diff --git a/cmd/zfs/zfs_main.c b/cmd/zfs/zfs_main.c index 42c180890fec..f609a4e70677 100644 --- a/cmd/zfs/zfs_main.c +++ b/cmd/zfs/zfs_main.c @@ -1,8640 +1,8640 @@ /* * 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) 2011, 2020 by Delphix. All rights reserved. * Copyright 2012 Milan Jurik. All rights reserved. * Copyright (c) 2012, Joyent, Inc. All rights reserved. * Copyright (c) 2013 Steven Hartland. All rights reserved. * Copyright 2016 Igor Kozhukhov . * Copyright 2016 Nexenta Systems, Inc. * Copyright (c) 2019 Datto Inc. * Copyright (c) 2019, loli10K * Copyright 2019 Joyent, Inc. * Copyright (c) 2019, 2020 by Christian Schwarz. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef HAVE_IDMAP #include #include #endif /* HAVE_IDMAP */ #include "zfs_iter.h" #include "zfs_util.h" #include "zfs_comutil.h" #include "libzfs_impl.h" #include "zfs_projectutil.h" libzfs_handle_t *g_zfs; 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); static int zfs_do_destroy(int argc, char **argv); static int zfs_do_get(int argc, char **argv); static int zfs_do_inherit(int argc, char **argv); static int zfs_do_list(int argc, char **argv); static int zfs_do_mount(int argc, char **argv); static int zfs_do_rename(int argc, char **argv); static int zfs_do_rollback(int argc, char **argv); static int zfs_do_set(int argc, char **argv); static int zfs_do_upgrade(int argc, char **argv); static int zfs_do_snapshot(int argc, char **argv); static int zfs_do_unmount(int argc, char **argv); static int zfs_do_share(int argc, char **argv); static int zfs_do_unshare(int argc, char **argv); static int zfs_do_send(int argc, char **argv); static int zfs_do_receive(int argc, char **argv); static int zfs_do_promote(int argc, char **argv); static int zfs_do_userspace(int argc, char **argv); static int zfs_do_allow(int argc, char **argv); static int zfs_do_unallow(int argc, char **argv); static int zfs_do_hold(int argc, char **argv); static int zfs_do_holds(int argc, char **argv); static int zfs_do_release(int argc, char **argv); static int zfs_do_diff(int argc, char **argv); static int zfs_do_bookmark(int argc, char **argv); static int zfs_do_channel_program(int argc, char **argv); static int zfs_do_load_key(int argc, char **argv); static int zfs_do_unload_key(int argc, char **argv); static int zfs_do_change_key(int argc, char **argv); static int zfs_do_project(int argc, char **argv); static int zfs_do_version(int argc, char **argv); static int zfs_do_redact(int argc, char **argv); static int zfs_do_wait(int argc, char **argv); #ifdef __FreeBSD__ static int zfs_do_jail(int argc, char **argv); static int zfs_do_unjail(int argc, char **argv); #endif /* * Enable a reasonable set of defaults for libumem debugging on DEBUG builds. */ #ifdef DEBUG const char * _umem_debug_init(void) { return ("default,verbose"); /* $UMEM_DEBUG setting */ } const char * _umem_logging_init(void) { return ("fail,contents"); /* $UMEM_LOGGING setting */ } #endif typedef enum { HELP_CLONE, HELP_CREATE, HELP_DESTROY, HELP_GET, HELP_INHERIT, HELP_UPGRADE, HELP_LIST, HELP_MOUNT, HELP_PROMOTE, HELP_RECEIVE, HELP_RENAME, HELP_ROLLBACK, HELP_SEND, HELP_SET, HELP_SHARE, HELP_SNAPSHOT, HELP_UNMOUNT, HELP_UNSHARE, HELP_ALLOW, HELP_UNALLOW, HELP_USERSPACE, HELP_GROUPSPACE, HELP_PROJECTSPACE, HELP_PROJECT, HELP_HOLD, HELP_HOLDS, HELP_RELEASE, HELP_DIFF, HELP_BOOKMARK, HELP_CHANNEL_PROGRAM, HELP_LOAD_KEY, HELP_UNLOAD_KEY, HELP_CHANGE_KEY, HELP_VERSION, HELP_REDACT, HELP_JAIL, HELP_UNJAIL, HELP_WAIT, } zfs_help_t; typedef struct zfs_command { const char *name; int (*func)(int argc, char **argv); zfs_help_t usage; } zfs_command_t; /* * Master command table. Each ZFS command has a name, associated function, and * usage message. The usage messages need to be internationalized, so we have * to have a function to return the usage message based on a command index. * * These commands are organized according to how they are displayed in the usage * message. An empty command (one with a NULL name) indicates an empty line in * the generic usage message. */ static zfs_command_t command_table[] = { { "version", zfs_do_version, HELP_VERSION }, { NULL }, { "create", zfs_do_create, HELP_CREATE }, { "destroy", zfs_do_destroy, HELP_DESTROY }, { NULL }, { "snapshot", zfs_do_snapshot, HELP_SNAPSHOT }, { "rollback", zfs_do_rollback, HELP_ROLLBACK }, { "clone", zfs_do_clone, HELP_CLONE }, { "promote", zfs_do_promote, HELP_PROMOTE }, { "rename", zfs_do_rename, HELP_RENAME }, { "bookmark", zfs_do_bookmark, HELP_BOOKMARK }, { "program", zfs_do_channel_program, HELP_CHANNEL_PROGRAM }, { NULL }, { "list", zfs_do_list, HELP_LIST }, { NULL }, { "set", zfs_do_set, HELP_SET }, { "get", zfs_do_get, HELP_GET }, { "inherit", zfs_do_inherit, HELP_INHERIT }, { "upgrade", zfs_do_upgrade, HELP_UPGRADE }, { NULL }, { "userspace", zfs_do_userspace, HELP_USERSPACE }, { "groupspace", zfs_do_userspace, HELP_GROUPSPACE }, { "projectspace", zfs_do_userspace, HELP_PROJECTSPACE }, { NULL }, { "project", zfs_do_project, HELP_PROJECT }, { NULL }, { "mount", zfs_do_mount, HELP_MOUNT }, { "unmount", zfs_do_unmount, HELP_UNMOUNT }, { "share", zfs_do_share, HELP_SHARE }, { "unshare", zfs_do_unshare, HELP_UNSHARE }, { NULL }, { "send", zfs_do_send, HELP_SEND }, { "receive", zfs_do_receive, HELP_RECEIVE }, { NULL }, { "allow", zfs_do_allow, HELP_ALLOW }, { NULL }, { "unallow", zfs_do_unallow, HELP_UNALLOW }, { NULL }, { "hold", zfs_do_hold, HELP_HOLD }, { "holds", zfs_do_holds, HELP_HOLDS }, { "release", zfs_do_release, HELP_RELEASE }, { "diff", zfs_do_diff, HELP_DIFF }, { "load-key", zfs_do_load_key, HELP_LOAD_KEY }, { "unload-key", zfs_do_unload_key, HELP_UNLOAD_KEY }, { "change-key", zfs_do_change_key, HELP_CHANGE_KEY }, { "redact", zfs_do_redact, HELP_REDACT }, { "wait", zfs_do_wait, HELP_WAIT }, #ifdef __FreeBSD__ { "jail", zfs_do_jail, HELP_JAIL }, { "unjail", zfs_do_unjail, HELP_UNJAIL }, #endif }; #define NCOMMAND (sizeof (command_table) / sizeof (command_table[0])) zfs_command_t *current_command; static const char * get_usage(zfs_help_t idx) { switch (idx) { case HELP_CLONE: return (gettext("\tclone [-p] [-o property=value] ... " " \n")); case HELP_CREATE: return (gettext("\tcreate [-Pnpv] [-o property=value] ... " "\n" "\tcreate [-Pnpsv] [-b blocksize] [-o property=value] ... " "-V \n")); case HELP_DESTROY: return (gettext("\tdestroy [-fnpRrv] \n" "\tdestroy [-dnpRrv] " "@[%][,...]\n" "\tdestroy #\n")); case HELP_GET: return (gettext("\tget [-rHp] [-d max] " "[-o \"all\" | field[,...]]\n" "\t [-t type[,...]] [-s source[,...]]\n" "\t <\"all\" | property[,...]> " "[filesystem|volume|snapshot|bookmark] ...\n")); case HELP_INHERIT: return (gettext("\tinherit [-rS] " " ...\n")); case HELP_UPGRADE: return (gettext("\tupgrade [-v]\n" "\tupgrade [-r] [-V version] <-a | filesystem ...>\n")); case HELP_LIST: return (gettext("\tlist [-Hp] [-r|-d max] [-o property[,...]] " "[-s property]...\n\t [-S property]... [-t type[,...]] " "[filesystem|volume|snapshot] ...\n")); case HELP_MOUNT: return (gettext("\tmount\n" "\tmount [-flvO] [-o opts] <-a | filesystem>\n")); case HELP_PROMOTE: return (gettext("\tpromote \n")); case HELP_RECEIVE: return (gettext("\treceive [-vMnsFhu] " "[-o =] ... [-x ] ...\n" "\t \n" "\treceive [-vMnsFhu] [-o =] ... " "[-x ] ... \n" "\t [-d | -e] \n" "\treceive -A \n")); case HELP_RENAME: return (gettext("\trename [-f] " "\n" "\trename -p [-f] \n" "\trename -u [-f] \n" "\trename -r \n")); case HELP_ROLLBACK: return (gettext("\trollback [-rRf] \n")); case HELP_SEND: return (gettext("\tsend [-DnPpRvLecwhb] [-[i|I] snapshot] " "\n" "\tsend [-nvPLecw] [-i snapshot|bookmark] " "\n" "\tsend [-DnPpvLec] [-i bookmark|snapshot] " "--redact \n" "\tsend [-nvPe] -t \n" "\tsend [-Pnv] --saved filesystem\n")); case HELP_SET: return (gettext("\tset ... " " ...\n")); case HELP_SHARE: return (gettext("\tshare [-l] <-a [nfs|smb] | filesystem>\n")); case HELP_SNAPSHOT: return (gettext("\tsnapshot [-r] [-o property=value] ... " "@ ...\n")); case HELP_UNMOUNT: return (gettext("\tunmount [-fu] " "<-a | filesystem|mountpoint>\n")); case HELP_UNSHARE: return (gettext("\tunshare " "<-a [nfs|smb] | filesystem|mountpoint>\n")); case HELP_ALLOW: return (gettext("\tallow \n" "\tallow [-ldug] " "<\"everyone\"|user|group>[,...] [,...]\n" "\t \n" "\tallow [-ld] -e [,...] " "\n" "\tallow -c [,...] \n" "\tallow -s @setname [,...] " "\n")); case HELP_UNALLOW: return (gettext("\tunallow [-rldug] " "<\"everyone\"|user|group>[,...]\n" "\t [[,...]] \n" "\tunallow [-rld] -e [[,...]] " "\n" "\tunallow [-r] -c [[,...]] " "\n" "\tunallow [-r] -s @setname [[,...]] " "\n")); case HELP_USERSPACE: return (gettext("\tuserspace [-Hinp] [-o field[,...]] " "[-s field] ...\n" "\t [-S field] ... [-t type[,...]] " "\n")); case HELP_GROUPSPACE: return (gettext("\tgroupspace [-Hinp] [-o field[,...]] " "[-s field] ...\n" "\t [-S field] ... [-t type[,...]] " "\n")); case HELP_PROJECTSPACE: return (gettext("\tprojectspace [-Hp] [-o field[,...]] " "[-s field] ... \n" "\t [-S field] ... \n")); case HELP_PROJECT: return (gettext("\tproject [-d|-r] \n" "\tproject -c [-0] [-d|-r] [-p id] \n" "\tproject -C [-k] [-r] \n" "\tproject [-p id] [-r] [-s] \n")); case HELP_HOLD: return (gettext("\thold [-r] ...\n")); case HELP_HOLDS: return (gettext("\tholds [-rH] ...\n")); case HELP_RELEASE: return (gettext("\trelease [-r] ...\n")); case HELP_DIFF: return (gettext("\tdiff [-FHt] " "[snapshot|filesystem]\n")); case HELP_BOOKMARK: return (gettext("\tbookmark " "\n")); case HELP_CHANNEL_PROGRAM: return (gettext("\tprogram [-jn] [-t ] " "[-m ]\n" "\t [lua args...]\n")); case HELP_LOAD_KEY: return (gettext("\tload-key [-rn] [-L ] " "<-a | filesystem|volume>\n")); case HELP_UNLOAD_KEY: return (gettext("\tunload-key [-r] " "<-a | filesystem|volume>\n")); case HELP_CHANGE_KEY: return (gettext("\tchange-key [-l] [-o keyformat=]\n" "\t [-o keylocation=] [-o pbkdf2iters=]\n" "\t \n" "\tchange-key -i [-l] \n")); case HELP_VERSION: return (gettext("\tversion\n")); case HELP_REDACT: return (gettext("\tredact " " ...\n")); case HELP_JAIL: return (gettext("\tjail \n")); case HELP_UNJAIL: return (gettext("\tunjail \n")); case HELP_WAIT: return (gettext("\twait [-t ] \n")); } abort(); /* NOTREACHED */ } void nomem(void) { (void) fprintf(stderr, gettext("internal error: out of memory\n")); exit(1); } /* * Utility function to guarantee malloc() success. */ void * safe_malloc(size_t size) { void *data; if ((data = calloc(1, size)) == NULL) nomem(); return (data); } static void * safe_realloc(void *data, size_t size) { void *newp; if ((newp = realloc(data, size)) == NULL) { free(data); nomem(); } return (newp); } static char * safe_strdup(char *str) { char *dupstr = strdup(str); if (dupstr == NULL) nomem(); return (dupstr); } /* * Callback routine that will print out information for each of * the properties. */ static int usage_prop_cb(int prop, void *cb) { FILE *fp = cb; (void) fprintf(fp, "\t%-15s ", zfs_prop_to_name(prop)); if (zfs_prop_readonly(prop)) (void) fprintf(fp, " NO "); else (void) fprintf(fp, "YES "); if (zfs_prop_inheritable(prop)) (void) fprintf(fp, " YES "); else (void) fprintf(fp, " NO "); if (zfs_prop_values(prop) == NULL) (void) fprintf(fp, "-\n"); else (void) fprintf(fp, "%s\n", zfs_prop_values(prop)); return (ZPROP_CONT); } /* * Display usage message. If we're inside a command, display only the usage for * that command. Otherwise, iterate over the entire command table and display * a complete usage message. */ static void usage(boolean_t requested) { int i; boolean_t show_properties = B_FALSE; FILE *fp = requested ? stdout : stderr; if (current_command == NULL) { (void) fprintf(fp, gettext("usage: zfs command args ...\n")); (void) fprintf(fp, gettext("where 'command' is one of the following:\n\n")); for (i = 0; i < NCOMMAND; i++) { if (command_table[i].name == NULL) (void) fprintf(fp, "\n"); else (void) fprintf(fp, "%s", get_usage(command_table[i].usage)); } (void) fprintf(fp, gettext("\nEach dataset is of the form: " "pool/[dataset/]*dataset[@name]\n")); } else { (void) fprintf(fp, gettext("usage:\n")); (void) fprintf(fp, "%s", get_usage(current_command->usage)); } if (current_command != NULL && (strcmp(current_command->name, "set") == 0 || strcmp(current_command->name, "get") == 0 || strcmp(current_command->name, "inherit") == 0 || strcmp(current_command->name, "list") == 0)) show_properties = B_TRUE; if (show_properties) { (void) fprintf(fp, gettext("\nThe following properties are supported:\n")); (void) fprintf(fp, "\n\t%-14s %s %s %s\n\n", "PROPERTY", "EDIT", "INHERIT", "VALUES"); /* Iterate over all properties */ (void) zprop_iter(usage_prop_cb, fp, B_FALSE, B_TRUE, ZFS_TYPE_DATASET); (void) fprintf(fp, "\t%-15s ", "userused@..."); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, "\t%-15s ", "groupused@..."); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, "\t%-15s ", "projectused@..."); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, "\t%-15s ", "userobjused@..."); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, "\t%-15s ", "groupobjused@..."); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, "\t%-15s ", "projectobjused@..."); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, "\t%-15s ", "userquota@..."); (void) fprintf(fp, "YES NO | none\n"); (void) fprintf(fp, "\t%-15s ", "groupquota@..."); (void) fprintf(fp, "YES NO | none\n"); (void) fprintf(fp, "\t%-15s ", "projectquota@..."); (void) fprintf(fp, "YES NO | none\n"); (void) fprintf(fp, "\t%-15s ", "userobjquota@..."); (void) fprintf(fp, "YES NO | none\n"); (void) fprintf(fp, "\t%-15s ", "groupobjquota@..."); (void) fprintf(fp, "YES NO | none\n"); (void) fprintf(fp, "\t%-15s ", "projectobjquota@..."); (void) fprintf(fp, "YES NO | none\n"); (void) fprintf(fp, "\t%-15s ", "written@"); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, "\t%-15s ", "written#"); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, gettext("\nSizes are specified in bytes " "with standard units such as K, M, G, etc.\n")); (void) fprintf(fp, gettext("\nUser-defined properties can " "be specified by using a name containing a colon (:).\n")); (void) fprintf(fp, gettext("\nThe {user|group|project}" "[obj]{used|quota}@ properties must be appended with\n" "a user|group|project specifier of one of these forms:\n" " POSIX name (eg: \"matt\")\n" " POSIX id (eg: \"126829\")\n" " SMB name@domain (eg: \"matt@sun\")\n" " SMB SID (eg: \"S-1-234-567-89\")\n")); } else { (void) fprintf(fp, gettext("\nFor the property list, run: %s\n"), "zfs set|get"); (void) fprintf(fp, gettext("\nFor the delegated permission list, run: %s\n"), "zfs allow|unallow"); } /* * See comments at end of main(). */ if (getenv("ZFS_ABORT") != NULL) { (void) printf("dumping core by request\n"); abort(); } exit(requested ? 0 : 2); } /* * Take a property=value argument string and add it to the given nvlist. * Modifies the argument inplace. */ static boolean_t parseprop(nvlist_t *props, char *propname) { char *propval; if ((propval = strchr(propname, '=')) == NULL) { (void) fprintf(stderr, gettext("missing " "'=' for property=value argument\n")); return (B_FALSE); } *propval = '\0'; propval++; if (nvlist_exists(props, propname)) { (void) fprintf(stderr, gettext("property '%s' " "specified multiple times\n"), propname); return (B_FALSE); } if (nvlist_add_string(props, propname, propval) != 0) nomem(); return (B_TRUE); } /* * Take a property name argument and add it to the given nvlist. * Modifies the argument inplace. */ static boolean_t parsepropname(nvlist_t *props, char *propname) { if (strchr(propname, '=') != NULL) { (void) fprintf(stderr, gettext("invalid character " "'=' in property argument\n")); return (B_FALSE); } if (nvlist_exists(props, propname)) { (void) fprintf(stderr, gettext("property '%s' " "specified multiple times\n"), propname); return (B_FALSE); } if (nvlist_add_boolean(props, propname) != 0) nomem(); return (B_TRUE); } static int parse_depth(char *opt, int *flags) { char *tmp; int depth; depth = (int)strtol(opt, &tmp, 0); if (*tmp) { (void) fprintf(stderr, gettext("%s is not an integer\n"), optarg); usage(B_FALSE); } if (depth < 0) { (void) fprintf(stderr, gettext("Depth can not be negative.\n")); usage(B_FALSE); } *flags |= (ZFS_ITER_DEPTH_LIMIT|ZFS_ITER_RECURSE); return (depth); } #define PROGRESS_DELAY 2 /* seconds */ static char *pt_reverse = "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b"; static time_t pt_begin; static char *pt_header = NULL; static boolean_t pt_shown; static void start_progress_timer(void) { pt_begin = time(NULL) + PROGRESS_DELAY; pt_shown = B_FALSE; } static void set_progress_header(char *header) { assert(pt_header == NULL); pt_header = safe_strdup(header); if (pt_shown) { (void) printf("%s: ", header); (void) fflush(stdout); } } static void update_progress(char *update) { if (!pt_shown && time(NULL) > pt_begin) { int len = strlen(update); (void) printf("%s: %s%*.*s", pt_header, update, len, len, pt_reverse); (void) fflush(stdout); pt_shown = B_TRUE; } else if (pt_shown) { int len = strlen(update); (void) printf("%s%*.*s", update, len, len, pt_reverse); (void) fflush(stdout); } } static void finish_progress(char *done) { if (pt_shown) { (void) printf("%s\n", done); (void) fflush(stdout); } free(pt_header); pt_header = NULL; } static int zfs_mount_and_share(libzfs_handle_t *hdl, const char *dataset, zfs_type_t type) { zfs_handle_t *zhp = NULL; int ret = 0; zhp = zfs_open(hdl, dataset, type); if (zhp == NULL) return (1); /* * Volumes may neither be mounted or shared. Potentially in the * future filesystems detected on these volumes could be mounted. */ if (zfs_get_type(zhp) == ZFS_TYPE_VOLUME) { zfs_close(zhp); return (0); } /* * Mount and/or share the new filesystem as appropriate. We provide a * verbose error message to let the user know that their filesystem was * in fact created, even if we failed to mount or share it. * * If the user doesn't want the dataset automatically mounted, then * skip the mount/share step */ if (zfs_prop_valid_for_type(ZFS_PROP_CANMOUNT, type, B_FALSE) && zfs_prop_get_int(zhp, ZFS_PROP_CANMOUNT) == ZFS_CANMOUNT_ON) { if (zfs_mount_delegation_check()) { (void) fprintf(stderr, gettext("filesystem " "successfully created, but it may only be " "mounted by root\n")); ret = 1; } else if (zfs_mount(zhp, NULL, 0) != 0) { (void) fprintf(stderr, gettext("filesystem " "successfully created, but not mounted\n")); ret = 1; } else if (zfs_share(zhp) != 0) { (void) fprintf(stderr, gettext("filesystem " "successfully created, but not shared\n")); ret = 1; } zfs_commit_all_shares(); } zfs_close(zhp); return (ret); } /* * zfs clone [-p] [-o prop=value] ... * * Given an existing dataset, create a writable copy whose initial contents * are the same as the source. The newly created dataset maintains a * dependency on the original; the original cannot be destroyed so long as * the clone exists. * * The '-p' flag creates all the non-existing ancestors of the target first. */ static int zfs_do_clone(int argc, char **argv) { zfs_handle_t *zhp = NULL; boolean_t parents = B_FALSE; nvlist_t *props; int ret = 0; int c; if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) nomem(); /* check options */ while ((c = getopt(argc, argv, "o:p")) != -1) { switch (c) { case 'o': if (!parseprop(props, optarg)) { nvlist_free(props); return (1); } break; case 'p': parents = B_TRUE; break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); goto usage; } } argc -= optind; argv += optind; /* check number of arguments */ if (argc < 1) { (void) fprintf(stderr, gettext("missing source dataset " "argument\n")); goto usage; } if (argc < 2) { (void) fprintf(stderr, gettext("missing target dataset " "argument\n")); goto usage; } if (argc > 2) { (void) fprintf(stderr, gettext("too many arguments\n")); goto usage; } /* open the source dataset */ if ((zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_SNAPSHOT)) == NULL) { nvlist_free(props); return (1); } if (parents && zfs_name_valid(argv[1], ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME)) { /* * Now create the ancestors of the target dataset. If the * target already exists and '-p' option was used we should not * complain. */ if (zfs_dataset_exists(g_zfs, argv[1], ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME)) { zfs_close(zhp); nvlist_free(props); return (0); } if (zfs_create_ancestors(g_zfs, argv[1]) != 0) { zfs_close(zhp); nvlist_free(props); return (1); } } /* pass to libzfs */ ret = zfs_clone(zhp, argv[1], props); /* create the mountpoint if necessary */ if (ret == 0) { if (log_history) { (void) zpool_log_history(g_zfs, history_str); log_history = B_FALSE; } ret = zfs_mount_and_share(g_zfs, argv[1], ZFS_TYPE_DATASET); } zfs_close(zhp); nvlist_free(props); return (!!ret); usage: ASSERT3P(zhp, ==, NULL); nvlist_free(props); usage(B_FALSE); return (-1); } /* * zfs create [-Pnpv] [-o prop=value] ... fs * zfs create [-Pnpsv] [-b blocksize] [-o prop=value] ... -V vol size * * Create a new dataset. This command can be used to create filesystems * and volumes. Snapshot creation is handled by 'zfs snapshot'. * For volumes, the user must specify a size to be used. * * The '-s' flag applies only to volumes, and indicates that we should not try * to set the reservation for this volume. By default we set a reservation * equal to the size for any volume. For pools with SPA_VERSION >= * SPA_VERSION_REFRESERVATION, we set a refreservation instead. * * The '-p' flag creates all the non-existing ancestors of the target first. * * The '-n' flag is no-op (dry run) mode. This will perform a user-space sanity * check of arguments and properties, but does not check for permissions, * available space, etc. * * The '-v' flag is for verbose output. * * The '-P' flag is used for parseable output. It implies '-v'. */ static int zfs_do_create(int argc, char **argv) { zfs_type_t type = ZFS_TYPE_FILESYSTEM; zpool_handle_t *zpool_handle = NULL; nvlist_t *real_props = NULL; uint64_t volsize = 0; int c; boolean_t noreserve = B_FALSE; boolean_t bflag = B_FALSE; boolean_t parents = B_FALSE; boolean_t dryrun = B_FALSE; boolean_t verbose = B_FALSE; boolean_t parseable = B_FALSE; int ret = 1; nvlist_t *props; uint64_t intval; if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) nomem(); /* check options */ while ((c = getopt(argc, argv, ":PV:b:nso:pv")) != -1) { switch (c) { case 'V': type = ZFS_TYPE_VOLUME; if (zfs_nicestrtonum(g_zfs, optarg, &intval) != 0) { (void) fprintf(stderr, gettext("bad volume " "size '%s': %s\n"), optarg, libzfs_error_description(g_zfs)); goto error; } if (nvlist_add_uint64(props, zfs_prop_to_name(ZFS_PROP_VOLSIZE), intval) != 0) nomem(); volsize = intval; break; case 'P': verbose = B_TRUE; parseable = B_TRUE; break; case 'p': parents = B_TRUE; break; case 'b': bflag = B_TRUE; if (zfs_nicestrtonum(g_zfs, optarg, &intval) != 0) { (void) fprintf(stderr, gettext("bad volume " "block size '%s': %s\n"), optarg, libzfs_error_description(g_zfs)); goto error; } if (nvlist_add_uint64(props, zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE), intval) != 0) nomem(); break; case 'n': dryrun = B_TRUE; break; case 'o': if (!parseprop(props, optarg)) goto error; break; case 's': noreserve = B_TRUE; break; case 'v': verbose = B_TRUE; break; case ':': (void) fprintf(stderr, gettext("missing size " "argument\n")); goto badusage; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); goto badusage; } } if ((bflag || noreserve) && type != ZFS_TYPE_VOLUME) { (void) fprintf(stderr, gettext("'-s' and '-b' can only be " "used when creating a volume\n")); goto badusage; } argc -= optind; argv += optind; /* check number of arguments */ if (argc == 0) { (void) fprintf(stderr, gettext("missing %s argument\n"), zfs_type_to_name(type)); goto badusage; } if (argc > 1) { (void) fprintf(stderr, gettext("too many arguments\n")); goto badusage; } if (dryrun || (type == ZFS_TYPE_VOLUME && !noreserve)) { char msg[ZFS_MAX_DATASET_NAME_LEN * 2]; char *p; if ((p = strchr(argv[0], '/')) != NULL) *p = '\0'; zpool_handle = zpool_open(g_zfs, argv[0]); if (p != NULL) *p = '/'; if (zpool_handle == NULL) goto error; (void) snprintf(msg, sizeof (msg), dryrun ? gettext("cannot verify '%s'") : gettext("cannot create '%s'"), argv[0]); if (props && (real_props = zfs_valid_proplist(g_zfs, type, props, 0, NULL, zpool_handle, B_TRUE, msg)) == NULL) { zpool_close(zpool_handle); goto error; } } /* * if volsize is not a multiple of volblocksize, round it up to the * nearest multiple of the volblocksize */ if (type == ZFS_TYPE_VOLUME) { uint64_t volblocksize; if (nvlist_lookup_uint64(props, zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE), &volblocksize) != 0) volblocksize = ZVOL_DEFAULT_BLOCKSIZE; if (volsize % volblocksize) { volsize = P2ROUNDUP_TYPED(volsize, volblocksize, uint64_t); if (nvlist_add_uint64(props, zfs_prop_to_name(ZFS_PROP_VOLSIZE), volsize) != 0) { nvlist_free(props); nomem(); } } } if (type == ZFS_TYPE_VOLUME && !noreserve) { uint64_t spa_version; zfs_prop_t resv_prop; char *strval; spa_version = zpool_get_prop_int(zpool_handle, ZPOOL_PROP_VERSION, NULL); if (spa_version >= SPA_VERSION_REFRESERVATION) resv_prop = ZFS_PROP_REFRESERVATION; else resv_prop = ZFS_PROP_RESERVATION; volsize = zvol_volsize_to_reservation(zpool_handle, volsize, real_props); if (nvlist_lookup_string(props, zfs_prop_to_name(resv_prop), &strval) != 0) { if (nvlist_add_uint64(props, zfs_prop_to_name(resv_prop), volsize) != 0) { nvlist_free(props); nomem(); } } } if (zpool_handle != NULL) { zpool_close(zpool_handle); nvlist_free(real_props); } if (parents && zfs_name_valid(argv[0], type)) { /* * Now create the ancestors of target dataset. If the target * already exists and '-p' option was used we should not * complain. */ if (zfs_dataset_exists(g_zfs, argv[0], type)) { ret = 0; goto error; } if (verbose) { (void) printf(parseable ? "create_ancestors\t%s\n" : dryrun ? "would create ancestors of %s\n" : "create ancestors of %s\n", argv[0]); } if (!dryrun) { if (zfs_create_ancestors(g_zfs, argv[0]) != 0) { goto error; } } } if (verbose) { nvpair_t *nvp = NULL; (void) printf(parseable ? "create\t%s\n" : dryrun ? "would create %s\n" : "create %s\n", argv[0]); while ((nvp = nvlist_next_nvpair(props, nvp)) != NULL) { uint64_t uval; char *sval; switch (nvpair_type(nvp)) { case DATA_TYPE_UINT64: VERIFY0(nvpair_value_uint64(nvp, &uval)); (void) printf(parseable ? "property\t%s\t%llu\n" : "\t%s=%llu\n", nvpair_name(nvp), (u_longlong_t)uval); break; case DATA_TYPE_STRING: VERIFY0(nvpair_value_string(nvp, &sval)); (void) printf(parseable ? "property\t%s\t%s\n" : "\t%s=%s\n", nvpair_name(nvp), sval); break; default: (void) fprintf(stderr, "property '%s' " "has illegal type %d\n", nvpair_name(nvp), nvpair_type(nvp)); abort(); } } } if (dryrun) { ret = 0; goto error; } /* pass to libzfs */ if (zfs_create(g_zfs, argv[0], type, props) != 0) goto error; if (log_history) { (void) zpool_log_history(g_zfs, history_str); log_history = B_FALSE; } ret = zfs_mount_and_share(g_zfs, argv[0], ZFS_TYPE_DATASET); error: nvlist_free(props); return (ret); badusage: nvlist_free(props); usage(B_FALSE); return (2); } /* * zfs destroy [-rRf] * zfs destroy [-rRd] * * -r Recursively destroy all children * -R Recursively destroy all dependents, including clones * -f Force unmounting of any dependents * -d If we can't destroy now, mark for deferred destruction * * Destroys the given dataset. By default, it will unmount any filesystems, * and refuse to destroy a dataset that has any dependents. A dependent can * either be a child, or a clone of a child. */ typedef struct destroy_cbdata { boolean_t cb_first; boolean_t cb_force; boolean_t cb_recurse; boolean_t cb_error; boolean_t cb_doclones; zfs_handle_t *cb_target; boolean_t cb_defer_destroy; boolean_t cb_verbose; boolean_t cb_parsable; boolean_t cb_dryrun; nvlist_t *cb_nvl; nvlist_t *cb_batchedsnaps; /* first snap in contiguous run */ char *cb_firstsnap; /* previous snap in contiguous run */ char *cb_prevsnap; int64_t cb_snapused; char *cb_snapspec; char *cb_bookmark; uint64_t cb_snap_count; } destroy_cbdata_t; /* * Check for any dependents based on the '-r' or '-R' flags. */ static int destroy_check_dependent(zfs_handle_t *zhp, void *data) { destroy_cbdata_t *cbp = data; const char *tname = zfs_get_name(cbp->cb_target); const char *name = zfs_get_name(zhp); if (strncmp(tname, name, strlen(tname)) == 0 && (name[strlen(tname)] == '/' || name[strlen(tname)] == '@')) { /* * This is a direct descendant, not a clone somewhere else in * the hierarchy. */ if (cbp->cb_recurse) goto out; if (cbp->cb_first) { (void) fprintf(stderr, gettext("cannot destroy '%s': " "%s has children\n"), zfs_get_name(cbp->cb_target), zfs_type_to_name(zfs_get_type(cbp->cb_target))); (void) fprintf(stderr, gettext("use '-r' to destroy " "the following datasets:\n")); cbp->cb_first = B_FALSE; cbp->cb_error = B_TRUE; } (void) fprintf(stderr, "%s\n", zfs_get_name(zhp)); } else { /* * This is a clone. We only want to report this if the '-r' * wasn't specified, or the target is a snapshot. */ if (!cbp->cb_recurse && zfs_get_type(cbp->cb_target) != ZFS_TYPE_SNAPSHOT) goto out; if (cbp->cb_first) { (void) fprintf(stderr, gettext("cannot destroy '%s': " "%s has dependent clones\n"), zfs_get_name(cbp->cb_target), zfs_type_to_name(zfs_get_type(cbp->cb_target))); (void) fprintf(stderr, gettext("use '-R' to destroy " "the following datasets:\n")); cbp->cb_first = B_FALSE; cbp->cb_error = B_TRUE; cbp->cb_dryrun = B_TRUE; } (void) fprintf(stderr, "%s\n", zfs_get_name(zhp)); } out: zfs_close(zhp); return (0); } static int destroy_batched(destroy_cbdata_t *cb) { int error = zfs_destroy_snaps_nvl(g_zfs, cb->cb_batchedsnaps, B_FALSE); fnvlist_free(cb->cb_batchedsnaps); cb->cb_batchedsnaps = fnvlist_alloc(); return (error); } static int destroy_callback(zfs_handle_t *zhp, void *data) { destroy_cbdata_t *cb = data; const char *name = zfs_get_name(zhp); int error; if (cb->cb_verbose) { if (cb->cb_parsable) { (void) printf("destroy\t%s\n", name); } else if (cb->cb_dryrun) { (void) printf(gettext("would destroy %s\n"), name); } else { (void) printf(gettext("will destroy %s\n"), name); } } /* * Ignore pools (which we've already flagged as an error before getting * here). */ if (strchr(zfs_get_name(zhp), '/') == NULL && zfs_get_type(zhp) == ZFS_TYPE_FILESYSTEM) { zfs_close(zhp); return (0); } if (cb->cb_dryrun) { zfs_close(zhp); return (0); } /* * We batch up all contiguous snapshots (even of different * filesystems) and destroy them with one ioctl. We can't * simply do all snap deletions and then all fs deletions, * because we must delete a clone before its origin. */ if (zfs_get_type(zhp) == ZFS_TYPE_SNAPSHOT) { cb->cb_snap_count++; fnvlist_add_boolean(cb->cb_batchedsnaps, name); if (cb->cb_snap_count % 10 == 0 && cb->cb_defer_destroy) error = destroy_batched(cb); } else { error = destroy_batched(cb); if (error != 0 || zfs_unmount(zhp, NULL, cb->cb_force ? MS_FORCE : 0) != 0 || zfs_destroy(zhp, cb->cb_defer_destroy) != 0) { zfs_close(zhp); /* * When performing a recursive destroy we ignore errors * so that the recursive destroy could continue * destroying past problem datasets */ if (cb->cb_recurse) { cb->cb_error = B_TRUE; return (0); } return (-1); } } zfs_close(zhp); return (0); } static int destroy_print_cb(zfs_handle_t *zhp, void *arg) { destroy_cbdata_t *cb = arg; const char *name = zfs_get_name(zhp); int err = 0; if (nvlist_exists(cb->cb_nvl, name)) { if (cb->cb_firstsnap == NULL) cb->cb_firstsnap = strdup(name); if (cb->cb_prevsnap != NULL) free(cb->cb_prevsnap); /* this snap continues the current range */ 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); } else if (cb->cb_dryrun) { (void) printf(gettext("would destroy %s\n"), name); } else { (void) printf(gettext("will destroy %s\n"), name); } } } else if (cb->cb_firstsnap != NULL) { /* end of this range */ uint64_t used = 0; err = lzc_snaprange_space(cb->cb_firstsnap, cb->cb_prevsnap, &used); cb->cb_snapused += used; free(cb->cb_firstsnap); cb->cb_firstsnap = NULL; free(cb->cb_prevsnap); cb->cb_prevsnap = NULL; } zfs_close(zhp); return (err); } static int destroy_print_snapshots(zfs_handle_t *fs_zhp, destroy_cbdata_t *cb) { int err; assert(cb->cb_firstsnap == NULL); assert(cb->cb_prevsnap == NULL); err = zfs_iter_snapshots_sorted(fs_zhp, destroy_print_cb, cb, 0, 0); if (cb->cb_firstsnap != NULL) { uint64_t used = 0; if (err == 0) { err = lzc_snaprange_space(cb->cb_firstsnap, cb->cb_prevsnap, &used); } cb->cb_snapused += used; free(cb->cb_firstsnap); cb->cb_firstsnap = NULL; free(cb->cb_prevsnap); cb->cb_prevsnap = NULL; } return (err); } static int snapshot_to_nvl_cb(zfs_handle_t *zhp, void *arg) { destroy_cbdata_t *cb = arg; int err = 0; /* Check for clones. */ if (!cb->cb_doclones && !cb->cb_defer_destroy) { cb->cb_target = zhp; cb->cb_first = B_TRUE; err = zfs_iter_dependents(zhp, B_TRUE, destroy_check_dependent, cb); } if (err == 0) { if (nvlist_add_boolean(cb->cb_nvl, zfs_get_name(zhp))) nomem(); } zfs_close(zhp); return (err); } static int gather_snapshots(zfs_handle_t *zhp, void *arg) { destroy_cbdata_t *cb = arg; int err = 0; err = zfs_iter_snapspec(zhp, cb->cb_snapspec, snapshot_to_nvl_cb, cb); if (err == ENOENT) err = 0; if (err != 0) goto out; if (cb->cb_verbose) { err = destroy_print_snapshots(zhp, cb); if (err != 0) goto out; } if (cb->cb_recurse) err = zfs_iter_filesystems(zhp, gather_snapshots, cb); out: zfs_close(zhp); return (err); } static int destroy_clones(destroy_cbdata_t *cb) { nvpair_t *pair; for (pair = nvlist_next_nvpair(cb->cb_nvl, NULL); pair != NULL; pair = nvlist_next_nvpair(cb->cb_nvl, pair)) { zfs_handle_t *zhp = zfs_open(g_zfs, nvpair_name(pair), ZFS_TYPE_SNAPSHOT); if (zhp != NULL) { boolean_t defer = cb->cb_defer_destroy; int err; /* * We can't defer destroy non-snapshots, so set it to * false while destroying the clones. */ cb->cb_defer_destroy = B_FALSE; err = zfs_iter_dependents(zhp, B_FALSE, destroy_callback, cb); cb->cb_defer_destroy = defer; zfs_close(zhp); if (err != 0) return (err); } } return (0); } static int zfs_do_destroy(int argc, char **argv) { destroy_cbdata_t cb = { 0 }; int rv = 0; int err = 0; int c; zfs_handle_t *zhp = NULL; char *at, *pound; zfs_type_t type = ZFS_TYPE_DATASET; /* check options */ while ((c = getopt(argc, argv, "vpndfrR")) != -1) { switch (c) { case 'v': cb.cb_verbose = B_TRUE; break; case 'p': cb.cb_verbose = B_TRUE; cb.cb_parsable = B_TRUE; break; case 'n': cb.cb_dryrun = B_TRUE; break; case 'd': cb.cb_defer_destroy = B_TRUE; type = ZFS_TYPE_SNAPSHOT; break; case 'f': cb.cb_force = B_TRUE; break; case 'r': cb.cb_recurse = B_TRUE; break; case 'R': cb.cb_recurse = B_TRUE; cb.cb_doclones = B_TRUE; break; case '?': default: (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; /* check number of arguments */ if (argc == 0) { (void) fprintf(stderr, gettext("missing dataset argument\n")); usage(B_FALSE); } if (argc > 1) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } at = strchr(argv[0], '@'); pound = strchr(argv[0], '#'); if (at != NULL) { /* Build the list of snaps to destroy in cb_nvl. */ cb.cb_nvl = fnvlist_alloc(); *at = '\0'; zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME); if (zhp == NULL) { nvlist_free(cb.cb_nvl); return (1); } cb.cb_snapspec = at + 1; if (gather_snapshots(zfs_handle_dup(zhp), &cb) != 0 || cb.cb_error) { rv = 1; goto out; } if (nvlist_empty(cb.cb_nvl)) { (void) fprintf(stderr, gettext("could not find any " "snapshots to destroy; check snapshot names.\n")); rv = 1; goto out; } if (cb.cb_verbose) { char buf[16]; zfs_nicebytes(cb.cb_snapused, buf, sizeof (buf)); if (cb.cb_parsable) { (void) printf("reclaim\t%llu\n", (u_longlong_t)cb.cb_snapused); } else if (cb.cb_dryrun) { (void) printf(gettext("would reclaim %s\n"), buf); } else { (void) printf(gettext("will reclaim %s\n"), buf); } } if (!cb.cb_dryrun) { if (cb.cb_doclones) { cb.cb_batchedsnaps = fnvlist_alloc(); err = destroy_clones(&cb); if (err == 0) { err = zfs_destroy_snaps_nvl(g_zfs, cb.cb_batchedsnaps, B_FALSE); } if (err != 0) { rv = 1; goto out; } } if (err == 0) { err = zfs_destroy_snaps_nvl(g_zfs, cb.cb_nvl, cb.cb_defer_destroy); } } if (err != 0) rv = 1; } else if (pound != NULL) { int err; nvlist_t *nvl; if (cb.cb_dryrun) { (void) fprintf(stderr, "dryrun is not supported with bookmark\n"); return (-1); } if (cb.cb_defer_destroy) { (void) fprintf(stderr, "defer destroy is not supported with bookmark\n"); return (-1); } if (cb.cb_recurse) { (void) fprintf(stderr, "recursive is not supported with bookmark\n"); return (-1); } /* * Unfortunately, zfs_bookmark() doesn't honor the * casesensitivity setting. However, we can't simply * remove this check, because lzc_destroy_bookmarks() * ignores non-existent bookmarks, so this is necessary * to get a proper error message. */ if (!zfs_bookmark_exists(argv[0])) { (void) fprintf(stderr, gettext("bookmark '%s' " "does not exist.\n"), argv[0]); return (1); } nvl = fnvlist_alloc(); fnvlist_add_boolean(nvl, argv[0]); err = lzc_destroy_bookmarks(nvl, NULL); if (err != 0) { (void) zfs_standard_error(g_zfs, err, "cannot destroy bookmark"); } nvlist_free(nvl); return (err); } else { /* Open the given dataset */ if ((zhp = zfs_open(g_zfs, argv[0], type)) == NULL) return (1); cb.cb_target = zhp; /* * Perform an explicit check for pools before going any further. */ if (!cb.cb_recurse && strchr(zfs_get_name(zhp), '/') == NULL && zfs_get_type(zhp) == ZFS_TYPE_FILESYSTEM) { (void) fprintf(stderr, gettext("cannot destroy '%s': " "operation does not apply to pools\n"), zfs_get_name(zhp)); (void) fprintf(stderr, gettext("use 'zfs destroy -r " "%s' to destroy all datasets in the pool\n"), zfs_get_name(zhp)); (void) fprintf(stderr, gettext("use 'zpool destroy %s' " "to destroy the pool itself\n"), zfs_get_name(zhp)); rv = 1; goto out; } /* * Check for any dependents and/or clones. */ cb.cb_first = B_TRUE; if (!cb.cb_doclones && zfs_iter_dependents(zhp, B_TRUE, destroy_check_dependent, &cb) != 0) { rv = 1; goto out; } if (cb.cb_error) { rv = 1; goto out; } cb.cb_batchedsnaps = fnvlist_alloc(); if (zfs_iter_dependents(zhp, B_FALSE, destroy_callback, &cb) != 0) { rv = 1; goto out; } /* * Do the real thing. The callback will close the * handle regardless of whether it succeeds or not. */ err = destroy_callback(zhp, &cb); zhp = NULL; if (err == 0) { err = zfs_destroy_snaps_nvl(g_zfs, cb.cb_batchedsnaps, cb.cb_defer_destroy); } if (err != 0 || cb.cb_error == B_TRUE) rv = 1; } out: fnvlist_free(cb.cb_batchedsnaps); fnvlist_free(cb.cb_nvl); if (zhp != NULL) zfs_close(zhp); return (rv); } static boolean_t is_recvd_column(zprop_get_cbdata_t *cbp) { int i; zfs_get_column_t col; for (i = 0; i < ZFS_GET_NCOLS && (col = cbp->cb_columns[i]) != GET_COL_NONE; i++) if (col == GET_COL_RECVD) return (B_TRUE); return (B_FALSE); } /* * zfs get [-rHp] [-o all | field[,field]...] [-s source[,source]...] * < all | property[,property]... > < fs | snap | vol > ... * * -r recurse over any child datasets * -H scripted mode. Headers are stripped, and fields are separated * by tabs instead of spaces. * -o Set of fields to display. One of "name,property,value, * received,source". Default is "name,property,value,source". * "all" is an alias for all five. * -s Set of sources to allow. One of * "local,default,inherited,received,temporary,none". Default is * all six. * -p Display values in parsable (literal) format. * * Prints properties for the given datasets. The user can control which * columns to display as well as which property types to allow. */ /* * Invoked to display the properties for a single dataset. */ static int get_callback(zfs_handle_t *zhp, void *data) { char buf[ZFS_MAXPROPLEN]; char rbuf[ZFS_MAXPROPLEN]; zprop_source_t sourcetype; char source[ZFS_MAX_DATASET_NAME_LEN]; zprop_get_cbdata_t *cbp = data; nvlist_t *user_props = zfs_get_user_props(zhp); zprop_list_t *pl = cbp->cb_proplist; nvlist_t *propval; char *strval; char *sourceval; boolean_t received = is_recvd_column(cbp); for (; pl != NULL; pl = pl->pl_next) { char *recvdval = NULL; /* * Skip the special fake placeholder. This will also skip over * the name property when 'all' is specified. */ if (pl->pl_prop == ZFS_PROP_NAME && pl == cbp->cb_proplist) continue; if (pl->pl_prop != ZPROP_INVAL) { if (zfs_prop_get(zhp, pl->pl_prop, buf, sizeof (buf), &sourcetype, source, sizeof (source), cbp->cb_literal) != 0) { if (pl->pl_all) continue; if (!zfs_prop_valid_for_type(pl->pl_prop, ZFS_TYPE_DATASET, B_FALSE)) { (void) fprintf(stderr, gettext("No such property '%s'\n"), zfs_prop_to_name(pl->pl_prop)); continue; } sourcetype = ZPROP_SRC_NONE; (void) strlcpy(buf, "-", sizeof (buf)); } if (received && (zfs_prop_get_recvd(zhp, zfs_prop_to_name(pl->pl_prop), rbuf, sizeof (rbuf), cbp->cb_literal) == 0)) recvdval = rbuf; zprop_print_one_property(zfs_get_name(zhp), cbp, zfs_prop_to_name(pl->pl_prop), buf, sourcetype, source, recvdval); } else if (zfs_prop_userquota(pl->pl_user_prop)) { sourcetype = ZPROP_SRC_LOCAL; if (zfs_prop_get_userquota(zhp, pl->pl_user_prop, buf, sizeof (buf), cbp->cb_literal) != 0) { sourcetype = ZPROP_SRC_NONE; (void) strlcpy(buf, "-", sizeof (buf)); } zprop_print_one_property(zfs_get_name(zhp), cbp, pl->pl_user_prop, buf, sourcetype, source, NULL); } else if (zfs_prop_written(pl->pl_user_prop)) { sourcetype = ZPROP_SRC_LOCAL; if (zfs_prop_get_written(zhp, pl->pl_user_prop, buf, sizeof (buf), cbp->cb_literal) != 0) { sourcetype = ZPROP_SRC_NONE; (void) strlcpy(buf, "-", sizeof (buf)); } zprop_print_one_property(zfs_get_name(zhp), cbp, pl->pl_user_prop, buf, sourcetype, source, NULL); } else { if (nvlist_lookup_nvlist(user_props, pl->pl_user_prop, &propval) != 0) { if (pl->pl_all) continue; sourcetype = ZPROP_SRC_NONE; strval = "-"; } else { verify(nvlist_lookup_string(propval, ZPROP_VALUE, &strval) == 0); verify(nvlist_lookup_string(propval, ZPROP_SOURCE, &sourceval) == 0); if (strcmp(sourceval, zfs_get_name(zhp)) == 0) { sourcetype = ZPROP_SRC_LOCAL; } else if (strcmp(sourceval, ZPROP_SOURCE_VAL_RECVD) == 0) { sourcetype = ZPROP_SRC_RECEIVED; } else { sourcetype = ZPROP_SRC_INHERITED; (void) strlcpy(source, sourceval, sizeof (source)); } } if (received && (zfs_prop_get_recvd(zhp, pl->pl_user_prop, rbuf, sizeof (rbuf), cbp->cb_literal) == 0)) recvdval = rbuf; zprop_print_one_property(zfs_get_name(zhp), cbp, pl->pl_user_prop, strval, sourcetype, source, recvdval); } } return (0); } static int zfs_do_get(int argc, char **argv) { zprop_get_cbdata_t cb = { 0 }; int i, c, flags = ZFS_ITER_ARGS_CAN_BE_PATHS; int types = ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK; char *value, *fields; int ret = 0; int limit = 0; zprop_list_t fake_name = { 0 }; /* * Set up default columns and sources. */ cb.cb_sources = ZPROP_SRC_ALL; cb.cb_columns[0] = GET_COL_NAME; cb.cb_columns[1] = GET_COL_PROPERTY; cb.cb_columns[2] = GET_COL_VALUE; cb.cb_columns[3] = GET_COL_SOURCE; cb.cb_type = ZFS_TYPE_DATASET; /* check options */ while ((c = getopt(argc, argv, ":d:o:s:rt:Hp")) != -1) { switch (c) { case 'p': cb.cb_literal = B_TRUE; break; case 'd': limit = parse_depth(optarg, &flags); break; case 'r': flags |= ZFS_ITER_RECURSE; break; case 'H': cb.cb_scripted = B_TRUE; break; case ':': (void) fprintf(stderr, gettext("missing argument for " "'%c' option\n"), optopt); usage(B_FALSE); break; case 'o': /* * Process the set of columns to display. We zero out * the structure to give us a blank slate. */ bzero(&cb.cb_columns, sizeof (cb.cb_columns)); i = 0; while (*optarg != '\0') { static char *col_subopts[] = { "name", "property", "value", "received", "source", "all", NULL }; if (i == ZFS_GET_NCOLS) { (void) fprintf(stderr, gettext("too " "many fields given to -o " "option\n")); usage(B_FALSE); } switch (getsubopt(&optarg, col_subopts, &value)) { case 0: cb.cb_columns[i++] = GET_COL_NAME; break; case 1: cb.cb_columns[i++] = GET_COL_PROPERTY; break; case 2: cb.cb_columns[i++] = GET_COL_VALUE; break; case 3: cb.cb_columns[i++] = GET_COL_RECVD; flags |= ZFS_ITER_RECVD_PROPS; break; case 4: cb.cb_columns[i++] = GET_COL_SOURCE; break; case 5: if (i > 0) { (void) fprintf(stderr, gettext("\"all\" conflicts " "with specific fields " "given to -o option\n")); usage(B_FALSE); } cb.cb_columns[0] = GET_COL_NAME; cb.cb_columns[1] = GET_COL_PROPERTY; cb.cb_columns[2] = GET_COL_VALUE; cb.cb_columns[3] = GET_COL_RECVD; cb.cb_columns[4] = GET_COL_SOURCE; flags |= ZFS_ITER_RECVD_PROPS; i = ZFS_GET_NCOLS; break; default: (void) fprintf(stderr, gettext("invalid column name " "'%s'\n"), value); usage(B_FALSE); } } break; case 's': cb.cb_sources = 0; while (*optarg != '\0') { static char *source_subopts[] = { "local", "default", "inherited", "received", "temporary", "none", NULL }; switch (getsubopt(&optarg, source_subopts, &value)) { case 0: cb.cb_sources |= ZPROP_SRC_LOCAL; break; case 1: cb.cb_sources |= ZPROP_SRC_DEFAULT; break; case 2: cb.cb_sources |= ZPROP_SRC_INHERITED; break; case 3: cb.cb_sources |= ZPROP_SRC_RECEIVED; break; case 4: cb.cb_sources |= ZPROP_SRC_TEMPORARY; break; case 5: cb.cb_sources |= ZPROP_SRC_NONE; break; default: (void) fprintf(stderr, gettext("invalid source " "'%s'\n"), value); usage(B_FALSE); } } break; case 't': types = 0; flags &= ~ZFS_ITER_PROP_LISTSNAPS; while (*optarg != '\0') { static char *type_subopts[] = { "filesystem", "volume", "snapshot", "snap", "bookmark", "all", NULL }; switch (getsubopt(&optarg, type_subopts, &value)) { case 0: types |= ZFS_TYPE_FILESYSTEM; break; case 1: types |= ZFS_TYPE_VOLUME; break; case 2: case 3: types |= ZFS_TYPE_SNAPSHOT; break; case 4: types |= ZFS_TYPE_BOOKMARK; break; case 5: types = ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK; break; default: (void) fprintf(stderr, gettext("invalid type '%s'\n"), value); usage(B_FALSE); } } break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; if (argc < 1) { (void) fprintf(stderr, gettext("missing property " "argument\n")); usage(B_FALSE); } fields = argv[0]; /* * Handle users who want to get all snapshots or bookmarks * of a dataset (ex. 'zfs get -t snapshot refer '). */ if ((types == ZFS_TYPE_SNAPSHOT || types == ZFS_TYPE_BOOKMARK) && argc > 1 && (flags & ZFS_ITER_RECURSE) == 0 && limit == 0) { flags |= (ZFS_ITER_DEPTH_LIMIT | ZFS_ITER_RECURSE); limit = 1; } if (zprop_get_list(g_zfs, fields, &cb.cb_proplist, ZFS_TYPE_DATASET) != 0) usage(B_FALSE); argc--; argv++; /* * As part of zfs_expand_proplist(), we keep track of the maximum column * width for each property. For the 'NAME' (and 'SOURCE') columns, we * need to know the maximum name length. However, the user likely did * not specify 'name' as one of the properties to fetch, so we need to * make sure we always include at least this property for * print_get_headers() to work properly. */ if (cb.cb_proplist != NULL) { fake_name.pl_prop = ZFS_PROP_NAME; fake_name.pl_width = strlen(gettext("NAME")); fake_name.pl_next = cb.cb_proplist; cb.cb_proplist = &fake_name; } cb.cb_first = B_TRUE; /* run for each object */ ret = zfs_for_each(argc, argv, flags, types, NULL, &cb.cb_proplist, limit, get_callback, &cb); if (cb.cb_proplist == &fake_name) zprop_free_list(fake_name.pl_next); else zprop_free_list(cb.cb_proplist); return (ret); } /* * inherit [-rS] ... * * -r Recurse over all children * -S Revert to received value, if any * * For each dataset specified on the command line, inherit the given property * from its parent. Inheriting a property at the pool level will cause it to * use the default value. The '-r' flag will recurse over all children, and is * useful for setting a property on a hierarchy-wide basis, regardless of any * local modifications for each dataset. */ typedef struct inherit_cbdata { const char *cb_propname; boolean_t cb_received; } inherit_cbdata_t; static int inherit_recurse_cb(zfs_handle_t *zhp, void *data) { inherit_cbdata_t *cb = data; zfs_prop_t prop = zfs_name_to_prop(cb->cb_propname); /* * If we're doing it recursively, then ignore properties that * are not valid for this type of dataset. */ if (prop != ZPROP_INVAL && !zfs_prop_valid_for_type(prop, zfs_get_type(zhp), B_FALSE)) return (0); return (zfs_prop_inherit(zhp, cb->cb_propname, cb->cb_received) != 0); } static int inherit_cb(zfs_handle_t *zhp, void *data) { inherit_cbdata_t *cb = data; return (zfs_prop_inherit(zhp, cb->cb_propname, cb->cb_received) != 0); } static int zfs_do_inherit(int argc, char **argv) { int c; zfs_prop_t prop; inherit_cbdata_t cb = { 0 }; char *propname; int ret = 0; int flags = 0; boolean_t received = B_FALSE; /* check options */ while ((c = getopt(argc, argv, "rS")) != -1) { switch (c) { case 'r': flags |= ZFS_ITER_RECURSE; break; case 'S': received = B_TRUE; break; case '?': default: (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; /* check number of arguments */ if (argc < 1) { (void) fprintf(stderr, gettext("missing property argument\n")); usage(B_FALSE); } if (argc < 2) { (void) fprintf(stderr, gettext("missing dataset argument\n")); usage(B_FALSE); } propname = argv[0]; argc--; argv++; if ((prop = zfs_name_to_prop(propname)) != ZPROP_INVAL) { if (zfs_prop_readonly(prop)) { (void) fprintf(stderr, gettext( "%s property is read-only\n"), propname); return (1); } if (!zfs_prop_inheritable(prop) && !received) { (void) fprintf(stderr, gettext("'%s' property cannot " "be inherited\n"), propname); if (prop == ZFS_PROP_QUOTA || prop == ZFS_PROP_RESERVATION || prop == ZFS_PROP_REFQUOTA || prop == ZFS_PROP_REFRESERVATION) { (void) fprintf(stderr, gettext("use 'zfs set " "%s=none' to clear\n"), propname); (void) fprintf(stderr, gettext("use 'zfs " "inherit -S %s' to revert to received " "value\n"), propname); } return (1); } if (received && (prop == ZFS_PROP_VOLSIZE || prop == ZFS_PROP_VERSION)) { (void) fprintf(stderr, gettext("'%s' property cannot " "be reverted to a received value\n"), propname); return (1); } } else if (!zfs_prop_user(propname)) { (void) fprintf(stderr, gettext("invalid property '%s'\n"), propname); usage(B_FALSE); } cb.cb_propname = propname; cb.cb_received = received; if (flags & ZFS_ITER_RECURSE) { ret = zfs_for_each(argc, argv, flags, ZFS_TYPE_DATASET, NULL, NULL, 0, inherit_recurse_cb, &cb); } else { ret = zfs_for_each(argc, argv, flags, ZFS_TYPE_DATASET, NULL, NULL, 0, inherit_cb, &cb); } return (ret); } typedef struct upgrade_cbdata { uint64_t cb_numupgraded; uint64_t cb_numsamegraded; uint64_t cb_numfailed; uint64_t cb_version; boolean_t cb_newer; boolean_t cb_foundone; char cb_lastfs[ZFS_MAX_DATASET_NAME_LEN]; } upgrade_cbdata_t; static int same_pool(zfs_handle_t *zhp, const char *name) { int len1 = strcspn(name, "/@"); const char *zhname = zfs_get_name(zhp); int len2 = strcspn(zhname, "/@"); if (len1 != len2) return (B_FALSE); return (strncmp(name, zhname, len1) == 0); } static int upgrade_list_callback(zfs_handle_t *zhp, void *data) { upgrade_cbdata_t *cb = data; int version = zfs_prop_get_int(zhp, ZFS_PROP_VERSION); /* list if it's old/new */ if ((!cb->cb_newer && version < ZPL_VERSION) || (cb->cb_newer && version > ZPL_VERSION)) { char *str; if (cb->cb_newer) { str = gettext("The following filesystems are " "formatted using a newer software version and\n" "cannot be accessed on the current system.\n\n"); } else { str = gettext("The following filesystems are " "out of date, and can be upgraded. After being\n" "upgraded, these filesystems (and any 'zfs send' " "streams generated from\n" "subsequent snapshots) will no longer be " "accessible by older software versions.\n\n"); } if (!cb->cb_foundone) { (void) puts(str); (void) printf(gettext("VER FILESYSTEM\n")); (void) printf(gettext("--- ------------\n")); cb->cb_foundone = B_TRUE; } (void) printf("%2u %s\n", version, zfs_get_name(zhp)); } return (0); } static int upgrade_set_callback(zfs_handle_t *zhp, void *data) { upgrade_cbdata_t *cb = data; int version = zfs_prop_get_int(zhp, ZFS_PROP_VERSION); int needed_spa_version; int spa_version; if (zfs_spa_version(zhp, &spa_version) < 0) return (-1); needed_spa_version = zfs_spa_version_map(cb->cb_version); if (needed_spa_version < 0) return (-1); if (spa_version < needed_spa_version) { /* can't upgrade */ (void) printf(gettext("%s: can not be " "upgraded; the pool version needs to first " "be upgraded\nto version %d\n\n"), zfs_get_name(zhp), needed_spa_version); cb->cb_numfailed++; return (0); } /* upgrade */ if (version < cb->cb_version) { char verstr[16]; (void) snprintf(verstr, sizeof (verstr), "%llu", (u_longlong_t)cb->cb_version); if (cb->cb_lastfs[0] && !same_pool(zhp, cb->cb_lastfs)) { /* * If they did "zfs upgrade -a", then we could * be doing ioctls to different pools. We need * to log this history once to each pool, and bypass * the normal history logging that happens in main(). */ (void) zpool_log_history(g_zfs, history_str); log_history = B_FALSE; } if (zfs_prop_set(zhp, "version", verstr) == 0) cb->cb_numupgraded++; else cb->cb_numfailed++; (void) strcpy(cb->cb_lastfs, zfs_get_name(zhp)); } else if (version > cb->cb_version) { /* can't downgrade */ (void) printf(gettext("%s: can not be downgraded; " "it is already at version %u\n"), zfs_get_name(zhp), version); cb->cb_numfailed++; } else { cb->cb_numsamegraded++; } return (0); } /* * zfs upgrade * zfs upgrade -v * zfs upgrade [-r] [-V ] <-a | filesystem> */ static int zfs_do_upgrade(int argc, char **argv) { boolean_t all = B_FALSE; boolean_t showversions = B_FALSE; int ret = 0; upgrade_cbdata_t cb = { 0 }; int c; int flags = ZFS_ITER_ARGS_CAN_BE_PATHS; /* check options */ while ((c = getopt(argc, argv, "rvV:a")) != -1) { switch (c) { case 'r': flags |= ZFS_ITER_RECURSE; break; case 'v': showversions = B_TRUE; break; case 'V': if (zfs_prop_string_to_index(ZFS_PROP_VERSION, optarg, &cb.cb_version) != 0) { (void) fprintf(stderr, gettext("invalid version %s\n"), optarg); usage(B_FALSE); } break; case 'a': all = B_TRUE; break; case '?': default: (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; if ((!all && !argc) && ((flags & ZFS_ITER_RECURSE) | cb.cb_version)) usage(B_FALSE); if (showversions && (flags & ZFS_ITER_RECURSE || all || cb.cb_version || argc)) usage(B_FALSE); if ((all || argc) && (showversions)) usage(B_FALSE); if (all && argc) usage(B_FALSE); if (showversions) { /* Show info on available versions. */ (void) printf(gettext("The following filesystem versions are " "supported:\n\n")); (void) printf(gettext("VER DESCRIPTION\n")); (void) printf("--- -----------------------------------------" "---------------\n"); (void) printf(gettext(" 1 Initial ZFS filesystem version\n")); (void) printf(gettext(" 2 Enhanced directory entries\n")); (void) printf(gettext(" 3 Case insensitive and filesystem " "user identifier (FUID)\n")); (void) printf(gettext(" 4 userquota, groupquota " "properties\n")); (void) printf(gettext(" 5 System attributes\n")); (void) printf(gettext("\nFor more information on a particular " "version, including supported releases,\n")); (void) printf("see the ZFS Administration Guide.\n\n"); ret = 0; } else if (argc || all) { /* Upgrade filesystems */ if (cb.cb_version == 0) cb.cb_version = ZPL_VERSION; ret = zfs_for_each(argc, argv, flags, ZFS_TYPE_FILESYSTEM, NULL, NULL, 0, upgrade_set_callback, &cb); (void) printf(gettext("%llu filesystems upgraded\n"), (u_longlong_t)cb.cb_numupgraded); if (cb.cb_numsamegraded) { (void) printf(gettext("%llu filesystems already at " "this version\n"), (u_longlong_t)cb.cb_numsamegraded); } if (cb.cb_numfailed != 0) ret = 1; } else { /* List old-version filesystems */ boolean_t found; (void) printf(gettext("This system is currently running " "ZFS filesystem version %llu.\n\n"), ZPL_VERSION); flags |= ZFS_ITER_RECURSE; ret = zfs_for_each(0, NULL, flags, ZFS_TYPE_FILESYSTEM, NULL, NULL, 0, upgrade_list_callback, &cb); found = cb.cb_foundone; cb.cb_foundone = B_FALSE; cb.cb_newer = B_TRUE; ret = zfs_for_each(0, NULL, flags, ZFS_TYPE_FILESYSTEM, NULL, NULL, 0, upgrade_list_callback, &cb); if (!cb.cb_foundone && !found) { (void) printf(gettext("All filesystems are " "formatted with the current version.\n")); } } return (ret); } /* * zfs userspace [-Hinp] [-o field[,...]] [-s field [-s field]...] * [-S field [-S field]...] [-t type[,...]] * filesystem | snapshot | path * zfs groupspace [-Hinp] [-o field[,...]] [-s field [-s field]...] * [-S field [-S field]...] [-t type[,...]] * filesystem | snapshot | path * zfs projectspace [-Hp] [-o field[,...]] [-s field [-s field]...] * [-S field [-S field]...] filesystem | snapshot | path * * -H Scripted mode; elide headers and separate columns by tabs. * -i Translate SID to POSIX ID. * -n Print numeric ID instead of user/group name. * -o Control which fields to display. * -p Use exact (parsable) numeric output. * -s Specify sort columns, descending order. * -S Specify sort columns, ascending order. * -t Control which object types to display. * * Displays space consumed by, and quotas on, each user in the specified * filesystem or snapshot. */ /* us_field_types, us_field_hdr and us_field_names should be kept in sync */ enum us_field_types { USFIELD_TYPE, USFIELD_NAME, USFIELD_USED, USFIELD_QUOTA, USFIELD_OBJUSED, USFIELD_OBJQUOTA }; static char *us_field_hdr[] = { "TYPE", "NAME", "USED", "QUOTA", "OBJUSED", "OBJQUOTA" }; static char *us_field_names[] = { "type", "name", "used", "quota", "objused", "objquota" }; #define USFIELD_LAST (sizeof (us_field_names) / sizeof (char *)) #define USTYPE_PSX_GRP (1 << 0) #define USTYPE_PSX_USR (1 << 1) #define USTYPE_SMB_GRP (1 << 2) #define USTYPE_SMB_USR (1 << 3) #define USTYPE_PROJ (1 << 4) #define USTYPE_ALL \ (USTYPE_PSX_GRP | USTYPE_PSX_USR | USTYPE_SMB_GRP | USTYPE_SMB_USR | \ USTYPE_PROJ) static int us_type_bits[] = { USTYPE_PSX_GRP, USTYPE_PSX_USR, USTYPE_SMB_GRP, USTYPE_SMB_USR, USTYPE_ALL }; static char *us_type_names[] = { "posixgroup", "posixuser", "smbgroup", "smbuser", "all" }; typedef struct us_node { nvlist_t *usn_nvl; uu_avl_node_t usn_avlnode; uu_list_node_t usn_listnode; } us_node_t; typedef struct us_cbdata { nvlist_t **cb_nvlp; uu_avl_pool_t *cb_avl_pool; uu_avl_t *cb_avl; boolean_t cb_numname; boolean_t cb_nicenum; boolean_t cb_sid2posix; zfs_userquota_prop_t cb_prop; zfs_sort_column_t *cb_sortcol; size_t cb_width[USFIELD_LAST]; } us_cbdata_t; static boolean_t us_populated = B_FALSE; typedef struct { zfs_sort_column_t *si_sortcol; boolean_t si_numname; } us_sort_info_t; static int us_field_index(char *field) { int i; for (i = 0; i < USFIELD_LAST; i++) { if (strcmp(field, us_field_names[i]) == 0) return (i); } return (-1); } static int us_compare(const void *larg, const void *rarg, void *unused) { const us_node_t *l = larg; const us_node_t *r = rarg; us_sort_info_t *si = (us_sort_info_t *)unused; zfs_sort_column_t *sortcol = si->si_sortcol; boolean_t numname = si->si_numname; nvlist_t *lnvl = l->usn_nvl; nvlist_t *rnvl = r->usn_nvl; int rc = 0; boolean_t lvb, rvb; for (; sortcol != NULL; sortcol = sortcol->sc_next) { char *lvstr = ""; char *rvstr = ""; uint32_t lv32 = 0; uint32_t rv32 = 0; uint64_t lv64 = 0; uint64_t rv64 = 0; zfs_prop_t prop = sortcol->sc_prop; const char *propname = NULL; boolean_t reverse = sortcol->sc_reverse; switch (prop) { case ZFS_PROP_TYPE: propname = "type"; (void) nvlist_lookup_uint32(lnvl, propname, &lv32); (void) nvlist_lookup_uint32(rnvl, propname, &rv32); if (rv32 != lv32) rc = (rv32 < lv32) ? 1 : -1; break; case ZFS_PROP_NAME: propname = "name"; if (numname) { compare_nums: (void) nvlist_lookup_uint64(lnvl, propname, &lv64); (void) nvlist_lookup_uint64(rnvl, propname, &rv64); if (rv64 != lv64) rc = (rv64 < lv64) ? 1 : -1; } else { if ((nvlist_lookup_string(lnvl, propname, &lvstr) == ENOENT) || (nvlist_lookup_string(rnvl, propname, &rvstr) == ENOENT)) { goto compare_nums; } rc = strcmp(lvstr, rvstr); } break; case ZFS_PROP_USED: case ZFS_PROP_QUOTA: if (!us_populated) break; if (prop == ZFS_PROP_USED) propname = "used"; else propname = "quota"; (void) nvlist_lookup_uint64(lnvl, propname, &lv64); (void) nvlist_lookup_uint64(rnvl, propname, &rv64); if (rv64 != lv64) rc = (rv64 < lv64) ? 1 : -1; break; default: break; } if (rc != 0) { if (rc < 0) return (reverse ? 1 : -1); else return (reverse ? -1 : 1); } } /* * If entries still seem to be the same, check if they are of the same * type (smbentity is added only if we are doing SID to POSIX ID * translation where we can have duplicate type/name combinations). */ if (nvlist_lookup_boolean_value(lnvl, "smbentity", &lvb) == 0 && nvlist_lookup_boolean_value(rnvl, "smbentity", &rvb) == 0 && lvb != rvb) return (lvb < rvb ? -1 : 1); return (0); } static boolean_t zfs_prop_is_user(unsigned p) { return (p == ZFS_PROP_USERUSED || p == ZFS_PROP_USERQUOTA || p == ZFS_PROP_USEROBJUSED || p == ZFS_PROP_USEROBJQUOTA); } static boolean_t zfs_prop_is_group(unsigned p) { return (p == ZFS_PROP_GROUPUSED || p == ZFS_PROP_GROUPQUOTA || p == ZFS_PROP_GROUPOBJUSED || p == ZFS_PROP_GROUPOBJQUOTA); } static boolean_t zfs_prop_is_project(unsigned p) { return (p == ZFS_PROP_PROJECTUSED || p == ZFS_PROP_PROJECTQUOTA || p == ZFS_PROP_PROJECTOBJUSED || p == ZFS_PROP_PROJECTOBJQUOTA); } static inline const char * us_type2str(unsigned field_type) { switch (field_type) { case USTYPE_PSX_USR: return ("POSIX User"); case USTYPE_PSX_GRP: return ("POSIX Group"); case USTYPE_SMB_USR: return ("SMB User"); case USTYPE_SMB_GRP: return ("SMB Group"); case USTYPE_PROJ: return ("Project"); default: return ("Undefined"); } } static int userspace_cb(void *arg, const char *domain, uid_t rid, uint64_t space) { us_cbdata_t *cb = (us_cbdata_t *)arg; zfs_userquota_prop_t prop = cb->cb_prop; char *name = NULL; char *propname; char sizebuf[32]; us_node_t *node; uu_avl_pool_t *avl_pool = cb->cb_avl_pool; uu_avl_t *avl = cb->cb_avl; uu_avl_index_t idx; nvlist_t *props; us_node_t *n; zfs_sort_column_t *sortcol = cb->cb_sortcol; unsigned type = 0; const char *typestr; size_t namelen; size_t typelen; size_t sizelen; int typeidx, nameidx, sizeidx; us_sort_info_t sortinfo = { sortcol, cb->cb_numname }; boolean_t smbentity = B_FALSE; if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) nomem(); node = safe_malloc(sizeof (us_node_t)); uu_avl_node_init(node, &node->usn_avlnode, avl_pool); node->usn_nvl = props; if (domain != NULL && domain[0] != '\0') { #ifdef HAVE_IDMAP /* SMB */ char sid[MAXNAMELEN + 32]; uid_t id; uint64_t classes; int err; directory_error_t e; smbentity = B_TRUE; (void) snprintf(sid, sizeof (sid), "%s-%u", domain, rid); if (prop == ZFS_PROP_GROUPUSED || prop == ZFS_PROP_GROUPQUOTA) { type = USTYPE_SMB_GRP; err = sid_to_id(sid, B_FALSE, &id); } else { type = USTYPE_SMB_USR; err = sid_to_id(sid, B_TRUE, &id); } if (err == 0) { rid = id; if (!cb->cb_sid2posix) { e = directory_name_from_sid(NULL, sid, &name, &classes); if (e != NULL) directory_error_free(e); if (name == NULL) name = sid; } } #else nvlist_free(props); free(node); return (-1); #endif /* HAVE_IDMAP */ } if (cb->cb_sid2posix || domain == NULL || domain[0] == '\0') { /* POSIX or -i */ if (zfs_prop_is_group(prop)) { type = USTYPE_PSX_GRP; if (!cb->cb_numname) { struct group *g; if ((g = getgrgid(rid)) != NULL) name = g->gr_name; } } else if (zfs_prop_is_user(prop)) { type = USTYPE_PSX_USR; if (!cb->cb_numname) { struct passwd *p; if ((p = getpwuid(rid)) != NULL) name = p->pw_name; } } else { type = USTYPE_PROJ; } } /* * Make sure that the type/name combination is unique when doing * SID to POSIX ID translation (hence changing the type from SMB to * POSIX). */ if (cb->cb_sid2posix && nvlist_add_boolean_value(props, "smbentity", smbentity) != 0) nomem(); /* Calculate/update width of TYPE field */ typestr = us_type2str(type); typelen = strlen(gettext(typestr)); typeidx = us_field_index("type"); if (typelen > cb->cb_width[typeidx]) cb->cb_width[typeidx] = typelen; if (nvlist_add_uint32(props, "type", type) != 0) nomem(); /* Calculate/update width of NAME field */ if ((cb->cb_numname && cb->cb_sid2posix) || name == NULL) { if (nvlist_add_uint64(props, "name", rid) != 0) nomem(); namelen = snprintf(NULL, 0, "%u", rid); } else { if (nvlist_add_string(props, "name", name) != 0) nomem(); namelen = strlen(name); } nameidx = us_field_index("name"); if (nameidx >= 0 && namelen > cb->cb_width[nameidx]) cb->cb_width[nameidx] = namelen; /* * Check if this type/name combination is in the list and update it; * otherwise add new node to the list. */ if ((n = uu_avl_find(avl, node, &sortinfo, &idx)) == NULL) { uu_avl_insert(avl, node, idx); } else { nvlist_free(props); free(node); node = n; props = node->usn_nvl; } /* Calculate/update width of USED/QUOTA fields */ if (cb->cb_nicenum) { if (prop == ZFS_PROP_USERUSED || prop == ZFS_PROP_GROUPUSED || prop == ZFS_PROP_USERQUOTA || prop == ZFS_PROP_GROUPQUOTA || prop == ZFS_PROP_PROJECTUSED || prop == ZFS_PROP_PROJECTQUOTA) { zfs_nicebytes(space, sizebuf, sizeof (sizebuf)); } else { zfs_nicenum(space, sizebuf, sizeof (sizebuf)); } } else { (void) snprintf(sizebuf, sizeof (sizebuf), "%llu", (u_longlong_t)space); } sizelen = strlen(sizebuf); if (prop == ZFS_PROP_USERUSED || prop == ZFS_PROP_GROUPUSED || prop == ZFS_PROP_PROJECTUSED) { propname = "used"; if (!nvlist_exists(props, "quota")) (void) nvlist_add_uint64(props, "quota", 0); } else if (prop == ZFS_PROP_USERQUOTA || prop == ZFS_PROP_GROUPQUOTA || prop == ZFS_PROP_PROJECTQUOTA) { propname = "quota"; if (!nvlist_exists(props, "used")) (void) nvlist_add_uint64(props, "used", 0); } else if (prop == ZFS_PROP_USEROBJUSED || prop == ZFS_PROP_GROUPOBJUSED || prop == ZFS_PROP_PROJECTOBJUSED) { propname = "objused"; if (!nvlist_exists(props, "objquota")) (void) nvlist_add_uint64(props, "objquota", 0); } else if (prop == ZFS_PROP_USEROBJQUOTA || prop == ZFS_PROP_GROUPOBJQUOTA || prop == ZFS_PROP_PROJECTOBJQUOTA) { propname = "objquota"; if (!nvlist_exists(props, "objused")) (void) nvlist_add_uint64(props, "objused", 0); } else { return (-1); } sizeidx = us_field_index(propname); if (sizeidx >= 0 && sizelen > cb->cb_width[sizeidx]) cb->cb_width[sizeidx] = sizelen; if (nvlist_add_uint64(props, propname, space) != 0) nomem(); return (0); } static void print_us_node(boolean_t scripted, boolean_t parsable, int *fields, int types, size_t *width, us_node_t *node) { nvlist_t *nvl = node->usn_nvl; char valstr[MAXNAMELEN]; boolean_t first = B_TRUE; int cfield = 0; int field; uint32_t ustype; /* Check type */ (void) nvlist_lookup_uint32(nvl, "type", &ustype); if (!