw_tsd_key, rrw_tsd_destroy); + tsd_create(&zfs_allow_log_key, zfs_allow_log_destroy); error = ldi_ident_from_mod(&modlinkage, &zfs_li); ASSERT(error == 0); diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/zfs_vfsops.c --- a/usr/src/uts/common/fs/zfs/zfs_vfsops.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/zfs_vfsops.c Sat Jun 30 13:00:47 2012 -0700 @@ -20,6 +20,7 @@ */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ /* Portions Copyright 2010 Robert Milkowski */ @@ -2248,9 +2249,8 @@ sa_register_update_callback(os, zfs_sa_upgrade); } - spa_history_log_internal(LOG_DS_UPGRADE, - dmu_objset_spa(os), tx, "oldver=%llu newver=%llu dataset = %llu", - zfsvfs->z_version, newvers, dmu_objset_id(os)); + spa_history_log_internal_ds(dmu_objset_ds(os), "upgrade", tx, + "from %llu to %llu", zfsvfs->z_version, newvers); dmu_tx_commit(tx); diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/fs/zfs/zvol.c --- a/usr/src/uts/common/fs/zfs/zvol.c Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/fs/zfs/zvol.c Sat Jun 30 13:00:47 2012 -0700 @@ -24,6 +24,7 @@ * Portions Copyright 2010 Robert Milkowski * * Copyright 2011 Nexenta Systems, Inc. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ /* @@ -135,7 +136,7 @@ int zvol_maxphys = DMU_MAX_ACCESS/2; extern int zfs_set_prop_nvlist(const char *, zprop_source_t, - nvlist_t *, nvlist_t **); + nvlist_t *, nvlist_t *); static int zvol_remove_zv(zvol_state_t *); static int zvol_get_data(void *arg, lr_write_t *lr, char *buf, zio_t *zio); static int zvol_dumpify(zvol_state_t *zv); @@ -1885,7 +1886,7 @@ if (zap_lookup(zv->zv_objset, ZVOL_ZAP_OBJ, ZVOL_DUMPSIZE, 8, 1, &dumpsize) != 0 || dumpsize != zv->zv_volsize) { - boolean_t resize = (dumpsize > 0) ? B_TRUE : B_FALSE; + boolean_t resize = (dumpsize > 0); if ((error = zvol_dump_init(zv, resize)) != 0) { (void) zvol_dump_fini(zv); diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/sys/fs/zfs.h --- a/usr/src/uts/common/sys/fs/zfs.h Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/sys/fs/zfs.h Sat Jun 30 13:00:47 2012 -0700 @@ -52,6 +52,16 @@ ZFS_TYPE_POOL = 0x8 } zfs_type_t; +typedef enum dmu_objset_type { + DMU_OST_NONE, + DMU_OST_META, + DMU_OST_ZFS, + DMU_OST_ZVOL, + DMU_OST_OTHER, /* For testing only! */ + DMU_OST_ANY, /* Be careful! */ + DMU_OST_NUMTYPES +} dmu_objset_type_t; + #define ZFS_TYPE_DATASET \ (ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME | ZFS_TYPE_SNAPSHOT) @@ -744,10 +754,10 @@ /* * /dev/zfs ioctl numbers. */ -#define ZFS_IOC ('Z' << 8) - typedef enum zfs_ioc { - ZFS_IOC_POOL_CREATE = ZFS_IOC, + ZFS_IOC_FIRST = ('Z' << 8), + ZFS_IOC = ZFS_IOC_FIRST, + ZFS_IOC_POOL_CREATE = ZFS_IOC_FIRST, ZFS_IOC_POOL_DESTROY, ZFS_IOC_POOL_IMPORT, ZFS_IOC_POOL_EXPORT, @@ -806,10 +816,15 @@ ZFS_IOC_OBJ_TO_STATS, ZFS_IOC_SPACE_WRITTEN, ZFS_IOC_SPACE_SNAPS, - ZFS_IOC_DESTROY_SNAPS_NVL, + ZFS_IOC_DESTROY_SNAPS, ZFS_IOC_POOL_REGUID, ZFS_IOC_POOL_REOPEN, - ZFS_IOC_SEND_PROGRESS + ZFS_IOC_SEND_PROGRESS, + ZFS_IOC_LOG_HISTORY, + ZFS_IOC_SEND_NEW, + ZFS_IOC_SEND_SPACE, + ZFS_IOC_CLONE, + ZFS_IOC_LAST } zfs_ioc_t; /* @@ -846,6 +861,12 @@ #define ZPOOL_HIST_TXG "history txg" #define ZPOOL_HIST_INT_EVENT "history internal event" #define ZPOOL_HIST_INT_STR "history internal str" +#define ZPOOL_HIST_INT_NAME "internal_name" +#define ZPOOL_HIST_IOCTL "ioctl" +#define ZPOOL_HIST_INPUT_NVL "in_nvl" +#define ZPOOL_HIST_OUTPUT_NVL "out_nvl" +#define ZPOOL_HIST_DSNAME "dsname" +#define ZPOOL_HIST_DSID "dsid" /* * Flags for ZFS_IOC_VDEV_SET_STATE @@ -891,56 +912,6 @@ #define ZFS_EV_VDEV_PATH "vdev_path" #define ZFS_EV_VDEV_GUID "vdev_guid" -/* - * Note: This is encoded on-disk, so new events must be added to the - * end, and unused events can not be removed. Be sure to edit - * libzfs_pool.c: hist_event_table[]. - */ -typedef enum history_internal_events { - LOG_NO_EVENT = 0, - LOG_POOL_CREATE, - LOG_POOL_VDEV_ADD, - LOG_POOL_REMOVE, - LOG_POOL_DESTROY, - LOG_POOL_EXPORT, - LOG_POOL_IMPORT, - LOG_POOL_VDEV_ATTACH, - LOG_POOL_VDEV_REPLACE, - LOG_POOL_VDEV_DETACH, - LOG_POOL_VDEV_ONLINE, - LOG_POOL_VDEV_OFFLINE, - LOG_POOL_UPGRADE, - LOG_POOL_CLEAR, - LOG_POOL_SCAN, - LOG_POOL_PROPSET, - LOG_DS_CREATE, - LOG_DS_CLONE, - LOG_DS_DESTROY, - LOG_DS_DESTROY_BEGIN, - LOG_DS_INHERIT, - LOG_DS_PROPSET, - LOG_DS_QUOTA, - LOG_DS_PERM_UPDATE, - LOG_DS_PERM_REMOVE, - LOG_DS_PERM_WHO_REMOVE, - LOG_DS_PROMOTE, - LOG_DS_RECEIVE, - LOG_DS_RENAME, - LOG_DS_RESERVATION, - LOG_DS_REPLAY_INC_SYNC, - LOG_DS_REPLAY_FULL_SYNC, - LOG_DS_ROLLBACK, - LOG_DS_SNAPSHOT, - LOG_DS_UPGRADE, - LOG_DS_REFQUOTA, - LOG_DS_REFRESERV, - LOG_POOL_SCAN_DONE, - LOG_DS_USER_HOLD, - LOG_DS_USER_RELEASE, - LOG_POOL_SPLIT, - LOG_END -} history_internal_events_t; - #ifdef __cplusplus } #endif diff -r b6bbdd77139c -r 95aba6e49b9f usr/src/uts/common/sys/sunddi.h --- a/usr/src/uts/common/sys/sunddi.h Thu Jun 14 12:09:31 2012 +0400 +++ b/usr/src/uts/common/sys/sunddi.h Sat Jun 30 13:00:47 2012 -0700 @@ -22,6 +22,7 @@ /* * Copyright (c) 1990, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright 2012 Garrett D'Amore . All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. */ #ifndef _SYS_SUNDDI_H @@ -460,6 +461,7 @@ extern size_t strlcat(char *, const char *, size_t); extern size_t strlcpy(char *, const char *, size_t); extern size_t strspn(const char *, const char *); +extern size_t strcspn(const char *, const char *); extern int bcmp(const void *, const void *, size_t) __PURE; extern int stoi(char **); extern void numtos(ulong_t, char *);