(ustype & types)) return; while ((field = fields[cfield]) != USFIELD_LAST) { nvpair_t *nvp = NULL; data_type_t type; uint32_t val32; uint64_t val64; char *strval = "-"; while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) { if (strcmp(nvpair_name(nvp), us_field_names[field]) == 0) break; } type = nvp == NULL ? DATA_TYPE_UNKNOWN : nvpair_type(nvp); switch (type) { case DATA_TYPE_UINT32: (void) nvpair_value_uint32(nvp, &val32); break; case DATA_TYPE_UINT64: (void) nvpair_value_uint64(nvp, &val64); break; case DATA_TYPE_STRING: (void) nvpair_value_string(nvp, &strval); break; case DATA_TYPE_UNKNOWN: break; default: (void) fprintf(stderr, "invalid data type\n"); } switch (field) { case USFIELD_TYPE: if (type == DATA_TYPE_UINT32) strval = (char *)us_type2str(val32); break; case USFIELD_NAME: if (type == DATA_TYPE_UINT64) { (void) sprintf(valstr, "%llu", (u_longlong_t)val64); strval = valstr; } break; case USFIELD_USED: case USFIELD_QUOTA: if (type == DATA_TYPE_UINT64) { if (parsable) { (void) sprintf(valstr, "%llu", (u_longlong_t)val64); strval = valstr; } else if (field == USFIELD_QUOTA && val64 == 0) { strval = "none"; } else { zfs_nicebytes(val64, valstr, sizeof (valstr)); strval = valstr; } } break; case USFIELD_OBJUSED: case USFIELD_OBJQUOTA: if (type == DATA_TYPE_UINT64) { if (parsable) { (void) sprintf(valstr, "%llu", (u_longlong_t)val64); strval = valstr; } else if (field == USFIELD_OBJQUOTA && val64 == 0) { strval = "none"; } else { zfs_nicenum(val64, valstr, sizeof (valstr)); strval = valstr; } } break; } if (!first) { if (scripted) (void) printf("\t"); else (void) printf(" "); } if (scripted) (void) printf("%s", strval); else if (field == USFIELD_TYPE || field == USFIELD_NAME) (void) printf("%-*s", (int)width[field], strval); else (void) printf("%*s", (int)width[field], strval); first = B_FALSE; cfield++; } (void) printf("\n"); } static void print_us(boolean_t scripted, boolean_t parsable, int *fields, int types, size_t *width, boolean_t rmnode, uu_avl_t *avl) { us_node_t *node; const char *col; int cfield = 0; int field; if (!scripted) { boolean_t first = B_TRUE; while ((field = fields[cfield]) != USFIELD_LAST) { col = gettext(us_field_hdr[field]); if (field == USFIELD_TYPE || field == USFIELD_NAME) { (void) printf(first ? "%-*s" : " %-*s", (int)width[field], col); } else { (void) printf(first ? "%*s" : " %*s", (int)width[field], col); } first = B_FALSE; cfield++; } (void) printf("\n"); } for (node = uu_avl_first(avl); node; node = uu_avl_next(avl, node)) { print_us_node(scripted, parsable, fields, types, width, node); if (rmnode) nvlist_free(node->usn_nvl); } } static int zfs_do_userspace(int argc, char **argv) { zfs_handle_t *zhp; zfs_userquota_prop_t p; uu_avl_pool_t *avl_pool; uu_avl_t *avl_tree; uu_avl_walk_t *walk; char *delim; char deffields[] = "type,name,used,quota,objused,objquota"; char *ofield = NULL; char *tfield = NULL; int cfield = 0; int fields[256]; int i; boolean_t scripted = B_FALSE; boolean_t prtnum = B_FALSE; boolean_t parsable = B_FALSE; boolean_t sid2posix = B_FALSE; int ret = 0; int c; zfs_sort_column_t *sortcol = NULL; int types = USTYPE_PSX_USR | USTYPE_SMB_USR; us_cbdata_t cb; us_node_t *node; us_node_t *rmnode; uu_list_pool_t *listpool; uu_list_t *list; uu_avl_index_t idx = 0; uu_list_index_t idx2 = 0; if (argc < 2) usage(B_FALSE); if (strcmp(argv[0], "groupspace") == 0) { /* Toggle default group types */ types = USTYPE_PSX_GRP | USTYPE_SMB_GRP; } else if (strcmp(argv[0], "projectspace") == 0) { types = USTYPE_PROJ; prtnum = B_TRUE; } while ((c = getopt(argc, argv, "nHpo:s:S:t:i")) != -1) { switch (c) { case 'n': if (types == USTYPE_PROJ) { (void) fprintf(stderr, gettext("invalid option 'n'\n")); usage(B_FALSE); } prtnum = B_TRUE; break; case 'H': scripted = B_TRUE; break; case 'p': parsable = B_TRUE; break; case 'o': ofield = optarg; break; case 's': case 'S': if (zfs_add_sort_column(&sortcol, optarg, c == 's' ? B_FALSE : B_TRUE) != 0) { (void) fprintf(stderr, gettext("invalid field '%s'\n"), optarg); usage(B_FALSE); } break; case 't': if (types == USTYPE_PROJ) { (void) fprintf(stderr, gettext("invalid option 't'\n")); usage(B_FALSE); } tfield = optarg; break; case 'i': if (types == USTYPE_PROJ) { (void) fprintf(stderr, gettext("invalid option 'i'\n")); usage(B_FALSE); } sid2posix = B_TRUE; break; case ':': (void) fprintf(stderr, gettext("missing argument for " "'%c' option\n"), optopt); usage(B_FALSE); break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; if (argc < 1) { (void) fprintf(stderr, gettext("missing dataset name\n")); usage(B_FALSE); } if (argc > 1) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } /* Use default output fields if not specified using -o */ if (ofield == NULL) ofield = deffields; do { if ((delim = strchr(ofield, ',')) != NULL) *delim = '\0'; if ((fields[cfield++] = us_field_index(ofield)) == -1) { (void) fprintf(stderr, gettext("invalid type '%s' " "for -o option\n"), ofield); return (-1); } if (delim != NULL) ofield = delim + 1; } while (delim != NULL); fields[cfield] = USFIELD_LAST; /* Override output types (-t option) */ if (tfield != NULL) { types = 0; do { boolean_t found = B_FALSE; if ((delim = strchr(tfield, ',')) != NULL) *delim = '\0'; for (i = 0; i < sizeof (us_type_bits) / sizeof (int); i++) { if (strcmp(tfield, us_type_names[i]) == 0) { found = B_TRUE; types |= us_type_bits[i]; break; } } if (!found) { (void) fprintf(stderr, gettext("invalid type " "'%s' for -t option\n"), tfield); return (-1); } if (delim != NULL) tfield = delim + 1; } while (delim != NULL); } if ((zhp = zfs_path_to_zhandle(g_zfs, argv[0], ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT)) == NULL) return (1); if (zhp->zfs_head_type != ZFS_TYPE_FILESYSTEM) { (void) fprintf(stderr, gettext("operation is only applicable " "to filesystems and their snapshots\n")); zfs_close(zhp); return (1); } if ((avl_pool = uu_avl_pool_create("us_avl_pool", sizeof (us_node_t), offsetof(us_node_t, usn_avlnode), us_compare, UU_DEFAULT)) == NULL) nomem(); if ((avl_tree = uu_avl_create(avl_pool, NULL, UU_DEFAULT)) == NULL) nomem(); /* Always add default sorting columns */ (void) zfs_add_sort_column(&sortcol, "type", B_FALSE); (void) zfs_add_sort_column(&sortcol, "name", B_FALSE); cb.cb_sortcol = sortcol; cb.cb_numname = prtnum; cb.cb_nicenum = !parsable; cb.cb_avl_pool = avl_pool; cb.cb_avl = avl_tree; cb.cb_sid2posix = sid2posix; for (i = 0; i < USFIELD_LAST; i++) cb.cb_width[i] = strlen(gettext(us_field_hdr[i])); for (p = 0; p < ZFS_NUM_USERQUOTA_PROPS; p++) { if ((zfs_prop_is_user(p) && !(types & (USTYPE_PSX_USR | USTYPE_SMB_USR))) || (zfs_prop_is_group(p) && !(types & (USTYPE_PSX_GRP | USTYPE_SMB_GRP))) || (zfs_prop_is_project(p) && types != USTYPE_PROJ)) continue; cb.cb_prop = p; if ((ret = zfs_userspace(zhp, p, userspace_cb, &cb)) != 0) { zfs_close(zhp); return (ret); } } zfs_close(zhp); /* Sort the list */ if ((node = uu_avl_first(avl_tree)) == NULL) return (0); us_populated = B_TRUE; listpool = uu_list_pool_create("tmplist", sizeof (us_node_t), offsetof(us_node_t, usn_listnode), NULL, UU_DEFAULT); list = uu_list_create(listpool, NULL, UU_DEFAULT); uu_list_node_init(node, &node->usn_listnode, listpool); while (node != NULL) { rmnode = node; node = uu_avl_next(avl_tree, node); uu_avl_remove(avl_tree, rmnode); if (uu_list_find(list, rmnode, NULL, &idx2) == NULL) uu_list_insert(list, rmnode, idx2); } for (node = uu_list_first(list); node != NULL; node = uu_list_next(list, node)) { us_sort_info_t sortinfo = { sortcol, cb.cb_numname }; if (uu_avl_find(avl_tree, node, &sortinfo, &idx) == NULL) uu_avl_insert(avl_tree, node, idx); } uu_list_destroy(list); uu_list_pool_destroy(listpool); /* Print and free node nvlist memory */ print_us(scripted, parsable, fields, types, cb.cb_width, B_TRUE, cb.cb_avl); zfs_free_sort_columns(sortcol); /* Clean up the AVL tree */ if ((walk = uu_avl_walk_start(cb.cb_avl, UU_WALK_ROBUST)) == NULL) nomem(); while ((node = uu_avl_walk_next(walk)) != NULL) { uu_avl_remove(cb.cb_avl, node); free(node); } uu_avl_walk_end(walk); uu_avl_destroy(avl_tree); uu_avl_pool_destroy(avl_pool); return (ret); } /* * list [-Hp][-r|-d max] [-o property[,...]] [-s property] ... [-S property] * [-t type[,...]] [filesystem|volume|snapshot] ... * * -H Scripted mode; elide headers and separate columns by tabs * -p Display values in parsable (literal) format. * -r Recurse over all children * -d Limit recursion by depth. * -o Control which fields to display. * -s Specify sort columns, descending order. * -S Specify sort columns, ascending order. * -t Control which object types to display. * * When given no arguments, list all filesystems in the system. * Otherwise, list the specified datasets, optionally recursing down them if * '-r' is specified. */ typedef struct list_cbdata { boolean_t cb_first; boolean_t cb_literal; boolean_t cb_scripted; zprop_list_t *cb_proplist; } list_cbdata_t; /* * Given a list of columns to display, output appropriate headers for each one. */ static void print_header(list_cbdata_t *cb) { zprop_list_t *pl = cb->cb_proplist; char headerbuf[ZFS_MAXPROPLEN]; const char *header; int i; boolean_t first = B_TRUE; boolean_t right_justify; for (; pl != NULL; pl = pl->pl_next) { if (!first) { (void) printf(" "); } else { first = B_FALSE; } right_justify = B_FALSE; if (pl->pl_prop != ZPROP_INVAL) { header = zfs_prop_column_name(pl->pl_prop); right_justify = zfs_prop_align_right(pl->pl_prop); } else { for (i = 0; pl->pl_user_prop[i] != '\0'; i++) headerbuf[i] = toupper(pl->pl_user_prop[i]); headerbuf[i] = '\0'; header = headerbuf; } if (pl->pl_next == NULL && !right_justify) (void) printf("%s", header); else if (right_justify) (void) printf("%*s", (int)pl->pl_width, header); else (void) printf("%-*s", (int)pl->pl_width, header); } (void) printf("\n"); } /* * Given a dataset and a list of fields, print out all the properties according * to the described layout. */ static void print_dataset(zfs_handle_t *zhp, list_cbdata_t *cb) { zprop_list_t *pl = cb->cb_proplist; boolean_t first = B_TRUE; char property[ZFS_MAXPROPLEN]; nvlist_t *userprops = zfs_get_user_props(zhp); nvlist_t *propval; char *propstr; boolean_t right_justify; for (; pl != NULL; pl = pl->pl_next) { if (!first) { if (cb->cb_scripted) (void) printf("\t"); else (void) printf(" "); } else { first = B_FALSE; } if (pl->pl_prop == ZFS_PROP_NAME) { (void) strlcpy(property, zfs_get_name(zhp), sizeof (property)); propstr = property; right_justify = zfs_prop_align_right(pl->pl_prop); } else if (pl->pl_prop != ZPROP_INVAL) { if (zfs_prop_get(zhp, pl->pl_prop, property, sizeof (property), NULL, NULL, 0, cb->cb_literal) != 0) propstr = "-"; else propstr = property; right_justify = zfs_prop_align_right(pl->pl_prop); } else if (zfs_prop_userquota(pl->pl_user_prop)) { if (zfs_prop_get_userquota(zhp, pl->pl_user_prop, property, sizeof (property), cb->cb_literal) != 0) propstr = "-"; else propstr = property; right_justify = B_TRUE; } else if (zfs_prop_written(pl->pl_user_prop)) { if (zfs_prop_get_written(zhp, pl->pl_user_prop, property, sizeof (property), cb->cb_literal) != 0) propstr = "-"; else propstr = property; right_justify = B_TRUE; } else { if (nvlist_lookup_nvlist(userprops, pl->pl_user_prop, &propval) != 0) propstr = "-"; else verify(nvlist_lookup_string(propval, ZPROP_VALUE, &propstr) == 0); right_justify = B_FALSE; } /* * If this is being called in scripted mode, or if this is the * last column and it is left-justified, don't include a width * format specifier. */ if (cb->cb_scripted || (pl->pl_next == NULL && !right_justify)) (void) printf("%s", propstr); else if (right_justify) (void) printf("%*s", (int)pl->pl_width, propstr); else (void) printf("%-*s", (int)pl->pl_width, propstr); } (void) printf("\n"); } /* * Generic callback function to list a dataset or snapshot. */ static int list_callback(zfs_handle_t *zhp, void *data) { list_cbdata_t *cbp = data; if (cbp->cb_first) { if (!cbp->cb_scripted) print_header(cbp); cbp->cb_first = B_FALSE; } print_dataset(zhp, cbp); return (0); } static int zfs_do_list(int argc, char **argv) { int c; static char default_fields[] = "name,used,available,referenced,mountpoint"; int types = ZFS_TYPE_DATASET; boolean_t types_specified = B_FALSE; char *fields = NULL; list_cbdata_t cb = { 0 }; char *value; int limit = 0; int ret = 0; zfs_sort_column_t *sortcol = NULL; int flags = ZFS_ITER_PROP_LISTSNAPS | ZFS_ITER_ARGS_CAN_BE_PATHS; /* check options */ while ((c = getopt(argc, argv, "HS:d:o:prs:t:")) != -1) { switch (c) { case 'o': fields = optarg; break; case 'p': cb.cb_literal = B_TRUE; flags |= ZFS_ITER_LITERAL_PROPS; break; case 'd': limit = parse_depth(optarg, &flags); break; case 'r': flags |= ZFS_ITER_RECURSE; break; case 'H': cb.cb_scripted = B_TRUE; break; case 's': if (zfs_add_sort_column(&sortcol, optarg, B_FALSE) != 0) { (void) fprintf(stderr, gettext("invalid property '%s'\n"), optarg); usage(B_FALSE); } break; case 'S': if (zfs_add_sort_column(&sortcol, optarg, B_TRUE) != 0) { (void) fprintf(stderr, gettext("invalid property '%s'\n"), optarg); usage(B_FALSE); } break; case 't': types = 0; types_specified = B_TRUE; flags &= ~ZFS_ITER_PROP_LISTSNAPS; while (*optarg != '\0') { static char *type_subopts[] = { "filesystem", "volume", "snapshot", "snap", "bookmark", "all", NULL }; switch (getsubopt(&optarg, type_subopts, &value)) { case 0: types |= ZFS_TYPE_FILESYSTEM; break; case 1: types |= ZFS_TYPE_VOLUME; break; case 2: case 3: types |= ZFS_TYPE_SNAPSHOT; break; case 4: types |= ZFS_TYPE_BOOKMARK; break; case 5: types = ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK; break; default: (void) fprintf(stderr, gettext("invalid type '%s'\n"), value); usage(B_FALSE); } } break; case ':': (void) fprintf(stderr, gettext("missing argument for " "'%c' option\n"), optopt); usage(B_FALSE); break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; if (fields == NULL) fields = default_fields; /* * If we are only going to list snapshot names and sort by name, * then we can use faster version. */ if (strcmp(fields, "name") == 0 && zfs_sort_only_by_name(sortcol)) flags |= ZFS_ITER_SIMPLE; /* * If "-o space" and no types were specified, don't display snapshots. */ if (strcmp(fields, "space") == 0 && types_specified == B_FALSE) types &= ~ZFS_TYPE_SNAPSHOT; /* * Handle users who want to list all snapshots or bookmarks * of the current dataset (ex. 'zfs list -t snapshot '). */ if ((types == ZFS_TYPE_SNAPSHOT || types == ZFS_TYPE_BOOKMARK) && argc > 0 && (flags & ZFS_ITER_RECURSE) == 0 && limit == 0) { flags |= (ZFS_ITER_DEPTH_LIMIT | ZFS_ITER_RECURSE); limit = 1; } /* * If the user specifies '-o all', the zprop_get_list() doesn't * normally include the name of the dataset. For 'zfs list', we always * want this property to be first. */ if (zprop_get_list(g_zfs, fields, &cb.cb_proplist, ZFS_TYPE_DATASET) != 0) usage(B_FALSE); cb.cb_first = B_TRUE; ret = zfs_for_each(argc, argv, flags, types, sortcol, &cb.cb_proplist, limit, list_callback, &cb); zprop_free_list(cb.cb_proplist); zfs_free_sort_columns(sortcol); if (ret == 0 && cb.cb_first && !cb.cb_scripted) (void) fprintf(stderr, gettext("no datasets available\n")); return (ret); } /* * zfs rename [-fu] * zfs rename [-f] -p * zfs rename [-u] -r * * Renames the given dataset to another of the same type. * * The '-p' flag creates all the non-existing ancestors of the target first. * The '-u' flag prevents file systems from being remounted during rename. */ /* ARGSUSED */ static int zfs_do_rename(int argc, char **argv) { zfs_handle_t *zhp; renameflags_t flags = { 0 }; int c; int ret = 0; int types; boolean_t parents = B_FALSE; /* check options */ while ((c = getopt(argc, argv, "pruf")) != -1) { switch (c) { case 'p': parents = B_TRUE; break; case 'r': flags.recursive = B_TRUE; break; case 'u': flags.nounmount = B_TRUE; break; case 'f': flags.forceunmount = B_TRUE; break; case '?': default: (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; /* check number of arguments */ if (argc < 1) { (void) fprintf(stderr, gettext("missing source dataset " "argument\n")); usage(B_FALSE); } if (argc < 2) { (void) fprintf(stderr, gettext("missing target dataset " "argument\n")); usage(B_FALSE); } if (argc > 2) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } if (flags.recursive && parents) { (void) fprintf(stderr, gettext("-p and -r options are mutually " "exclusive\n")); usage(B_FALSE); } if (flags.nounmount && parents) { (void) fprintf(stderr, gettext("-u and -p options are mutually " "exclusive\n")); usage(B_FALSE); } if (flags.recursive && strchr(argv[0], '@') == 0) { (void) fprintf(stderr, gettext("source dataset for recursive " "rename must be a snapshot\n")); usage(B_FALSE); } if (flags.nounmount) types = ZFS_TYPE_FILESYSTEM; else if (parents) types = ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME; else types = ZFS_TYPE_DATASET; if ((zhp = zfs_open(g_zfs, argv[0], types)) == NULL) return (1); /* If we were asked and the name looks good, try to create ancestors. */ if (parents && zfs_name_valid(argv[1], zfs_get_type(zhp)) && zfs_create_ancestors(g_zfs, argv[1]) != 0) { zfs_close(zhp); return (1); } ret = (zfs_rename(zhp, argv[1], flags) != 0); zfs_close(zhp); return (ret); } /* * zfs promote * * Promotes the given clone fs to be the parent */ /* ARGSUSED */ static int zfs_do_promote(int argc, char **argv) { zfs_handle_t *zhp; int ret = 0; /* check options */ if (argc > 1 && argv[1][0] == '-') { (void) fprintf(stderr, gettext("invalid option '%c'\n"), argv[1][1]); usage(B_FALSE); } /* check number of arguments */ if (argc < 2) { (void) fprintf(stderr, gettext("missing clone filesystem" " argument\n")); usage(B_FALSE); } if (argc > 2) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } zhp = zfs_open(g_zfs, argv[1], ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME); if (zhp == NULL) return (1); ret = (zfs_promote(zhp) != 0); zfs_close(zhp); return (ret); } static int zfs_do_redact(int argc, char **argv) { char *snap = NULL; char *bookname = NULL; char **rsnaps = NULL; int numrsnaps = 0; argv++; argc--; if (argc < 3) { (void) fprintf(stderr, gettext("too few arguments\n")); usage(B_FALSE); } snap = argv[0]; bookname = argv[1]; rsnaps = argv + 2; numrsnaps = argc - 2; nvlist_t *rsnapnv = fnvlist_alloc(); for (int i = 0; i < numrsnaps; i++) { fnvlist_add_boolean(rsnapnv, rsnaps[i]); } int err = lzc_redact(snap, bookname, rsnapnv); fnvlist_free(rsnapnv); switch (err) { case 0: break; case ENOENT: (void) fprintf(stderr, gettext("provided snapshot %s does not exist\n"), snap); break; case EEXIST: (void) fprintf(stderr, gettext("specified redaction bookmark " "(%s) provided already exists\n"), bookname); break; case ENAMETOOLONG: (void) fprintf(stderr, gettext("provided bookmark name cannot " "be used, final name would be too long\n")); break; case E2BIG: (void) fprintf(stderr, gettext("too many redaction snapshots " "specified\n")); break; case EINVAL: if (strchr(bookname, '#') != NULL) (void) fprintf(stderr, gettext( "redaction bookmark name must not contain '#'\n")); else (void) fprintf(stderr, gettext( "redaction snapshot must be descendent of " "snapshot being redacted\n")); break; case EALREADY: (void) fprintf(stderr, gettext("attempted to redact redacted " "dataset or with respect to redacted dataset\n")); break; case ENOTSUP: (void) fprintf(stderr, gettext("redaction bookmarks feature " "not enabled\n")); break; case EXDEV: (void) fprintf(stderr, gettext("potentially invalid redaction " "snapshot; full dataset names required\n")); break; default: (void) fprintf(stderr, gettext("internal error: %s\n"), strerror(errno)); } return (err); } /* * zfs rollback [-rRf] * * -r Delete any intervening snapshots before doing rollback * -R Delete any snapshots and their clones * -f ignored for backwards compatibility * * Given a filesystem, rollback to a specific snapshot, discarding any changes * since then and making it the active dataset. If more recent snapshots exist, * the command will complain unless the '-r' flag is given. */ typedef struct rollback_cbdata { uint64_t cb_create; uint8_t cb_younger_ds_printed; boolean_t cb_first; int cb_doclones; char *cb_target; int cb_error; boolean_t cb_recurse; } rollback_cbdata_t; static int rollback_check_dependent(zfs_handle_t *zhp, void *data) { rollback_cbdata_t *cbp = data; if (cbp->cb_first && cbp->cb_recurse) { (void) fprintf(stderr, gettext("cannot rollback to " "'%s': clones of previous snapshots exist\n"), cbp->cb_target); (void) fprintf(stderr, gettext("use '-R' to " "force deletion of the following clones and " "dependents:\n")); cbp->cb_first = 0; cbp->cb_error = 1; } (void) fprintf(stderr, "%s\n", zfs_get_name(zhp)); zfs_close(zhp); return (0); } /* * Report some snapshots/bookmarks more recent than the one specified. * Used when '-r' is not specified. We reuse this same callback for the * snapshot dependents - if 'cb_dependent' is set, then this is a * dependent and we should report it without checking the transaction group. */ static int rollback_check(zfs_handle_t *zhp, void *data) { rollback_cbdata_t *cbp = data; /* * Max number of younger snapshots and/or bookmarks to display before * we stop the iteration. */ const uint8_t max_younger = 32; if (cbp->cb_doclones) { zfs_close(zhp); return (0); } if (zfs_prop_get_int(zhp, ZFS_PROP_CREATETXG) > cbp->cb_create) { if (cbp->cb_first && !cbp->cb_recurse) { (void) fprintf(stderr, gettext("cannot " "rollback to '%s': more recent snapshots " "or bookmarks exist\n"), cbp->cb_target); (void) fprintf(stderr, gettext("use '-r' to " "force deletion of the following " "snapshots and bookmarks:\n")); cbp->cb_first = 0; cbp->cb_error = 1; } if (cbp->cb_recurse) { if (zfs_iter_dependents(zhp, B_TRUE, rollback_check_dependent, cbp) != 0) { zfs_close(zhp); return (-1); } } else { (void) fprintf(stderr, "%s\n", zfs_get_name(zhp)); cbp->cb_younger_ds_printed++; } } zfs_close(zhp); if (cbp->cb_younger_ds_printed == max_younger) { /* * This non-recursive rollback is going to fail due to the * presence of snapshots and/or bookmarks that are younger than * the rollback target. * We printed some of the offending objects, now we stop * zfs_iter_snapshot/bookmark iteration so we can fail fast and * avoid iterating over the rest of the younger objects */ (void) fprintf(stderr, gettext("Output limited to %d " "snapshots/bookmarks\n"), max_younger); return (-1); } return (0); } static int zfs_do_rollback(int argc, char **argv) { int ret = 0; int c; boolean_t force = B_FALSE; rollback_cbdata_t cb = { 0 }; zfs_handle_t *zhp, *snap; char parentname[ZFS_MAX_DATASET_NAME_LEN]; char *delim; uint64_t min_txg = 0; /* check options */ while ((c = getopt(argc, argv, "rRf")) != -1) { switch (c) { case 'r': cb.cb_recurse = 1; break; case 'R': cb.cb_recurse = 1; cb.cb_doclones = 1; break; case 'f': force = B_TRUE; break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; /* check number of arguments */ if (argc < 1) { (void) fprintf(stderr, gettext("missing dataset argument\n")); usage(B_FALSE); } if (argc > 1) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } /* open the snapshot */ if ((snap = zfs_open(g_zfs, argv[0], ZFS_TYPE_SNAPSHOT)) == NULL) return (1); /* open the parent dataset */ (void) strlcpy(parentname, argv[0], sizeof (parentname)); verify((delim = strrchr(parentname, '@')) != NULL); *delim = '\0'; if ((zhp = zfs_open(g_zfs, parentname, ZFS_TYPE_DATASET)) == NULL) { zfs_close(snap); return (1); } /* * Check for more recent snapshots and/or clones based on the presence * of '-r' and '-R'. */ cb.cb_target = argv[0]; cb.cb_create = zfs_prop_get_int(snap, ZFS_PROP_CREATETXG); cb.cb_first = B_TRUE; cb.cb_error = 0; if (cb.cb_create > 0) min_txg = cb.cb_create; if ((ret = zfs_iter_snapshots(zhp, B_FALSE, rollback_check, &cb, min_txg, 0)) != 0) goto out; if ((ret = zfs_iter_bookmarks(zhp, rollback_check, &cb)) != 0) goto out; if ((ret = cb.cb_error) != 0) goto out; /* * Rollback parent to the given snapshot. */ ret = zfs_rollback(zhp, snap, force); out: zfs_close(snap); zfs_close(zhp); if (ret == 0) return (0); else return (1); } /* * zfs set property=value ... { fs | snap | vol } ... * * Sets the given properties for all datasets specified on the command line. */ static int set_callback(zfs_handle_t *zhp, void *data) { nvlist_t *props = data; if (zfs_prop_set_list(zhp, props) != 0) { switch (libzfs_errno(g_zfs)) { case EZFS_MOUNTFAILED: (void) fprintf(stderr, gettext("property may be set " "but unable to remount filesystem\n")); break; case EZFS_SHARENFSFAILED: (void) fprintf(stderr, gettext("property may be set " "but unable to reshare filesystem\n")); break; } return (1); } return (0); } static int zfs_do_set(int argc, char **argv) { nvlist_t *props = NULL; int ds_start = -1; /* argv idx of first dataset arg */ int ret = 0; int i; /* check for options */ if (argc > 1 && argv[1][0] == '-') { (void) fprintf(stderr, gettext("invalid option '%c'\n"), argv[1][1]); usage(B_FALSE); } /* check number of arguments */ if (argc < 2) { (void) fprintf(stderr, gettext("missing arguments\n")); usage(B_FALSE); } if (argc < 3) { if (strchr(argv[1], '=') == NULL) { (void) fprintf(stderr, gettext("missing property=value " "argument(s)\n")); } else { (void) fprintf(stderr, gettext("missing dataset " "name(s)\n")); } usage(B_FALSE); } /* validate argument order: prop=val args followed by dataset args */ for (i = 1; i < argc; i++) { if (strchr(argv[i], '=') != NULL) { if (ds_start > 0) { /* out-of-order prop=val argument */ (void) fprintf(stderr, gettext("invalid " "argument order\n")); usage(B_FALSE); } } else if (ds_start < 0) { ds_start = i; } } if (ds_start < 0) { (void) fprintf(stderr, gettext("missing dataset name(s)\n")); usage(B_FALSE); } /* Populate a list of property settings */ if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) nomem(); for (i = 1; i < ds_start; i++) { if (!parseprop(props, argv[i])) { ret = -1; goto error; } } ret = zfs_for_each(argc - ds_start, argv + ds_start, 0, ZFS_TYPE_DATASET, NULL, NULL, 0, set_callback, props); error: nvlist_free(props); 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; if (sd->sd_recursive && zfs_prop_get_int(zhp, ZFS_PROP_INCONSISTENT) != 0) { zfs_close(zhp); return (0); } 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] ... * * Creates a snapshot with the given name. While functionally equivalent to * 'zfs create', it is a separate command to differentiate intent. */ static int zfs_do_snapshot(int argc, char **argv) { int ret = 0; int 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) { switch (c) { case 'o': if (!parseprop(props, optarg)) { nvlist_free(sd.sd_nvl); nvlist_free(props); return (1); } break; case 'r': sd.sd_recursive = B_TRUE; multiple_snaps = B_TRUE; break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); goto usage; } } argc -= optind; argv += optind; /* check number of arguments */ if (argc < 1) { (void) fprintf(stderr, gettext("missing snapshot argument\n")); goto usage; } if (argc > 1) 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 != 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); } /* * Send a backup stream to stdout. */ static int zfs_do_send(int argc, char **argv) { char *fromname = NULL; char *toname = NULL; char *resume_token = NULL; char *cp; zfs_handle_t *zhp; sendflags_t flags = { 0 }; int c, err; nvlist_t *dbgnv = NULL; char *redactbook = NULL; struct option long_options[] = { {"replicate", no_argument, NULL, 'R'}, {"redact", required_argument, NULL, 'd'}, {"props", no_argument, NULL, 'p'}, {"parsable", no_argument, NULL, 'P'}, {"dedup", no_argument, NULL, 'D'}, {"verbose", no_argument, NULL, 'v'}, {"dryrun", no_argument, NULL, 'n'}, {"large-block", no_argument, NULL, 'L'}, {"embed", no_argument, NULL, 'e'}, {"resume", required_argument, NULL, 't'}, {"compressed", no_argument, NULL, 'c'}, {"raw", no_argument, NULL, 'w'}, {"backup", no_argument, NULL, 'b'}, {"holds", no_argument, NULL, 'h'}, {"saved", no_argument, NULL, 'S'}, {0, 0, 0, 0} }; /* check options */ while ((c = getopt_long(argc, argv, ":i:I:RDpvnPLeht:cwbd:S", long_options, NULL)) != -1) { switch (c) { case 'i': if (fromname) usage(B_FALSE); fromname = optarg; break; case 'I': if (fromname) usage(B_FALSE); fromname = optarg; flags.doall = B_TRUE; break; case 'R': flags.replicate = B_TRUE; break; case 'd': redactbook = optarg; break; case 'p': flags.props = B_TRUE; break; case 'b': flags.backup = B_TRUE; break; case 'h': flags.holds = B_TRUE; break; case 'P': flags.parsable = B_TRUE; break; case 'v': flags.verbosity++; flags.progress = B_TRUE; break; case 'D': (void) fprintf(stderr, gettext("WARNING: deduplicated send is no " "longer supported. A regular,\n" "non-deduplicated stream will be generated.\n\n")); break; case 'n': flags.dryrun = B_TRUE; break; case 'L': flags.largeblock = B_TRUE; break; case 'e': flags.embed_data = B_TRUE; break; case 't': resume_token = optarg; break; case 'c': flags.compress = B_TRUE; break; case 'w': flags.raw = B_TRUE; flags.compress = B_TRUE; flags.embed_data = B_TRUE; flags.largeblock = B_TRUE; break; case 'S': flags.saved = B_TRUE; break; case ':': /* * If a parameter was not passed, optopt contains the * value that would normally lead us into the * appropriate case statement. If it's > 256, then this * must be a longopt and we should look at argv to get * the string. Otherwise it's just the character, so we * should use it directly. */ if (optopt <= UINT8_MAX) { (void) fprintf(stderr, gettext("missing argument for '%c' " "option\n"), optopt); } else { (void) fprintf(stderr, gettext("missing argument for '%s' " "option\n"), argv[optind - 1]); } usage(B_FALSE); break; case '?': /*FALLTHROUGH*/ default: /* * If an invalid flag was passed, optopt contains the * character if it was a short flag, or 0 if it was a * longopt. */ if (optopt != 0) { (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); } else { (void) fprintf(stderr, gettext("invalid option '%s'\n"), argv[optind - 1]); } usage(B_FALSE); } } if (flags.parsable && flags.verbosity == 0) flags.verbosity = 1; argc -= optind; argv += optind; if (resume_token != NULL) { if (fromname != NULL || flags.replicate || flags.props || flags.backup || flags.holds || flags.saved || redactbook != NULL) { (void) fprintf(stderr, gettext("invalid flags combined with -t\n")); usage(B_FALSE); } if (argc > 0) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } } else { if (argc < 1) { (void) fprintf(stderr, gettext("missing snapshot argument\n")); usage(B_FALSE); } if (argc > 1) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } } if (flags.saved) { if (fromname != NULL || flags.replicate || flags.props || flags.doall || flags.backup || flags.holds || flags.largeblock || flags.embed_data || flags.compress || flags.raw || redactbook != NULL) { (void) fprintf(stderr, gettext("incompatible flags " "combined with saved send flag\n")); usage(B_FALSE); } if (strchr(argv[0], '@') != NULL) { (void) fprintf(stderr, gettext("saved send must " "specify the dataset with partially-received " "state\n")); usage(B_FALSE); } } if (flags.raw && redactbook != NULL) { (void) fprintf(stderr, gettext("Error: raw sends may not be redacted.\n")); return (1); } if (!flags.dryrun && isatty(STDOUT_FILENO)) { (void) fprintf(stderr, gettext("Error: Stream can not be written to a terminal.\n" "You must redirect standard output.\n")); return (1); } if (flags.saved) { zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_DATASET); if (zhp == NULL) return (1); err = zfs_send_saved(zhp, &flags, STDOUT_FILENO, resume_token); zfs_close(zhp); return (err != 0); } else if (resume_token != NULL) { return (zfs_send_resume(g_zfs, &flags, STDOUT_FILENO, resume_token)); } /* * For everything except -R and -I, use the new, cleaner code path. */ if (!(flags.replicate || flags.doall)) { char frombuf[ZFS_MAX_DATASET_NAME_LEN]; if (fromname != NULL && (strchr(fromname, '#') == NULL && strchr(fromname, '@') == NULL)) { /* * Neither bookmark or snapshot was specified. Print a * warning, and assume snapshot. */ (void) fprintf(stderr, "Warning: incremental source " "didn't specify type, assuming snapshot. Use '@' " "or '#' prefix to avoid ambiguity.\n"); (void) snprintf(frombuf, sizeof (frombuf), "@%s", fromname); fromname = frombuf; } if (fromname != NULL && (fromname[0] == '#' || fromname[0] == '@')) { /* * Incremental source name begins with # or @. * Default to same fs as target. */ char tmpbuf[ZFS_MAX_DATASET_NAME_LEN]; (void) strlcpy(tmpbuf, fromname, sizeof (tmpbuf)); (void) strlcpy(frombuf, argv[0], sizeof (frombuf)); cp = strchr(frombuf, '@'); if (cp != NULL) *cp = '\0'; (void) strlcat(frombuf, tmpbuf, sizeof (frombuf)); fromname = frombuf; } zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_DATASET); if (zhp == NULL) return (1); err = zfs_send_one(zhp, fromname, STDOUT_FILENO, &flags, redactbook); zfs_close(zhp); return (err != 0); } if (fromname != NULL && strchr(fromname, '#')) { (void) fprintf(stderr, gettext("Error: multiple snapshots cannot be " "sent from a bookmark.\n")); return (1); } if (redactbook != NULL) { (void) fprintf(stderr, gettext("Error: multiple snapshots " "cannot be sent redacted.\n")); return (1); } if ((cp = strchr(argv[0], '@')) == NULL) { (void) fprintf(stderr, gettext("Error: " "Unsupported flag with filesystem or bookmark.\n")); return (1); } *cp = '\0'; toname = cp + 1; zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME); if (zhp == NULL) return (1); /* * If they specified the full path to the snapshot, chop off * everything except the short name of the snapshot, but special * case if they specify the origin. */ if (fromname && (cp = strchr(fromname, '@')) != NULL) { char origin[ZFS_MAX_DATASET_NAME_LEN]; zprop_source_t src; (void) zfs_prop_get(zhp, ZFS_PROP_ORIGIN, origin, sizeof (origin), &src, NULL, 0, B_FALSE); if (strcmp(origin, fromname) == 0) { fromname = NULL; flags.fromorigin = B_TRUE; } else { *cp = '\0'; if (cp != fromname && strcmp(argv[0], fromname)) { (void) fprintf(stderr, gettext("incremental source must be " "in same filesystem\n")); usage(B_FALSE); } fromname = cp + 1; if (strchr(fromname, '@') || strchr(fromname, '/')) { (void) fprintf(stderr, gettext("invalid incremental source\n")); usage(B_FALSE); } } } if (flags.replicate && fromname == NULL) flags.doall = B_TRUE; err = zfs_send(zhp, fromname, toname, &flags, STDOUT_FILENO, NULL, 0, flags.verbosity >= 3 ? &dbgnv : NULL); if (flags.verbosity >= 3 && dbgnv != NULL) { /* * dump_nvlist prints to stdout, but that's been * redirected to a file. Make it print to stderr * instead. */ (void) dup2(STDERR_FILENO, STDOUT_FILENO); dump_nvlist(dbgnv, 0); nvlist_free(dbgnv); } zfs_close(zhp); return (err != 0); } /* * Restore a backup stream from stdin. */ static int zfs_do_receive(int argc, char **argv) { int c, err = 0; recvflags_t flags = { 0 }; boolean_t abort_resumable = B_FALSE; nvlist_t *props; if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) nomem(); /* check options */ while ((c = getopt(argc, argv, ":o:x:dehMnuvFsA")) != -1) { switch (c) { case 'o': if (!parseprop(props, optarg)) { nvlist_free(props); usage(B_FALSE); } break; case 'x': if (!parsepropname(props, optarg)) { nvlist_free(props); usage(B_FALSE); } break; case 'd': if (flags.istail) { (void) fprintf(stderr, gettext("invalid option " "combination: -d and -e are mutually " "exclusive\n")); usage(B_FALSE); } flags.isprefix = B_TRUE; break; case 'e': if (flags.isprefix) { (void) fprintf(stderr, gettext("invalid option " "combination: -d and -e are mutually " "exclusive\n")); usage(B_FALSE); } flags.istail = B_TRUE; break; case 'h': flags.skipholds = B_TRUE; break; case 'M': flags.forceunmount = B_TRUE; break; case 'n': flags.dryrun = B_TRUE; break; case 'u': flags.nomount = B_TRUE; break; case 'v': flags.verbose = B_TRUE; break; case 's': flags.resumable = B_TRUE; break; case 'F': flags.force = B_TRUE; break; case 'A': abort_resumable = B_TRUE; break; case ':': (void) fprintf(stderr, gettext("missing argument for " "'%c' option\n"), optopt); usage(B_FALSE); break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; /* zfs recv -e (use "tail" name) implies -d (remove dataset "head") */ if (flags.istail) flags.isprefix = B_TRUE; /* check number of arguments */ if (argc < 1) { (void) fprintf(stderr, gettext("missing snapshot argument\n")); usage(B_FALSE); } if (argc > 1) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } if (abort_resumable) { if (flags.isprefix || flags.istail || flags.dryrun || flags.resumable || flags.nomount) { (void) fprintf(stderr, gettext("invalid option\n")); usage(B_FALSE); } char namebuf[ZFS_MAX_DATASET_NAME_LEN]; (void) snprintf(namebuf, sizeof (namebuf), "%s/%%recv", argv[0]); if (zfs_dataset_exists(g_zfs, namebuf, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME)) { zfs_handle_t *zhp = zfs_open(g_zfs, namebuf, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME); if (zhp == NULL) { nvlist_free(props); return (1); } err = zfs_destroy(zhp, B_FALSE); zfs_close(zhp); } else { zfs_handle_t *zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME); if (zhp == NULL) usage(B_FALSE); if (!zfs_prop_get_int(zhp, ZFS_PROP_INCONSISTENT) || zfs_prop_get(zhp, ZFS_PROP_RECEIVE_RESUME_TOKEN, NULL, 0, NULL, NULL, 0, B_TRUE) == -1) { (void) fprintf(stderr, gettext("'%s' does not have any " "resumable receive state to abort\n"), argv[0]); nvlist_free(props); zfs_close(zhp); return (1); } err = zfs_destroy(zhp, B_FALSE); zfs_close(zhp); } nvlist_free(props); return (err != 0); } if (isatty(STDIN_FILENO)) { (void) fprintf(stderr, gettext("Error: Backup stream can not be read " "from a terminal.\n" "You must redirect standard input.\n")); nvlist_free(props); return (1); } err = zfs_receive(g_zfs, argv[0], props, &flags, STDIN_FILENO, NULL); nvlist_free(props); return (err != 0); } /* * allow/unallow stuff */ /* copied from zfs/sys/dsl_deleg.h */ #define ZFS_DELEG_PERM_CREATE "create" #define ZFS_DELEG_PERM_DESTROY "destroy" #define ZFS_DELEG_PERM_SNAPSHOT "snapshot" #define ZFS_DELEG_PERM_ROLLBACK "rollback" #define ZFS_DELEG_PERM_CLONE "clone" #define ZFS_DELEG_PERM_PROMOTE "promote" #define ZFS_DELEG_PERM_RENAME "rename" #define ZFS_DELEG_PERM_MOUNT "mount" #define ZFS_DELEG_PERM_SHARE "share" #define ZFS_DELEG_PERM_SEND "send" #define ZFS_DELEG_PERM_RECEIVE "receive" #define ZFS_DELEG_PERM_ALLOW "allow" #define ZFS_DELEG_PERM_USERPROP "userprop" #define ZFS_DELEG_PERM_VSCAN "vscan" /* ??? */ #define ZFS_DELEG_PERM_USERQUOTA "userquota" #define ZFS_DELEG_PERM_GROUPQUOTA "groupquota" #define ZFS_DELEG_PERM_USERUSED "userused" #define ZFS_DELEG_PERM_GROUPUSED "groupused" #define ZFS_DELEG_PERM_USEROBJQUOTA "userobjquota" #define ZFS_DELEG_PERM_GROUPOBJQUOTA "groupobjquota" #define ZFS_DELEG_PERM_USEROBJUSED "userobjused" #define ZFS_DELEG_PERM_GROUPOBJUSED "groupobjused" #define ZFS_DELEG_PERM_HOLD "hold" #define ZFS_DELEG_PERM_RELEASE "release" #define ZFS_DELEG_PERM_DIFF "diff" #define ZFS_DELEG_PERM_BOOKMARK "bookmark" #define ZFS_DELEG_PERM_LOAD_KEY "load-key" #define ZFS_DELEG_PERM_CHANGE_KEY "change-key" #define ZFS_DELEG_PERM_PROJECTUSED "projectused" #define ZFS_DELEG_PERM_PROJECTQUOTA "projectquota" #define ZFS_DELEG_PERM_PROJECTOBJUSED "projectobjused" #define ZFS_DELEG_PERM_PROJECTOBJQUOTA "projectobjquota" #define ZFS_NUM_DELEG_NOTES ZFS_DELEG_NOTE_NONE static zfs_deleg_perm_tab_t zfs_deleg_perm_tbl[] = { { ZFS_DELEG_PERM_ALLOW, ZFS_DELEG_NOTE_ALLOW }, { ZFS_DELEG_PERM_CLONE, ZFS_DELEG_NOTE_CLONE }, { ZFS_DELEG_PERM_CREATE, ZFS_DELEG_NOTE_CREATE }, { ZFS_DELEG_PERM_DESTROY, ZFS_DELEG_NOTE_DESTROY }, { ZFS_DELEG_PERM_DIFF, ZFS_DELEG_NOTE_DIFF}, { ZFS_DELEG_PERM_HOLD, ZFS_DELEG_NOTE_HOLD }, { ZFS_DELEG_PERM_MOUNT, ZFS_DELEG_NOTE_MOUNT }, { ZFS_DELEG_PERM_PROMOTE, ZFS_DELEG_NOTE_PROMOTE }, { ZFS_DELEG_PERM_RECEIVE, ZFS_DELEG_NOTE_RECEIVE }, { ZFS_DELEG_PERM_RELEASE, ZFS_DELEG_NOTE_RELEASE }, { ZFS_DELEG_PERM_RENAME, ZFS_DELEG_NOTE_RENAME }, { ZFS_DELEG_PERM_ROLLBACK, ZFS_DELEG_NOTE_ROLLBACK }, { ZFS_DELEG_PERM_SEND, ZFS_DELEG_NOTE_SEND }, { ZFS_DELEG_PERM_SHARE, ZFS_DELEG_NOTE_SHARE }, { ZFS_DELEG_PERM_SNAPSHOT, ZFS_DELEG_NOTE_SNAPSHOT }, { ZFS_DELEG_PERM_BOOKMARK, ZFS_DELEG_NOTE_BOOKMARK }, { ZFS_DELEG_PERM_LOAD_KEY, ZFS_DELEG_NOTE_LOAD_KEY }, { ZFS_DELEG_PERM_CHANGE_KEY, ZFS_DELEG_NOTE_CHANGE_KEY }, { ZFS_DELEG_PERM_GROUPQUOTA, ZFS_DELEG_NOTE_GROUPQUOTA }, { ZFS_DELEG_PERM_GROUPUSED, ZFS_DELEG_NOTE_GROUPUSED }, { ZFS_DELEG_PERM_USERPROP, ZFS_DELEG_NOTE_USERPROP }, { ZFS_DELEG_PERM_USERQUOTA, ZFS_DELEG_NOTE_USERQUOTA }, { ZFS_DELEG_PERM_USERUSED, ZFS_DELEG_NOTE_USERUSED }, { ZFS_DELEG_PERM_USEROBJQUOTA, ZFS_DELEG_NOTE_USEROBJQUOTA }, { ZFS_DELEG_PERM_USEROBJUSED, ZFS_DELEG_NOTE_USEROBJUSED }, { ZFS_DELEG_PERM_GROUPOBJQUOTA, ZFS_DELEG_NOTE_GROUPOBJQUOTA }, { ZFS_DELEG_PERM_GROUPOBJUSED, ZFS_DELEG_NOTE_GROUPOBJUSED }, { ZFS_DELEG_PERM_PROJECTUSED, ZFS_DELEG_NOTE_PROJECTUSED }, { ZFS_DELEG_PERM_PROJECTQUOTA, ZFS_DELEG_NOTE_PROJECTQUOTA }, { ZFS_DELEG_PERM_PROJECTOBJUSED, ZFS_DELEG_NOTE_PROJECTOBJUSED }, { ZFS_DELEG_PERM_PROJECTOBJQUOTA, ZFS_DELEG_NOTE_PROJECTOBJQUOTA }, { NULL, ZFS_DELEG_NOTE_NONE } }; /* permission structure */ typedef struct deleg_perm { zfs_deleg_who_type_t dp_who_type; const char *dp_name; boolean_t dp_local; boolean_t dp_descend; } deleg_perm_t; /* */ typedef struct deleg_perm_node { deleg_perm_t dpn_perm; uu_avl_node_t dpn_avl_node; } deleg_perm_node_t; typedef struct fs_perm fs_perm_t; /* permissions set */ typedef struct who_perm { zfs_deleg_who_type_t who_type; const char *who_name; /* id */ char who_ug_name[256]; /* user/group name */ fs_perm_t *who_fsperm; /* uplink */ uu_avl_t *who_deleg_perm_avl; /* permissions */ } who_perm_t; /* */ typedef struct who_perm_node { who_perm_t who_perm; uu_avl_node_t who_avl_node; } who_perm_node_t; typedef struct fs_perm_set fs_perm_set_t; /* fs permissions */ struct fs_perm { const char *fsp_name; uu_avl_t *fsp_sc_avl; /* sets,create */ uu_avl_t *fsp_uge_avl; /* user,group,everyone */ fs_perm_set_t *fsp_set; /* uplink */ }; /* */ typedef struct fs_perm_node { fs_perm_t fspn_fsperm; uu_avl_t *fspn_avl; uu_list_node_t fspn_list_node; } fs_perm_node_t; /* top level structure */ struct fs_perm_set { uu_list_pool_t *fsps_list_pool; uu_list_t *fsps_list; /* list of fs_perms */ uu_avl_pool_t *fsps_named_set_avl_pool; uu_avl_pool_t *fsps_who_perm_avl_pool; uu_avl_pool_t *fsps_deleg_perm_avl_pool; }; static inline const char * deleg_perm_type(zfs_deleg_note_t note) { /* subcommands */ switch (note) { /* SUBCOMMANDS */ /* OTHER */ case ZFS_DELEG_NOTE_GROUPQUOTA: case ZFS_DELEG_NOTE_GROUPUSED: case ZFS_DELEG_NOTE_USERPROP: case ZFS_DELEG_NOTE_USERQUOTA: case ZFS_DELEG_NOTE_USERUSED: case ZFS_DELEG_NOTE_USEROBJQUOTA: case ZFS_DELEG_NOTE_USEROBJUSED: case ZFS_DELEG_NOTE_GROUPOBJQUOTA: case ZFS_DELEG_NOTE_GROUPOBJUSED: case ZFS_DELEG_NOTE_PROJECTUSED: case ZFS_DELEG_NOTE_PROJECTQUOTA: case ZFS_DELEG_NOTE_PROJECTOBJUSED: case ZFS_DELEG_NOTE_PROJECTOBJQUOTA: /* other */ return (gettext("other")); default: return (gettext("subcommand")); } } static int who_type2weight(zfs_deleg_who_type_t who_type) { int res; switch (who_type) { case ZFS_DELEG_NAMED_SET_SETS: case ZFS_DELEG_NAMED_SET: res = 0; break; case ZFS_DELEG_CREATE_SETS: case ZFS_DELEG_CREATE: res = 1; break; case ZFS_DELEG_USER_SETS: case ZFS_DELEG_USER: res = 2; break; case ZFS_DELEG_GROUP_SETS: case ZFS_DELEG_GROUP: res = 3; break; case ZFS_DELEG_EVERYONE_SETS: case ZFS_DELEG_EVERYONE: res = 4; break; default: res = -1; } return (res); } /* ARGSUSED */ static int who_perm_compare(const void *larg, const void *rarg, void *unused) { const who_perm_node_t *l = larg; const who_perm_node_t *r = rarg; zfs_deleg_who_type_t ltype = l->who_perm.who_type; zfs_deleg_who_type_t rtype = r->who_perm.who_type; int lweight = who_type2weight(ltype); int rweight = who_type2weight(rtype); int res = lweight - rweight; if (res == 0) res = strncmp(l->who_perm.who_name, r->who_perm.who_name, ZFS_MAX_DELEG_NAME-1); if (res == 0) return (0); if (res > 0) return (1); else return (-1); } /* ARGSUSED */ static int deleg_perm_compare(const void *larg, const void *rarg, void *unused) { const deleg_perm_node_t *l = larg; const deleg_perm_node_t *r = rarg; int res = strncmp(l->dpn_perm.dp_name, r->dpn_perm.dp_name, ZFS_MAX_DELEG_NAME-1); if (res == 0) return (0); if (res > 0) return (1); else return (-1); } static inline void fs_perm_set_init(fs_perm_set_t *fspset) { bzero(fspset, sizeof (fs_perm_set_t)); if ((fspset->fsps_list_pool = uu_list_pool_create("fsps_list_pool", sizeof (fs_perm_node_t), offsetof(fs_perm_node_t, fspn_list_node), NULL, UU_DEFAULT)) == NULL) nomem(); if ((fspset->fsps_list = uu_list_create(fspset->fsps_list_pool, NULL, UU_DEFAULT)) == NULL) nomem(); if ((fspset->fsps_named_set_avl_pool = uu_avl_pool_create( "named_set_avl_pool", sizeof (who_perm_node_t), offsetof( who_perm_node_t, who_avl_node), who_perm_compare, UU_DEFAULT)) == NULL) nomem(); if ((fspset->fsps_who_perm_avl_pool = uu_avl_pool_create( "who_perm_avl_pool", sizeof (who_perm_node_t), offsetof( who_perm_node_t, who_avl_node), who_perm_compare, UU_DEFAULT)) == NULL) nomem(); if ((fspset->fsps_deleg_perm_avl_pool = uu_avl_pool_create( "deleg_perm_avl_pool", sizeof (deleg_perm_node_t), offsetof( deleg_perm_node_t, dpn_avl_node), deleg_perm_compare, UU_DEFAULT)) == NULL) nomem(); } static inline void fs_perm_fini(fs_perm_t *); static inline void who_perm_fini(who_perm_t *); static inline void fs_perm_set_fini(fs_perm_set_t *fspset) { fs_perm_node_t *node = uu_list_first(fspset->fsps_list); while (node != NULL) { fs_perm_node_t *next_node = uu_list_next(fspset->fsps_list, node); fs_perm_t *fsperm = &node->fspn_fsperm; fs_perm_fini(fsperm); uu_list_remove(fspset->fsps_list, node); free(node); node = next_node; } uu_avl_pool_destroy(fspset->fsps_named_set_avl_pool); uu_avl_pool_destroy(fspset->fsps_who_perm_avl_pool); uu_avl_pool_destroy(fspset->fsps_deleg_perm_avl_pool); } static inline void deleg_perm_init(deleg_perm_t *deleg_perm, zfs_deleg_who_type_t type, const char *name) { deleg_perm->dp_who_type = type; deleg_perm->dp_name = name; } static inline void who_perm_init(who_perm_t *who_perm, fs_perm_t *fsperm, zfs_deleg_who_type_t type, const char *name) { uu_avl_pool_t *pool; pool = fsperm->fsp_set->fsps_deleg_perm_avl_pool; bzero(who_perm, sizeof (who_perm_t)); if ((who_perm->who_deleg_perm_avl = uu_avl_create(pool, NULL, UU_DEFAULT)) == NULL) nomem(); who_perm->who_type = type; who_perm->who_name = name; who_perm->who_fsperm = fsperm; } static inline void who_perm_fini(who_perm_t *who_perm) { deleg_perm_node_t *node = uu_avl_first(who_perm->who_deleg_perm_avl); while (node != NULL) { deleg_perm_node_t *next_node = uu_avl_next(who_perm->who_deleg_perm_avl, node); uu_avl_remove(who_perm->who_deleg_perm_avl, node); free(node); node = next_node; } uu_avl_destroy(who_perm->who_deleg_perm_avl); } static inline void fs_perm_init(fs_perm_t *fsperm, fs_perm_set_t *fspset, const char *fsname) { uu_avl_pool_t *nset_pool = fspset->fsps_named_set_avl_pool; uu_avl_pool_t *who_pool = fspset->fsps_who_perm_avl_pool; bzero(fsperm, sizeof (fs_perm_t)); if ((fsperm->fsp_sc_avl = uu_avl_create(nset_pool, NULL, UU_DEFAULT)) == NULL) nomem(); if ((fsperm->fsp_uge_avl = uu_avl_create(who_pool, NULL, UU_DEFAULT)) == NULL) nomem(); fsperm->fsp_set = fspset; fsperm->fsp_name = fsname; } static inline void fs_perm_fini(fs_perm_t *fsperm) { who_perm_node_t *node = uu_avl_first(fsperm->fsp_sc_avl); while (node != NULL) { who_perm_node_t *next_node = uu_avl_next(fsperm->fsp_sc_avl, node); who_perm_t *who_perm = &node->who_perm; who_perm_fini(who_perm); uu_avl_remove(fsperm->fsp_sc_avl, node); free(node); node = next_node; } node = uu_avl_first(fsperm->fsp_uge_avl); while (node != NULL) { who_perm_node_t *next_node = uu_avl_next(fsperm->fsp_uge_avl, node); who_perm_t *who_perm = &node->who_perm; who_perm_fini(who_perm); uu_avl_remove(fsperm->fsp_uge_avl, node); free(node); node = next_node; } uu_avl_destroy(fsperm->fsp_sc_avl); uu_avl_destroy(fsperm->fsp_uge_avl); } static void set_deleg_perm_node(uu_avl_t *avl, deleg_perm_node_t *node, zfs_deleg_who_type_t who_type, const char *name, char locality) { uu_avl_index_t idx = 0; deleg_perm_node_t *found_node = NULL; deleg_perm_t *deleg_perm = &node->dpn_perm; deleg_perm_init(deleg_perm, who_type, name); if ((found_node = uu_avl_find(avl, node, NULL, &idx)) == NULL) uu_avl_insert(avl, node, idx); else { node = found_node; deleg_perm = &node->dpn_perm; } switch (locality) { case ZFS_DELEG_LOCAL: deleg_perm->dp_local = B_TRUE; break; case ZFS_DELEG_DESCENDENT: deleg_perm->dp_descend = B_TRUE; break; case ZFS_DELEG_NA: break; default: assert(B_FALSE); /* invalid locality */ } } static inline int parse_who_perm(who_perm_t *who_perm, nvlist_t *nvl, char locality) { nvpair_t *nvp = NULL; fs_perm_set_t *fspset = who_perm->who_fsperm->fsp_set; uu_avl_t *avl = who_perm->who_deleg_perm_avl; zfs_deleg_who_type_t who_type = who_perm->who_type; while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) { const char *name = nvpair_name(nvp); data_type_t type = nvpair_type(nvp); uu_avl_pool_t *avl_pool = fspset->fsps_deleg_perm_avl_pool; deleg_perm_node_t *node = safe_malloc(sizeof (deleg_perm_node_t)); VERIFY(type == DATA_TYPE_BOOLEAN); uu_avl_node_init(node, &node->dpn_avl_node, avl_pool); set_deleg_perm_node(avl, node, who_type, name, locality); } return (0); } static inline int parse_fs_perm(fs_perm_t *fsperm, nvlist_t *nvl) { nvpair_t *nvp = NULL; fs_perm_set_t *fspset = fsperm->fsp_set; while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) { nvlist_t *nvl2 = NULL; const char *name = nvpair_name(nvp); uu_avl_t *avl = NULL; uu_avl_pool_t *avl_pool = NULL; zfs_deleg_who_type_t perm_type = name[0]; char perm_locality = name[1]; const char *perm_name = name + 3; who_perm_t *who_perm = NULL; assert('$' == name[2]); if (nvpair_value_nvlist(nvp, &nvl2) != 0) return (-1); switch (perm_type) { case ZFS_DELEG_CREATE: case ZFS_DELEG_CREATE_SETS: case ZFS_DELEG_NAMED_SET: case ZFS_DELEG_NAMED_SET_SETS: avl_pool = fspset->fsps_named_set_avl_pool; avl = fsperm->fsp_sc_avl; break; case ZFS_DELEG_USER: case ZFS_DELEG_USER_SETS: case ZFS_DELEG_GROUP: case ZFS_DELEG_GROUP_SETS: case ZFS_DELEG_EVERYONE: case ZFS_DELEG_EVERYONE_SETS: avl_pool = fspset->fsps_who_perm_avl_pool; avl = fsperm->fsp_uge_avl; break; default: assert(!"unhandled zfs_deleg_who_type_t"); } who_perm_node_t *found_node = NULL; who_perm_node_t *node = safe_malloc( sizeof (who_perm_node_t)); who_perm = &node->who_perm; uu_avl_index_t idx = 0; uu_avl_node_init(node, &node->who_avl_node, avl_pool); who_perm_init(who_perm, fsperm, perm_type, perm_name); if ((found_node = uu_avl_find(avl, node, NULL, &idx)) == NULL) { if (avl == fsperm->fsp_uge_avl) { uid_t rid = 0; struct passwd *p = NULL; struct group *g = NULL; const char *nice_name = NULL; switch (perm_type) { case ZFS_DELEG_USER_SETS: case ZFS_DELEG_USER: rid = atoi(perm_name); p = getpwuid(rid); if (p) nice_name = p->pw_name; break; case ZFS_DELEG_GROUP_SETS: case ZFS_DELEG_GROUP: rid = atoi(perm_name); g = getgrgid(rid); if (g) nice_name = g->gr_name; break; default: break; } if (nice_name != NULL) { (void) strlcpy( node->who_perm.who_ug_name, nice_name, 256); } else { /* User or group unknown */ (void) snprintf( node->who_perm.who_ug_name, sizeof (node->who_perm.who_ug_name), "(unknown: %d)", rid); } } uu_avl_insert(avl, node, idx); } else { node = found_node; who_perm = &node->who_perm; } assert(who_perm != NULL); (void) parse_who_perm(who_perm, nvl2, perm_locality); } return (0); } static inline int parse_fs_perm_set(fs_perm_set_t *fspset, nvlist_t *nvl) { nvpair_t *nvp = NULL; uu_avl_index_t idx = 0; while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) { nvlist_t *nvl2 = NULL; const char *fsname = nvpair_name(nvp); data_type_t type = nvpair_type(nvp); fs_perm_t *fsperm = NULL; fs_perm_node_t *node = safe_malloc(sizeof (fs_perm_node_t)); if (node == NULL) nomem(); fsperm = &node->fspn_fsperm; VERIFY(DATA_TYPE_NVLIST == type); uu_list_node_init(node, &node->fspn_list_node, fspset->fsps_list_pool); idx = uu_list_numnodes(fspset->fsps_list); fs_perm_init(fsperm, fspset, fsname); if (nvpair_value_nvlist(nvp, &nvl2) != 0) return (-1); (void) parse_fs_perm(fsperm, nvl2); uu_list_insert(fspset->fsps_list, node, idx); } return (0); } static inline const char * deleg_perm_comment(zfs_deleg_note_t note) { const char *str = ""; /* subcommands */ switch (note) { /* SUBCOMMANDS */ case ZFS_DELEG_NOTE_ALLOW: str = gettext("Must also have the permission that is being" "\n\t\t\t\tallowed"); break; case ZFS_DELEG_NOTE_CLONE: str = gettext("Must also have the 'create' ability and 'mount'" "\n\t\t\t\tability in the origin file system"); break; case ZFS_DELEG_NOTE_CREATE: str = gettext("Must also have the 'mount' ability"); break; case ZFS_DELEG_NOTE_DESTROY: str = gettext("Must also have the 'mount' ability"); break; case ZFS_DELEG_NOTE_DIFF: str = gettext("Allows lookup of paths within a dataset;" "\n\t\t\t\tgiven an object number. Ordinary users need this" "\n\t\t\t\tin order to use zfs diff"); break; case ZFS_DELEG_NOTE_HOLD: str = gettext("Allows adding a user hold to a snapshot"); break; case ZFS_DELEG_NOTE_MOUNT: str = gettext("Allows mount/umount of ZFS datasets"); break; case ZFS_DELEG_NOTE_PROMOTE: str = gettext("Must also have the 'mount'\n\t\t\t\tand" " 'promote' ability in the origin file system"); break; case ZFS_DELEG_NOTE_RECEIVE: str = gettext("Must also have the 'mount' and 'create'" " ability"); break; case ZFS_DELEG_NOTE_RELEASE: str = gettext("Allows releasing a user hold which\n\t\t\t\t" "might destroy the snapshot"); break; case ZFS_DELEG_NOTE_RENAME: str = gettext("Must also have the 'mount' and 'create'" "\n\t\t\t\tability in the new parent"); break; case ZFS_DELEG_NOTE_ROLLBACK: str = gettext(""); break; case ZFS_DELEG_NOTE_SEND: str = gettext(""); break; case ZFS_DELEG_NOTE_SHARE: str = gettext("Allows sharing file systems over NFS or SMB" "\n\t\t\t\tprotocols"); break; case ZFS_DELEG_NOTE_SNAPSHOT: str = gettext(""); break; case ZFS_DELEG_NOTE_LOAD_KEY: str = gettext("Allows loading or unloading an encryption key"); break; case ZFS_DELEG_NOTE_CHANGE_KEY: str = gettext("Allows changing or adding an encryption key"); break; /* * case ZFS_DELEG_NOTE_VSCAN: * str = gettext(""); * break; */ /* OTHER */ case ZFS_DELEG_NOTE_GROUPQUOTA: str = gettext("Allows accessing any groupquota@... property"); break; case ZFS_DELEG_NOTE_GROUPUSED: str = gettext("Allows reading any groupused@... property"); break; case ZFS_DELEG_NOTE_USERPROP: str = gettext("Allows changing any user property"); break; case ZFS_DELEG_NOTE_USERQUOTA: str = gettext("Allows accessing any userquota@... property"); break; case ZFS_DELEG_NOTE_USERUSED: str = gettext("Allows reading any userused@... property"); break; case ZFS_DELEG_NOTE_USEROBJQUOTA: str = gettext("Allows accessing any userobjquota@... property"); break; case ZFS_DELEG_NOTE_GROUPOBJQUOTA: str = gettext("Allows accessing any \n\t\t\t\t" "groupobjquota@... property"); break; case ZFS_DELEG_NOTE_GROUPOBJUSED: str = gettext("Allows reading any groupobjused@... property"); break; case ZFS_DELEG_NOTE_USEROBJUSED: str = gettext("Allows reading any userobjused@... property"); break; case ZFS_DELEG_NOTE_PROJECTQUOTA: str = gettext("Allows accessing any projectquota@... property"); break; case ZFS_DELEG_NOTE_PROJECTOBJQUOTA: str = gettext("Allows accessing any \n\t\t\t\t" "projectobjquota@... property"); break; case ZFS_DELEG_NOTE_PROJECTUSED: str = gettext("Allows reading any projectused@... property"); break; case ZFS_DELEG_NOTE_PROJECTOBJUSED: str = gettext("Allows accessing any \n\t\t\t\t" "projectobjused@... property"); break; /* other */ default: str = ""; } return (str); } struct allow_opts { boolean_t local; boolean_t descend; boolean_t user; boolean_t group; boolean_t everyone; boolean_t create; boolean_t set; boolean_t recursive; /* unallow only */ boolean_t prt_usage; boolean_t prt_perms; char *who; char *perms; const char *dataset; }; static inline int prop_cmp(const void *a, const void *b) { const char *str1 = *(const char **)a; const char *str2 = *(const char **)b; return (strcmp(str1, str2)); } static void allow_usage(boolean_t un, boolean_t requested, const char *msg) { const char *opt_desc[] = { "-h", gettext("show this help message and exit"), "-l", gettext("set permission locally"), "-d", gettext("set permission for descents"), "-u", gettext("set permission for user"), "-g", gettext("set permission for group"), "-e", gettext("set permission for everyone"), "-c", gettext("set create time permission"), "-s", gettext("define permission set"), /* unallow only */ "-r", gettext("remove permissions recursively"), }; size_t unallow_size = sizeof (opt_desc) / sizeof (char *); size_t allow_size = unallow_size - 2; const char *props[ZFS_NUM_PROPS]; int i; size_t count = 0; FILE *fp = requested ? stdout : stderr; zprop_desc_t *pdtbl = zfs_prop_get_table(); const char *fmt = gettext("%-16s %-14s\t%s\n"); (void) fprintf(fp, gettext("Usage: %s\n"), get_usage(un ? HELP_UNALLOW : HELP_ALLOW)); (void) fprintf(fp, gettext("Options:\n")); for (i = 0; i < (un ? unallow_size : allow_size); i += 2) { const char *opt = opt_desc[i]; const char *optdsc = opt_desc[i + 1]; (void) fprintf(fp, gettext(" %-10s %s\n"), opt, optdsc); } (void) fprintf(fp, gettext("\nThe following permissions are " "supported:\n\n")); (void) fprintf(fp, fmt, gettext("NAME"), gettext("TYPE"), gettext("NOTES")); for (i = 0; i < ZFS_NUM_DELEG_NOTES; i++) { const char *perm_name = zfs_deleg_perm_tbl[i].z_perm; zfs_deleg_note_t perm_note = zfs_deleg_perm_tbl[i].z_note; const char *perm_type = deleg_perm_type(perm_note); const char *perm_comment = deleg_perm_comment(perm_note); (void) fprintf(fp, fmt, perm_name, perm_type, perm_comment); } for (i = 0; i < ZFS_NUM_PROPS; i++) { zprop_desc_t *pd = &pdtbl[i]; if (pd->pd_visible != B_TRUE) continue; if (pd->pd_attr == PROP_READONLY) continue; props[count++] = pd->pd_name; } props[count] = NULL; qsort(props, count, sizeof (char *), prop_cmp); for (i = 0; i < count; i++) (void) fprintf(fp, fmt, props[i], gettext("property"), ""); if (msg != NULL) (void) fprintf(fp, gettext("\nzfs: error: %s"), msg); exit(requested ? 0 : 2); } static inline const char * munge_args(int argc, char **argv, boolean_t un, size_t expected_argc, char **permsp) { if (un && argc == expected_argc - 1) *permsp = NULL; else if (argc == expected_argc) *permsp = argv[argc - 2]; else allow_usage(un, B_FALSE, gettext("wrong number of parameters\n")); return (argv[argc - 1]); } static void parse_allow_args(int argc, char **argv, boolean_t un, struct allow_opts *opts) { int uge_sum = opts->user + opts->group + opts->everyone; int csuge_sum = opts->create + opts->set + uge_sum; int ldcsuge_sum = csuge_sum + opts->local + opts->descend; int all_sum = un ? ldcsuge_sum + opts->recursive : ldcsuge_sum; if (uge_sum > 1) allow_usage(un, B_FALSE, gettext("-u, -g, and -e are mutually exclusive\n")); if (opts->prt_usage) { if (argc == 0 && all_sum == 0) allow_usage(un, B_TRUE, NULL); else usage(B_FALSE); } if (opts->set) { if (csuge_sum > 1) allow_usage(un, B_FALSE, gettext("invalid options combined with -s\n")); opts->dataset = munge_args(argc, argv, un, 3, &opts->perms); if (argv[0][0] != '@') allow_usage(un, B_FALSE, gettext("invalid set name: missing '@' prefix\n")); opts->who = argv[0]; } else if (opts->create) { if (ldcsuge_sum > 1) allow_usage(un, B_FALSE, gettext("invalid options combined with -c\n")); opts->dataset = munge_args(argc, argv, un, 2, &opts->perms); } else if (opts->everyone) { if (csuge_sum > 1) allow_usage(un, B_FALSE, gettext("invalid options combined with -e\n")); opts->dataset = munge_args(argc, argv, un, 2, &opts->perms); } else if (uge_sum == 0 && argc > 0 && strcmp(argv[0], "everyone") == 0) { opts->everyone = B_TRUE; argc--; argv++; opts->dataset = munge_args(argc, argv, un, 2, &opts->perms); } else if (argc == 1 && !un) { opts->prt_perms = B_TRUE; opts->dataset = argv[argc-1]; } else { opts->dataset = munge_args(argc, argv, un, 3, &opts->perms); opts->who = argv[0]; } if (!opts->local && !opts->descend) { opts->local = B_TRUE; opts->descend = B_TRUE; } } static void store_allow_perm(zfs_deleg_who_type_t type, boolean_t local, boolean_t descend, const char *who, char *perms, nvlist_t *top_nvl) { int i; char ld[2] = { '\0', '\0' }; char who_buf[MAXNAMELEN + 32]; char base_type = '\0'; char set_type = '\0'; nvlist_t *base_nvl = NULL; nvlist_t *set_nvl = NULL; nvlist_t *nvl; if (nvlist_alloc(&base_nvl, NV_UNIQUE_NAME, 0) != 0) nomem(); if (nvlist_alloc(&set_nvl, NV_UNIQUE_NAME, 0) != 0) nomem(); switch (type) { case ZFS_DELEG_NAMED_SET_SETS: case ZFS_DELEG_NAMED_SET: set_type = ZFS_DELEG_NAMED_SET_SETS; base_type = ZFS_DELEG_NAMED_SET; ld[0] = ZFS_DELEG_NA; break; case ZFS_DELEG_CREATE_SETS: case ZFS_DELEG_CREATE: set_type = ZFS_DELEG_CREATE_SETS; base_type = ZFS_DELEG_CREATE; ld[0] = ZFS_DELEG_NA; break; case ZFS_DELEG_USER_SETS: case ZFS_DELEG_USER: set_type = ZFS_DELEG_USER_SETS; base_type = ZFS_DELEG_USER; if (local) ld[0] = ZFS_DELEG_LOCAL; if (descend) ld[1] = ZFS_DELEG_DESCENDENT; break; case ZFS_DELEG_GROUP_SETS: case ZFS_DELEG_GROUP: set_type = ZFS_DELEG_GROUP_SETS; base_type = ZFS_DELEG_GROUP; if (local) ld[0] = ZFS_DELEG_LOCAL; if (descend) ld[1] = ZFS_DELEG_DESCENDENT; break; case ZFS_DELEG_EVERYONE_SETS: case ZFS_DELEG_EVERYONE: set_type = ZFS_DELEG_EVERYONE_SETS; base_type = ZFS_DELEG_EVERYONE; if (local) ld[0] = ZFS_DELEG_LOCAL; if (descend) ld[1] = ZFS_DELEG_DESCENDENT; break; default: assert(set_type != '\0' && base_type != '\0'); } if (perms != NULL) { char *curr = perms; char *end = curr + strlen(perms); while (curr < end) { char *delim = strchr(curr, ','); if (delim == NULL) delim = end; else *delim = '\0'; if (curr[0] == '@') nvl = set_nvl; else nvl = base_nvl; (void) nvlist_add_boolean(nvl, curr); if (delim != end) *delim = ','; curr = delim + 1; } for (i = 0; i < 2; i++) { char locality = ld[i]; if (locality == 0) continue; if (!nvlist_empty(base_nvl)) { if (who != NULL) (void) snprintf(who_buf, sizeof (who_buf), "%c%c$%s", base_type, locality, who); else (void) snprintf(who_buf, sizeof (who_buf), "%c%c$", base_type, locality); (void) nvlist_add_nvlist(top_nvl, who_buf, base_nvl); } if (!nvlist_empty(set_nvl)) { if (who != NULL) (void) snprintf(who_buf, sizeof (who_buf), "%c%c$%s", set_type, locality, who); else (void) snprintf(who_buf, sizeof (who_buf), "%c%c$", set_type, locality); (void) nvlist_add_nvlist(top_nvl, who_buf, set_nvl); } } } else { for (i = 0; i < 2; i++) { char locality = ld[i]; if (locality == 0) continue; if (who != NULL) (void) snprintf(who_buf, sizeof (who_buf), "%c%c$%s", base_type, locality, who); else (void) snprintf(who_buf, sizeof (who_buf), "%c%c$", base_type, locality); (void) nvlist_add_boolean(top_nvl, who_buf); if (who != NULL) (void) snprintf(who_buf, sizeof (who_buf), "%c%c$%s", set_type, locality, who); else (void) snprintf(who_buf, sizeof (who_buf), "%c%c$", set_type, locality); (void) nvlist_add_boolean(top_nvl, who_buf); } } } static int construct_fsacl_list(boolean_t un, struct allow_opts *opts, nvlist_t **nvlp) { if (nvlist_alloc(nvlp, NV_UNIQUE_NAME, 0) != 0) nomem(); if (opts->set) { store_allow_perm(ZFS_DELEG_NAMED_SET, opts->local, opts->descend, opts->who, opts->perms, *nvlp); } else if (opts->create) { store_allow_perm(ZFS_DELEG_CREATE, opts->local, opts->descend, NULL, opts->perms, *nvlp); } else if (opts->everyone) { store_allow_perm(ZFS_DELEG_EVERYONE, opts->local, opts->descend, NULL, opts->perms, *nvlp); } else { char *curr = opts->who; char *end = curr + strlen(curr); while (curr < end) { const char *who; zfs_deleg_who_type_t who_type = ZFS_DELEG_WHO_UNKNOWN; char *endch; char *delim = strchr(curr, ','); char errbuf[256]; char id[64]; struct passwd *p = NULL; struct group *g = NULL; uid_t rid; if (delim == NULL) delim = end; else *delim = '\0'; rid = (uid_t)strtol(curr, &endch, 0); if (opts->user) { who_type = ZFS_DELEG_USER; if (*endch != '\0') p = getpwnam(curr); else p = getpwuid(rid); if (p != NULL) rid = p->pw_uid; else if (*endch != '\0') { (void) snprintf(errbuf, 256, gettext( "invalid user %s\n"), curr); allow_usage(un, B_TRUE, errbuf); } } else if (opts->group) { who_type = ZFS_DELEG_GROUP; if (*endch != '\0') g = getgrnam(curr); else g = getgrgid(rid); if (g != NULL) rid = g->gr_gid; else if (*endch != '\0') { (void) snprintf(errbuf, 256, gettext( "invalid group %s\n"), curr); allow_usage(un, B_TRUE, errbuf); } } else { if (*endch != '\0') { p = getpwnam(curr); } else { p = getpwuid(rid); } if (p == NULL) { if (*endch != '\0') { g = getgrnam(curr); } else { g = getgrgid(rid); } } if (p != NULL) { who_type = ZFS_DELEG_USER; rid = p->pw_uid; } else if (g != NULL) { who_type = ZFS_DELEG_GROUP; rid = g->gr_gid; } else { (void) snprintf(errbuf, 256, gettext( "invalid user/group %s\n"), curr); allow_usage(un, B_TRUE, errbuf); } } (void) sprintf(id, "%u", rid); who = id; store_allow_perm(who_type, opts->local, opts->descend, who, opts->perms, *nvlp); curr = delim + 1; } } return (0); } static void print_set_creat_perms(uu_avl_t *who_avl) { const char *sc_title[] = { gettext("Permission sets:\n"), gettext("Create time permissions:\n"), NULL }; who_perm_node_t *who_node = NULL; int prev_weight = -1; for (who_node = uu_avl_first(who_avl); who_node != NULL; who_node = uu_avl_next(who_avl, who_node)) { uu_avl_t *avl = who_node->who_perm.who_deleg_perm_avl; zfs_deleg_who_type_t who_type = who_node->who_perm.who_type; const char *who_name = who_node->who_perm.who_name; int weight = who_type2weight(who_type); boolean_t first = B_TRUE; deleg_perm_node_t *deleg_node; if (prev_weight != weight) { (void) printf("%s", sc_title[weight]); prev_weight = weight; } if (who_name == NULL || strnlen(who_name, 1) == 0) (void) printf("\t"); else (void) printf("\t%s ", who_name); for (deleg_node = uu_avl_first(avl); deleg_node != NULL; deleg_node = uu_avl_next(avl, deleg_node)) { if (first) { (void) printf("%s", deleg_node->dpn_perm.dp_name); first = B_FALSE; } else (void) printf(",%s", deleg_node->dpn_perm.dp_name); } (void) printf("\n"); } } static void print_uge_deleg_perms(uu_avl_t *who_avl, boolean_t local, boolean_t descend, const char *title) { who_perm_node_t *who_node = NULL; boolean_t prt_title = B_TRUE; uu_avl_walk_t *walk; if ((walk = uu_avl_walk_start(who_avl, UU_WALK_ROBUST)) == NULL) nomem(); while ((who_node = uu_avl_walk_next(walk)) != NULL) { const char *who_name = who_node->who_perm.who_name; const char *nice_who_name = who_node->who_perm.who_ug_name; uu_avl_t *avl = who_node->who_perm.who_deleg_perm_avl; zfs_deleg_who_type_t who_type = who_node->who_perm.who_type; char delim = ' '; deleg_perm_node_t *deleg_node; boolean_t prt_who = B_TRUE; for (deleg_node = uu_avl_first(avl); deleg_node != NULL; deleg_node = uu_avl_next(avl, deleg_node)) { if (local != deleg_node->dpn_perm.dp_local || descend != deleg_node->dpn_perm.dp_descend) continue; if (prt_who) { const char *who = NULL; if (prt_title) { prt_title = B_FALSE; (void) printf("%s", title); } switch (who_type) { case ZFS_DELEG_USER_SETS: case ZFS_DELEG_USER: who = gettext("user"); if (nice_who_name) who_name = nice_who_name; break; case ZFS_DELEG_GROUP_SETS: case ZFS_DELEG_GROUP: who = gettext("group"); if (nice_who_name) who_name = nice_who_name; break; case ZFS_DELEG_EVERYONE_SETS: case ZFS_DELEG_EVERYONE: who = gettext("everyone"); who_name = NULL; break; default: assert(who != NULL); } prt_who = B_FALSE; if (who_name == NULL) (void) printf("\t%s", who); else (void) printf("\t%s %s", who, who_name); } (void) printf("%c%s", delim, deleg_node->dpn_perm.dp_name); delim = ','; } if (!prt_who) (void) printf("\n"); } uu_avl_walk_end(walk); } static void print_fs_perms(fs_perm_set_t *fspset) { fs_perm_node_t *node = NULL; char buf[MAXNAMELEN + 32]; const char *dsname = buf; for (node = uu_list_first(fspset->fsps_list); node != NULL; node = uu_list_next(fspset->fsps_list, node)) { uu_avl_t *sc_avl = node->fspn_fsperm.fsp_sc_avl; uu_avl_t *uge_avl = node->fspn_fsperm.fsp_uge_avl; int left = 0; (void) snprintf(buf, sizeof (buf), gettext("---- Permissions on %s "), node->fspn_fsperm.fsp_name); (void) printf("%s", dsname); left = 70 - strlen(buf); while (left-- > 0) (void) printf("-"); (void) printf("\n"); print_set_creat_perms(sc_avl); print_uge_deleg_perms(uge_avl, B_TRUE, B_FALSE, gettext("Local permissions:\n")); print_uge_deleg_perms(uge_avl, B_FALSE, B_TRUE, gettext("Descendent permissions:\n")); print_uge_deleg_perms(uge_avl, B_TRUE, B_TRUE, gettext("Local+Descendent permissions:\n")); } } static fs_perm_set_t fs_perm_set = { NULL, NULL, NULL, NULL }; struct deleg_perms { boolean_t un; nvlist_t *nvl; }; static int set_deleg_perms(zfs_handle_t *zhp, void *data) { struct deleg_perms *perms = (struct deleg_perms *)data; zfs_type_t zfs_type = zfs_get_type(zhp); if (zfs_type != ZFS_TYPE_FILESYSTEM && zfs_type != ZFS_TYPE_VOLUME) return (0); return (zfs_set_fsacl(zhp, perms->un, perms->nvl)); } static int zfs_do_allow_unallow_impl(int argc, char **argv, boolean_t un) { zfs_handle_t *zhp; nvlist_t *perm_nvl = NULL; nvlist_t *update_perm_nvl = NULL; int error = 1; int c; struct allow_opts opts = { 0 }; const char *optstr = un ? "ldugecsrh" : "ldugecsh"; /* check opts */ while ((c = getopt(argc, argv, optstr)) != -1) { switch (c) { case 'l': opts.local = B_TRUE; break; case 'd': opts.descend = B_TRUE; break; case 'u': opts.user = B_TRUE; break; case 'g': opts.group = B_TRUE; break; case 'e': opts.everyone = B_TRUE; break; case 's': opts.set = B_TRUE; break; case 'c': opts.create = B_TRUE; break; case 'r': opts.recursive = B_TRUE; break; case ':': (void) fprintf(stderr, gettext("missing argument for " "'%c' option\n"), optopt); usage(B_FALSE); break; case 'h': opts.prt_usage = B_TRUE; break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; /* check arguments */ parse_allow_args(argc, argv, un, &opts); /* try to open the dataset */ if ((zhp = zfs_open(g_zfs, opts.dataset, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME)) == NULL) { (void) fprintf(stderr, "Failed to open dataset: %s\n", opts.dataset); return (-1); } if (zfs_get_fsacl(zhp, &perm_nvl) != 0) goto cleanup2; fs_perm_set_init(&fs_perm_set); if (parse_fs_perm_set(&fs_perm_set, perm_nvl) != 0) { (void) fprintf(stderr, "Failed to parse fsacl permissions\n"); goto cleanup1; } if (opts.prt_perms) print_fs_perms(&fs_perm_set); else { (void) construct_fsacl_list(un, &opts, &update_perm_nvl); if (zfs_set_fsacl(zhp, un, update_perm_nvl) != 0) goto cleanup0; if (un && opts.recursive) { struct deleg_perms data = { un, update_perm_nvl }; if (zfs_iter_filesystems(zhp, set_deleg_perms, &data) != 0) goto cleanup0; } } error = 0; cleanup0: nvlist_free(perm_nvl); nvlist_free(update_perm_nvl); cleanup1: fs_perm_set_fini(&fs_perm_set); cleanup2: zfs_close(zhp); return (error); } static int zfs_do_allow(int argc, char **argv) { return (zfs_do_allow_unallow_impl(argc, argv, B_FALSE)); } static int zfs_do_unallow(int argc, char **argv) { return (zfs_do_allow_unallow_impl(argc, argv, B_TRUE)); } static int zfs_do_hold_rele_impl(int argc, char **argv, boolean_t holding) { int errors = 0; int i; const char *tag; boolean_t recursive = B_FALSE; const char *opts = holding ? "rt" : "r"; int c; /* check options */ while ((c = getopt(argc, argv, opts)) != -1) { switch (c) { case 'r': recursive = B_TRUE; break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; /* check number of arguments */ if (argc < 2) usage(B_FALSE); tag = argv[0]; --argc; ++argv; if (holding && tag[0] == '.') { /* tags starting with '.' are reserved for libzfs */ (void) fprintf(stderr, gettext("tag may not start with '.'\n")); usage(B_FALSE); } for (i = 0; i < argc; ++i) { zfs_handle_t *zhp; char parent[ZFS_MAX_DATASET_NAME_LEN]; const char *delim; char *path = argv[i]; delim = strchr(path, '@'); if (delim == NULL) { (void) fprintf(stderr, gettext("'%s' is not a snapshot\n"), path); ++errors; continue; } (void) strncpy(parent, path, delim - path); parent[delim - path] = '\0'; zhp = zfs_open(g_zfs, parent, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME); if (zhp == NULL) { ++errors; continue; } if (holding) { if (zfs_hold(zhp, delim+1, tag, recursive, -1) != 0) ++errors; } else { if (zfs_release(zhp, delim+1, tag, recursive) != 0) ++errors; } zfs_close(zhp); } return (errors != 0); } /* * zfs hold [-r] [-t] ... * * -r Recursively hold * * Apply a user-hold with the given tag to the list of snapshots. */ static int zfs_do_hold(int argc, char **argv) { return (zfs_do_hold_rele_impl(argc, argv, B_TRUE)); } /* * zfs release [-r] ... * * -r Recursively release * * Release a user-hold with the given tag from the list of snapshots. */ static int zfs_do_release(int argc, char **argv) { return (zfs_do_hold_rele_impl(argc, argv, B_FALSE)); } typedef struct holds_cbdata { boolean_t cb_recursive; const char *cb_snapname; nvlist_t **cb_nvlp; size_t cb_max_namelen; size_t cb_max_taglen; } holds_cbdata_t; #define STRFTIME_FMT_STR "%a %b %e %H:%M %Y" #define DATETIME_BUF_LEN (32) /* * */ static void print_holds(boolean_t scripted, int nwidth, int tagwidth, nvlist_t *nvl) { int i; nvpair_t *nvp = NULL; char *hdr_cols[] = { "NAME", "TAG", "TIMESTAMP" }; const char *col; if (!scripted) { for (i = 0; i < 3; i++) { col = gettext(hdr_cols[i]); if (i < 2) (void) printf("%-*s ", i ? tagwidth : nwidth, col); else (void) printf("%s\n", col); } } while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) { char *zname = nvpair_name(nvp); nvlist_t *nvl2; nvpair_t *nvp2 = NULL; (void) nvpair_value_nvlist(nvp, &nvl2); while ((nvp2 = nvlist_next_nvpair(nvl2, nvp2)) != NULL) { char tsbuf[DATETIME_BUF_LEN]; char *tagname = nvpair_name(nvp2); uint64_t val = 0; time_t time; struct tm t; (void) nvpair_value_uint64(nvp2, &val); time = (time_t)val; (void) localtime_r(&time, &t); (void) strftime(tsbuf, DATETIME_BUF_LEN, gettext(STRFTIME_FMT_STR), &t); if (scripted) { (void) printf("%s\t%s\t%s\n", zname, tagname, tsbuf); } else { (void) printf("%-*s %-*s %s\n", nwidth, zname, tagwidth, tagname, tsbuf); } } } } /* * Generic callback function to list a dataset or snapshot. */ static int holds_callback(zfs_handle_t *zhp, void *data) { holds_cbdata_t *cbp = data; nvlist_t *top_nvl = *cbp->cb_nvlp; nvlist_t *nvl = NULL; nvpair_t *nvp = NULL; const char *zname = zfs_get_name(zhp); size_t znamelen = strlen(zname); if (cbp->cb_recursive) { const char *snapname; char *delim = strchr(zname, '@'); if (delim == NULL) return (0); snapname = delim + 1; if (strcmp(cbp->cb_snapname, snapname)) return (0); } if (zfs_get_holds(zhp, &nvl) != 0) return (-1); if (znamelen > cbp->cb_max_namelen) cbp->cb_max_namelen = znamelen; while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) { const char *tag = nvpair_name(nvp); size_t taglen = strlen(tag); if (taglen > cbp->cb_max_taglen) cbp->cb_max_taglen = taglen; } return (nvlist_add_nvlist(top_nvl, zname, nvl)); } /* * zfs holds [-rH] ... * * -r Lists holds that are set on the named snapshots recursively. * -H Scripted mode; elide headers and separate columns by tabs. */ static int zfs_do_holds(int argc, char **argv) { int errors = 0; int c; int i; boolean_t scripted = B_FALSE; boolean_t recursive = B_FALSE; const char *opts = "rH"; nvlist_t *nvl; int types = ZFS_TYPE_SNAPSHOT; holds_cbdata_t cb = { 0 }; int limit = 0; int ret = 0; int flags = 0; /* check options */ while ((c = getopt(argc, argv, opts)) != -1) { switch (c) { case 'r': recursive = B_TRUE; break; case 'H': scripted = B_TRUE; break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } if (recursive) { types |= ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME; flags |= ZFS_ITER_RECURSE; } argc -= optind; argv += optind; /* check number of arguments */ if (argc < 1) usage(B_FALSE); if (nvlist_alloc(&nvl, NV_UNIQUE_NAME, 0) != 0) nomem(); for (i = 0; i < argc; ++i) { char *snapshot = argv[i]; const char *delim; const char *snapname; delim = strchr(snapshot, '@'); if (delim == NULL) { (void) fprintf(stderr, gettext("'%s' is not a snapshot\n"), snapshot); ++errors; continue; } snapname = delim + 1; if (recursive) snapshot[delim - snapshot] = '\0'; cb.cb_recursive = recursive; cb.cb_snapname = snapname; cb.cb_nvlp = &nvl; /* * 1. collect holds data, set format options */ ret = zfs_for_each(argc, argv, flags, types, NULL, NULL, limit, holds_callback, &cb); if (ret != 0) ++errors; } /* * 2. print holds data */ print_holds(scripted, cb.cb_max_namelen, cb.cb_max_taglen, nvl); if (nvlist_empty(nvl)) (void) fprintf(stderr, gettext("no datasets available\n")); nvlist_free(nvl); return (0 != errors); } #define CHECK_SPINNER 30 #define SPINNER_TIME 3 /* seconds */ #define MOUNT_TIME 1 /* seconds */ typedef struct get_all_state { boolean_t ga_verbose; get_all_cb_t *ga_cbp; } get_all_state_t; static int get_one_dataset(zfs_handle_t *zhp, void *data) { static char *spin[] = { "-", "\\", "|", "/" }; static int spinval = 0; static int spincheck = 0; static time_t last_spin_time = (time_t)0; get_all_state_t *state = data; zfs_type_t type = zfs_get_type(zhp); if (state->ga_verbose) { if (--spincheck < 0) { time_t now = time(NULL); if (last_spin_time + SPINNER_TIME < now) { update_progress(spin[spinval++ % 4]); last_spin_time = now; } spincheck = CHECK_SPINNER; } } /* * Iterate over any nested datasets. */ if (zfs_iter_filesystems(zhp, get_one_dataset, data) != 0) { zfs_close(zhp); return (1); } /* * Skip any datasets whose type does not match. */ if ((type & ZFS_TYPE_FILESYSTEM) == 0) { zfs_close(zhp); return (0); } libzfs_add_handle(state->ga_cbp, zhp); assert(state->ga_cbp->cb_used <= state->ga_cbp->cb_alloc); return (0); } static void get_all_datasets(get_all_cb_t *cbp, boolean_t verbose) { get_all_state_t state = { .ga_verbose = verbose, .ga_cbp = cbp }; if (verbose) set_progress_header(gettext("Reading ZFS config")); (void) zfs_iter_root(g_zfs, get_one_dataset, &state); if (verbose) finish_progress(gettext("done.")); } /* * Generic callback for sharing or mounting filesystems. Because the code is so * similar, we have a common function with an extra parameter to determine which * mode we are using. */ typedef enum { OP_SHARE, OP_MOUNT } share_mount_op_t; typedef struct share_mount_state { share_mount_op_t sm_op; boolean_t sm_verbose; int sm_flags; char *sm_options; char *sm_proto; /* only valid for OP_SHARE */ pthread_mutex_t sm_lock; /* protects the remaining fields */ uint_t sm_total; /* number of filesystems to process */ uint_t sm_done; /* number of filesystems processed */ int sm_status; /* -1 if any of the share/mount operations failed */ } share_mount_state_t; /* * Share or mount a dataset. */ static int share_mount_one(zfs_handle_t *zhp, int op, int flags, char *protocol, boolean_t explicit, const char *options) { char mountpoint[ZFS_MAXPROPLEN]; char shareopts[ZFS_MAXPROPLEN]; char smbshareopts[ZFS_MAXPROPLEN]; const char *cmdname = op == OP_SHARE ? "share" : "mount"; struct mnttab mnt; uint64_t zoned, canmount; boolean_t shared_nfs, shared_smb; assert(zfs_get_type(zhp) & ZFS_TYPE_FILESYSTEM); /* * Check to make sure we can mount/share this dataset. If we * are in the global zone and the filesystem is exported to a * local zone, or if we are in a local zone and the * filesystem is not exported, then it is an error. */ zoned = zfs_prop_get_int(zhp, ZFS_PROP_ZONED); if (zoned && getzoneid() == GLOBAL_ZONEID) { if (!explicit) return (0); (void) fprintf(stderr, gettext("cannot %s '%s': " "dataset is exported to a local zone\n"), cmdname, zfs_get_name(zhp)); return (1); } else if (!zoned && getzoneid() != GLOBAL_ZONEID) { if (!explicit) return (0); (void) fprintf(stderr, gettext("cannot %s '%s': " "permission denied\n"), cmdname, zfs_get_name(zhp)); return (1); } /* * Ignore any filesystems which don't apply to us. This * includes those with a legacy mountpoint, or those with * legacy share options. */ verify(zfs_prop_get(zhp, ZFS_PROP_MOUNTPOINT, mountpoint, sizeof (mountpoint), NULL, NULL, 0, B_FALSE) == 0); verify(zfs_prop_get(zhp, ZFS_PROP_SHARENFS, shareopts, sizeof (shareopts), NULL, NULL, 0, B_FALSE) == 0); verify(zfs_prop_get(zhp, ZFS_PROP_SHARESMB, smbshareopts, sizeof (smbshareopts), NULL, NULL, 0, B_FALSE) == 0); if (op == OP_SHARE && strcmp(shareopts, "off") == 0 && strcmp(smbshareopts, "off") == 0) { if (!explicit) return (0); (void) fprintf(stderr, gettext("cannot share '%s': " "legacy share\n"), zfs_get_name(zhp)); - (void) fprintf(stderr, gettext("use share(1M) to " - "share this filesystem, or set " - "sharenfs property on\n")); + (void) fprintf(stderr, gettext("use exports(5) or " + "smb.conf(5) to share this filesystem, or set " + "the sharenfs or sharesmb property\n")); return (1); } /* * We cannot share or mount legacy filesystems. If the * shareopts is non-legacy but the mountpoint is legacy, we * treat it as a legacy share. */ if (strcmp(mountpoint, "legacy") == 0) { if (!explicit) return (0); (void) fprintf(stderr, gettext("cannot %s '%s': " "legacy mountpoint\n"), cmdname, zfs_get_name(zhp)); - (void) fprintf(stderr, gettext("use %s(1M) to " + (void) fprintf(stderr, gettext("use %s(8) to " "%s this filesystem\n"), cmdname, cmdname); return (1); } if (strcmp(mountpoint, "none") == 0) { if (!explicit) return (0); (void) fprintf(stderr, gettext("cannot %s '%s': no " "mountpoint set\n"), cmdname, zfs_get_name(zhp)); return (1); } /* * canmount explicit outcome * on no pass through * on yes pass through * off no return 0 * off yes display error, return 1 * noauto no return 0 * noauto yes pass through */ canmount = zfs_prop_get_int(zhp, ZFS_PROP_CANMOUNT); if (canmount == ZFS_CANMOUNT_OFF) { if (!explicit) return (0); (void) fprintf(stderr, gettext("cannot %s '%s': " "'canmount' property is set to 'off'\n"), cmdname, zfs_get_name(zhp)); return (1); } else if (canmount == ZFS_CANMOUNT_NOAUTO && !explicit) { /* * When performing a 'zfs mount -a', we skip any mounts for * datasets that have 'noauto' set. Sharing a dataset with * 'noauto' set is only allowed if it's mounted. */ if (op == OP_MOUNT) return (0); if (op == OP_SHARE && !zfs_is_mounted(zhp, NULL)) { /* also purge it from existing exports */ zfs_unshareall_bypath(zhp, mountpoint); return (0); } } /* * If this filesystem is encrypted and does not have * a loaded key, we can not mount it. */ if ((flags & MS_CRYPT) == 0 && zfs_prop_get_int(zhp, ZFS_PROP_ENCRYPTION) != ZIO_CRYPT_OFF && zfs_prop_get_int(zhp, ZFS_PROP_KEYSTATUS) == ZFS_KEYSTATUS_UNAVAILABLE) { if (!explicit) return (0); (void) fprintf(stderr, gettext("cannot %s '%s': " "encryption key not loaded\n"), cmdname, zfs_get_name(zhp)); return (1); } /* * If this filesystem is inconsistent and has a receive resume * token, we can not mount it. */ if (zfs_prop_get_int(zhp, ZFS_PROP_INCONSISTENT) && zfs_prop_get(zhp, ZFS_PROP_RECEIVE_RESUME_TOKEN, NULL, 0, NULL, NULL, 0, B_TRUE) == 0) { if (!explicit) return (0); (void) fprintf(stderr, gettext("cannot %s '%s': " "Contains partially-completed state from " "\"zfs receive -s\", which can be resumed with " "\"zfs send -t\"\n"), cmdname, zfs_get_name(zhp)); return (1); } if (zfs_prop_get_int(zhp, ZFS_PROP_REDACTED) && !(flags & MS_FORCE)) { if (!explicit) return (0); (void) fprintf(stderr, gettext("cannot %s '%s': " "Dataset is not complete, was created by receiving " "a redacted zfs send stream.\n"), cmdname, zfs_get_name(zhp)); return (1); } /* * At this point, we have verified that the mountpoint and/or * shareopts are appropriate for auto management. If the * filesystem is already mounted or shared, return (failing * for explicit requests); otherwise mount or share the * filesystem. */ switch (op) { case OP_SHARE: shared_nfs = zfs_is_shared_nfs(zhp, NULL); shared_smb = zfs_is_shared_smb(zhp, NULL); if ((shared_nfs && shared_smb) || (shared_nfs && strcmp(shareopts, "on") == 0 && strcmp(smbshareopts, "off") == 0) || (shared_smb && strcmp(smbshareopts, "on") == 0 && strcmp(shareopts, "off") == 0)) { if (!explicit) return (0); (void) fprintf(stderr, gettext("cannot share " "'%s': filesystem already shared\n"), zfs_get_name(zhp)); return (1); } if (!zfs_is_mounted(zhp, NULL) && zfs_mount(zhp, NULL, flags) != 0) return (1); if (protocol == NULL) { if (zfs_shareall(zhp) != 0) return (1); } else if (strcmp(protocol, "nfs") == 0) { if (zfs_share_nfs(zhp)) return (1); } else if (strcmp(protocol, "smb") == 0) { if (zfs_share_smb(zhp)) return (1); } else { (void) fprintf(stderr, gettext("cannot share " "'%s': invalid share type '%s' " "specified\n"), zfs_get_name(zhp), protocol); return (1); } break; case OP_MOUNT: if (options == NULL) mnt.mnt_mntopts = ""; else mnt.mnt_mntopts = (char *)options; if (!hasmntopt(&mnt, MNTOPT_REMOUNT) && zfs_is_mounted(zhp, NULL)) { if (!explicit) return (0); (void) fprintf(stderr, gettext("cannot mount " "'%s': filesystem already mounted\n"), zfs_get_name(zhp)); return (1); } if (zfs_mount(zhp, options, flags) != 0) return (1); break; } return (0); } /* * Reports progress in the form "(current/total)". Not thread-safe. */ static void report_mount_progress(int current, int total) { static time_t last_progress_time = 0; time_t now = time(NULL); char info[32]; /* report 1..n instead of 0..n-1 */ ++current; /* display header if we're here for the first time */ if (current == 1) { set_progress_header(gettext("Mounting ZFS filesystems")); } else if (current != total && last_progress_time + MOUNT_TIME >= now) { /* too soon to report again */ return; } last_progress_time = now; (void) sprintf(info, "(%d/%d)", current, total); if (current == total) finish_progress(info); else update_progress(info); } /* * zfs_foreach_mountpoint() callback that mounts or shares one filesystem and * updates the progress meter. */ static int share_mount_one_cb(zfs_handle_t *zhp, void *arg) { share_mount_state_t *sms = arg; int ret; ret = share_mount_one(zhp, sms->sm_op, sms->sm_flags, sms->sm_proto, B_FALSE, sms->sm_options); pthread_mutex_lock(&sms->sm_lock); if (ret != 0) sms->sm_status = ret; sms->sm_done++; if (sms->sm_verbose) report_mount_progress(sms->sm_done, sms->sm_total); pthread_mutex_unlock(&sms->sm_lock); return (ret); } static void append_options(char *mntopts, char *newopts) { int len = strlen(mntopts); /* original length plus new string to append plus 1 for the comma */ if (len + 1 + strlen(newopts) >= MNT_LINE_MAX) { (void) fprintf(stderr, gettext("the opts argument for " "'%s' option is too long (more than %d chars)\n"), "-o", MNT_LINE_MAX); usage(B_FALSE); } if (*mntopts) mntopts[len++] = ','; (void) strcpy(&mntopts[len], newopts); } static int share_mount(int op, int argc, char **argv) { int do_all = 0; boolean_t verbose = B_FALSE; int c, ret = 0; char *options = NULL; int flags = 0; /* check options */ while ((c = getopt(argc, argv, op == OP_MOUNT ? ":alvo:Of" : "al")) != -1) { switch (c) { case 'a': do_all = 1; break; case 'v': verbose = B_TRUE; break; case 'l': flags |= MS_CRYPT; break; case 'o': if (*optarg == '\0') { (void) fprintf(stderr, gettext("empty mount " "options (-o) specified\n")); usage(B_FALSE); } if (options == NULL) options = safe_malloc(MNT_LINE_MAX + 1); /* option validation is done later */ append_options(options, optarg); break; case 'O': flags |= MS_OVERLAY; break; case 'f': flags |= MS_FORCE; break; case ':': (void) fprintf(stderr, gettext("missing argument for " "'%c' option\n"), optopt); usage(B_FALSE); break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; /* check number of arguments */ if (do_all) { char *protocol = NULL; if (op == OP_SHARE && argc > 0) { if (strcmp(argv[0], "nfs") != 0 && strcmp(argv[0], "smb") != 0) { (void) fprintf(stderr, gettext("share type " "must be 'nfs' or 'smb'\n")); usage(B_FALSE); } protocol = argv[0]; argc--; argv++; } if (argc != 0) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } start_progress_timer(); get_all_cb_t cb = { 0 }; get_all_datasets(&cb, verbose); if (cb.cb_used == 0) { if (options != NULL) free(options); return (0); } share_mount_state_t share_mount_state = { 0 }; share_mount_state.sm_op = op; share_mount_state.sm_verbose = verbose; share_mount_state.sm_flags = flags; share_mount_state.sm_options = options; share_mount_state.sm_proto = protocol; share_mount_state.sm_total = cb.cb_used; pthread_mutex_init(&share_mount_state.sm_lock, NULL); /* * libshare isn't mt-safe, so only do the operation in parallel * if we're mounting. Additionally, the key-loading option must * be serialized so that we can prompt the user for their keys * in a consistent manner. */ zfs_foreach_mountpoint(g_zfs, cb.cb_handles, cb.cb_used, share_mount_one_cb, &share_mount_state, op == OP_MOUNT && !(flags & MS_CRYPT)); zfs_commit_all_shares(); ret = share_mount_state.sm_status; for (int i = 0; i < cb.cb_used; i++) zfs_close(cb.cb_handles[i]); free(cb.cb_handles); } else if (argc == 0) { struct mnttab entry; if ((op == OP_SHARE) || (options != NULL)) { (void) fprintf(stderr, gettext("missing filesystem " "argument (specify -a for all)\n")); usage(B_FALSE); } /* * When mount is given no arguments, go through * /proc/self/mounts and display any active ZFS mounts. * We hide any snapshots, since they are controlled * automatically. */ /* Reopen MNTTAB to prevent reading stale data from open file */ if (freopen(MNTTAB, "r", mnttab_file) == NULL) { if (options != NULL) free(options); return (ENOENT); } while (getmntent(mnttab_file, &entry) == 0) { if (strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0 || strchr(entry.mnt_special, '@') != NULL) continue; (void) printf("%-30s %s\n", entry.mnt_special, entry.mnt_mountp); } } else { zfs_handle_t *zhp; if (argc > 1) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } if ((zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_FILESYSTEM)) == NULL) { ret = 1; } else { ret = share_mount_one(zhp, op, flags, NULL, B_TRUE, options); zfs_commit_all_shares(); zfs_close(zhp); } } if (options != NULL) free(options); return (ret); } /* * zfs mount -a [nfs] * zfs mount filesystem * * Mount all filesystems, or mount the given filesystem. */ static int zfs_do_mount(int argc, char **argv) { return (share_mount(OP_MOUNT, argc, argv)); } /* * zfs share -a [nfs | smb] * zfs share filesystem * * Share all filesystems, or share the given filesystem. */ static int zfs_do_share(int argc, char **argv) { return (share_mount(OP_SHARE, argc, argv)); } typedef struct unshare_unmount_node { zfs_handle_t *un_zhp; char *un_mountp; uu_avl_node_t un_avlnode; } unshare_unmount_node_t; /* ARGSUSED */ static int unshare_unmount_compare(const void *larg, const void *rarg, void *unused) { const unshare_unmount_node_t *l = larg; const unshare_unmount_node_t *r = rarg; return (strcmp(l->un_mountp, r->un_mountp)); } /* * Convenience routine used by zfs_do_umount() and manual_unmount(). Given an * absolute path, find the entry /proc/self/mounts, verify that it's a * ZFS filesystem, and unmount it appropriately. */ static int unshare_unmount_path(int op, char *path, int flags, boolean_t is_manual) { zfs_handle_t *zhp; int ret = 0; struct stat64 statbuf; struct extmnttab entry; const char *cmdname = (op == OP_SHARE) ? "unshare" : "unmount"; ino_t path_inode; /* * Search for the given (major,minor) pair in the mount table. */ /* Reopen MNTTAB to prevent reading stale data from open file */ if (freopen(MNTTAB, "r", mnttab_file) == NULL) return (ENOENT); if (getextmntent(path, &entry, &statbuf) != 0) { if (op == OP_SHARE) { (void) fprintf(stderr, gettext("cannot %s '%s': not " "currently mounted\n"), cmdname, path); return (1); } (void) fprintf(stderr, gettext("warning: %s not in" "/proc/self/mounts\n"), path); if ((ret = umount2(path, flags)) != 0) (void) fprintf(stderr, gettext("%s: %s\n"), path, strerror(errno)); return (ret != 0); } path_inode = statbuf.st_ino; if (strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0) { (void) fprintf(stderr, gettext("cannot %s '%s': not a ZFS " "filesystem\n"), cmdname, path); return (1); } if ((zhp = zfs_open(g_zfs, entry.mnt_special, ZFS_TYPE_FILESYSTEM)) == NULL) return (1); ret = 1; if (stat64(entry.mnt_mountp, &statbuf) != 0) { (void) fprintf(stderr, gettext("cannot %s '%s': %s\n"), cmdname, path, strerror(errno)); goto out; } else if (statbuf.st_ino != path_inode) { (void) fprintf(stderr, gettext("cannot " "%s '%s': not a mountpoint\n"), cmdname, path); goto out; } if (op == OP_SHARE) { char nfs_mnt_prop[ZFS_MAXPROPLEN]; char smbshare_prop[ZFS_MAXPROPLEN]; verify(zfs_prop_get(zhp, ZFS_PROP_SHARENFS, nfs_mnt_prop, sizeof (nfs_mnt_prop), NULL, NULL, 0, B_FALSE) == 0); verify(zfs_prop_get(zhp, ZFS_PROP_SHARESMB, smbshare_prop, sizeof (smbshare_prop), NULL, NULL, 0, B_FALSE) == 0); if (strcmp(nfs_mnt_prop, "off") == 0 && strcmp(smbshare_prop, "off") == 0) { (void) fprintf(stderr, gettext("cannot unshare " "'%s': legacy share\n"), path); (void) fprintf(stderr, gettext("use exportfs(8) " "or smbcontrol(1) to unshare this filesystem\n")); } else if (!zfs_is_shared(zhp)) { (void) fprintf(stderr, gettext("cannot unshare '%s': " "not currently shared\n"), path); } else { ret = zfs_unshareall_bypath(zhp, path); zfs_commit_all_shares(); } } else { char mtpt_prop[ZFS_MAXPROPLEN]; verify(zfs_prop_get(zhp, ZFS_PROP_MOUNTPOINT, mtpt_prop, sizeof (mtpt_prop), NULL, NULL, 0, B_FALSE) == 0); if (is_manual) { ret = zfs_unmount(zhp, NULL, flags); } else if (strcmp(mtpt_prop, "legacy") == 0) { (void) fprintf(stderr, gettext("cannot unmount " "'%s': legacy mountpoint\n"), zfs_get_name(zhp)); (void) fprintf(stderr, gettext("use umount(8) " "to unmount this filesystem\n")); } else { ret = zfs_unmountall(zhp, flags); } } out: zfs_close(zhp); return (ret != 0); } /* * Generic callback for unsharing or unmounting a filesystem. */ static int unshare_unmount(int op, int argc, char **argv) { int do_all = 0; int flags = 0; int ret = 0; int c; zfs_handle_t *zhp; char nfs_mnt_prop[ZFS_MAXPROPLEN]; char sharesmb[ZFS_MAXPROPLEN]; /* check options */ while ((c = getopt(argc, argv, op == OP_SHARE ? ":a" : "afu")) != -1) { switch (c) { case 'a': do_all = 1; break; case 'f': flags |= MS_FORCE; break; case 'u': flags |= MS_CRYPT; break; case ':': (void) fprintf(stderr, gettext("missing argument for " "'%c' option\n"), optopt); usage(B_FALSE); break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; if (do_all) { /* * We could make use of zfs_for_each() to walk all datasets in * the system, but this would be very inefficient, especially * since we would have to linearly search /proc/self/mounts for * each one. Instead, do one pass through /proc/self/mounts * looking for zfs entries and call zfs_unmount() for each one. * * Things get a little tricky if the administrator has created * mountpoints beneath other ZFS filesystems. In this case, we * have to unmount the deepest filesystems first. To accomplish * this, we place all the mountpoints in an AVL tree sorted by * the special type (dataset name), and walk the result in * reverse to make sure to get any snapshots first. */ struct mnttab entry; uu_avl_pool_t *pool; uu_avl_t *tree = NULL; unshare_unmount_node_t *node; uu_avl_index_t idx; uu_avl_walk_t *walk; char *protocol = NULL; if (op == OP_SHARE && argc > 0) { if (strcmp(argv[0], "nfs") != 0 && strcmp(argv[0], "smb") != 0) { (void) fprintf(stderr, gettext("share type " "must be 'nfs' or 'smb'\n")); usage(B_FALSE); } protocol = argv[0]; argc--; argv++; } if (argc != 0) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } if (((pool = uu_avl_pool_create("unmount_pool", sizeof (unshare_unmount_node_t), offsetof(unshare_unmount_node_t, un_avlnode), unshare_unmount_compare, UU_DEFAULT)) == NULL) || ((tree = uu_avl_create(pool, NULL, UU_DEFAULT)) == NULL)) nomem(); /* Reopen MNTTAB to prevent reading stale data from open file */ if (freopen(MNTTAB, "r", mnttab_file) == NULL) return (ENOENT); while (getmntent(mnttab_file, &entry) == 0) { /* ignore non-ZFS entries */ if (strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0) continue; /* ignore snapshots */ if (strchr(entry.mnt_special, '@') != NULL) continue; if ((zhp = zfs_open(g_zfs, entry.mnt_special, ZFS_TYPE_FILESYSTEM)) == NULL) { ret = 1; continue; } /* * Ignore datasets that are excluded/restricted by * parent pool name. */ if (zpool_skip_pool(zfs_get_pool_name(zhp))) { zfs_close(zhp); continue; } switch (op) { case OP_SHARE: verify(zfs_prop_get(zhp, ZFS_PROP_SHARENFS, nfs_mnt_prop, sizeof (nfs_mnt_prop), NULL, NULL, 0, B_FALSE) == 0); if (strcmp(nfs_mnt_prop, "off") != 0) break; verify(zfs_prop_get(zhp, ZFS_PROP_SHARESMB, nfs_mnt_prop, sizeof (nfs_mnt_prop), NULL, NULL, 0, B_FALSE) == 0); if (strcmp(nfs_mnt_prop, "off") == 0) continue; break; case OP_MOUNT: /* Ignore legacy mounts */ verify(zfs_prop_get(zhp, ZFS_PROP_MOUNTPOINT, nfs_mnt_prop, sizeof (nfs_mnt_prop), NULL, NULL, 0, B_FALSE) == 0); if (strcmp(nfs_mnt_prop, "legacy") == 0) continue; /* Ignore canmount=noauto mounts */ if (zfs_prop_get_int(zhp, ZFS_PROP_CANMOUNT) == ZFS_CANMOUNT_NOAUTO) continue; default: break; } node = safe_malloc(sizeof (unshare_unmount_node_t)); node->un_zhp = zhp; node->un_mountp = safe_strdup(entry.mnt_mountp); uu_avl_node_init(node, &node->un_avlnode, pool); if (uu_avl_find(tree, node, NULL, &idx) == NULL) { uu_avl_insert(tree, node, idx); } else { zfs_close(node->un_zhp); free(node->un_mountp); free(node); } } /* * Walk the AVL tree in reverse, unmounting each filesystem and * removing it from the AVL tree in the process. */ if ((walk = uu_avl_walk_start(tree, UU_WALK_REVERSE | UU_WALK_ROBUST)) == NULL) nomem(); while ((node = uu_avl_walk_next(walk)) != NULL) { const char *mntarg = NULL; uu_avl_remove(tree, node); switch (op) { case OP_SHARE: if (zfs_unshareall_bytype(node->un_zhp, node->un_mountp, protocol) != 0) ret = 1; break; case OP_MOUNT: if (zfs_unmount(node->un_zhp, mntarg, flags) != 0) ret = 1; break; } zfs_close(node->un_zhp); free(node->un_mountp); free(node); } if (op == OP_SHARE) zfs_commit_shares(protocol); uu_avl_walk_end(walk); uu_avl_destroy(tree); uu_avl_pool_destroy(pool); } else { if (argc != 1) { if (argc == 0) (void) fprintf(stderr, gettext("missing filesystem argument\n")); else (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } /* * We have an argument, but it may be a full path or a ZFS * filesystem. Pass full paths off to unmount_path() (shared by * manual_unmount), otherwise open the filesystem and pass to * zfs_unmount(). */ if (argv[0][0] == '/') return (unshare_unmount_path(op, argv[0], flags, B_FALSE)); if ((zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_FILESYSTEM)) == NULL) return (1); verify(zfs_prop_get(zhp, op == OP_SHARE ? ZFS_PROP_SHARENFS : ZFS_PROP_MOUNTPOINT, nfs_mnt_prop, sizeof (nfs_mnt_prop), NULL, NULL, 0, B_FALSE) == 0); switch (op) { case OP_SHARE: verify(zfs_prop_get(zhp, ZFS_PROP_SHARENFS, nfs_mnt_prop, sizeof (nfs_mnt_prop), NULL, NULL, 0, B_FALSE) == 0); verify(zfs_prop_get(zhp, ZFS_PROP_SHARESMB, sharesmb, sizeof (sharesmb), NULL, NULL, 0, B_FALSE) == 0); if (strcmp(nfs_mnt_prop, "off") == 0 && strcmp(sharesmb, "off") == 0) { (void) fprintf(stderr, gettext("cannot " "unshare '%s': legacy share\n"), zfs_get_name(zhp)); (void) fprintf(stderr, gettext("use " - "unshare(1M) to unshare this " - "filesystem\n")); + "exports(5) or smb.conf(5) to unshare " + "this filesystem\n")); ret = 1; } else if (!zfs_is_shared(zhp)) { (void) fprintf(stderr, gettext("cannot " "unshare '%s': not currently " "shared\n"), zfs_get_name(zhp)); ret = 1; } else if (zfs_unshareall(zhp) != 0) { ret = 1; } break; case OP_MOUNT: if (strcmp(nfs_mnt_prop, "legacy") == 0) { (void) fprintf(stderr, gettext("cannot " "unmount '%s': legacy " "mountpoint\n"), zfs_get_name(zhp)); (void) fprintf(stderr, gettext("use " - "umount(1M) to unmount this " + "umount(8) to unmount this " "filesystem\n")); ret = 1; } else if (!zfs_is_mounted(zhp, NULL)) { (void) fprintf(stderr, gettext("cannot " "unmount '%s': not currently " "mounted\n"), zfs_get_name(zhp)); ret = 1; } else if (zfs_unmountall(zhp, flags) != 0) { ret = 1; } break; } zfs_close(zhp); } return (ret); } /* * zfs unmount [-fu] -a * zfs unmount [-fu] filesystem * * Unmount all filesystems, or a specific ZFS filesystem. */ static int zfs_do_unmount(int argc, char **argv) { return (unshare_unmount(OP_MOUNT, argc, argv)); } /* * zfs unshare -a * zfs unshare filesystem * * Unshare all filesystems, or a specific ZFS filesystem. */ static int zfs_do_unshare(int argc, char **argv) { return (unshare_unmount(OP_SHARE, argc, argv)); } static int find_command_idx(char *command, int *idx) { int i; for (i = 0; i < NCOMMAND; i++) { if (command_table[i].name == NULL) continue; if (strcmp(command, command_table[i].name) == 0) { *idx = i; return (0); } } return (1); } static int zfs_do_diff(int argc, char **argv) { zfs_handle_t *zhp; int flags = 0; char *tosnap = NULL; char *fromsnap = NULL; char *atp, *copy; int err = 0; int c; struct sigaction sa; while ((c = getopt(argc, argv, "FHt")) != -1) { switch (c) { case 'F': flags |= ZFS_DIFF_CLASSIFY; break; case 'H': flags |= ZFS_DIFF_PARSEABLE; break; case 't': flags |= ZFS_DIFF_TIMESTAMP; break; default: (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; if (argc < 1) { (void) fprintf(stderr, gettext("must provide at least one snapshot name\n")); usage(B_FALSE); } if (argc > 2) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } fromsnap = argv[0]; tosnap = (argc == 2) ? argv[1] : NULL; copy = NULL; if (*fromsnap != '@') copy = strdup(fromsnap); else if (tosnap) copy = strdup(tosnap); if (copy == NULL) usage(B_FALSE); if ((atp = strchr(copy, '@')) != NULL) *atp = '\0'; if ((zhp = zfs_open(g_zfs, copy, ZFS_TYPE_FILESYSTEM)) == NULL) { free(copy); return (1); } free(copy); /* * Ignore SIGPIPE so that the library can give us * information on any failure */ if (sigemptyset(&sa.sa_mask) == -1) { err = errno; goto out; } sa.sa_flags = 0; sa.sa_handler = SIG_IGN; if (sigaction(SIGPIPE, &sa, NULL) == -1) { err = errno; goto out; } err = zfs_show_diffs(zhp, STDOUT_FILENO, fromsnap, tosnap, flags); out: zfs_close(zhp); return (err != 0); } /* * zfs bookmark | * * Creates a bookmark with the given name from the source snapshot * or creates a copy of an existing source bookmark. */ static int zfs_do_bookmark(int argc, char **argv) { char *source, *bookname; char expbuf[ZFS_MAX_DATASET_NAME_LEN]; int source_type; nvlist_t *nvl; int ret = 0; int c; /* check options */ while ((c = getopt(argc, argv, "")) != -1) { switch (c) { case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); goto usage; } } argc -= optind; argv += optind; /* check number of arguments */ if (argc < 1) { (void) fprintf(stderr, gettext("missing source argument\n")); goto usage; } if (argc < 2) { (void) fprintf(stderr, gettext("missing bookmark argument\n")); goto usage; } source = argv[0]; bookname = argv[1]; if (strchr(source, '@') == NULL && strchr(source, '#') == NULL) { (void) fprintf(stderr, gettext("invalid source name '%s': " "must contain a '@' or '#'\n"), source); goto usage; } if (strchr(bookname, '#') == NULL) { (void) fprintf(stderr, gettext("invalid bookmark name '%s': " "must contain a '#'\n"), bookname); goto usage; } /* * expand source or bookname to full path: * one of them may be specified as short name */ { char **expand; char *source_short, *bookname_short; source_short = strpbrk(source, "@#"); bookname_short = strpbrk(bookname, "#"); if (source_short == source && bookname_short == bookname) { (void) fprintf(stderr, gettext( "either source or bookmark must be specified as " "full dataset paths")); goto usage; } else if (source_short != source && bookname_short != bookname) { expand = NULL; } else if (source_short != source) { strlcpy(expbuf, source, sizeof (expbuf)); expand = &bookname; } else if (bookname_short != bookname) { strlcpy(expbuf, bookname, sizeof (expbuf)); expand = &source; } else { abort(); } if (expand != NULL) { *strpbrk(expbuf, "@#") = '\0'; /* dataset name in buf */ (void) strlcat(expbuf, *expand, sizeof (expbuf)); *expand = expbuf; } } /* determine source type */ switch (*strpbrk(source, "@#")) { case '@': source_type = ZFS_TYPE_SNAPSHOT; break; case '#': source_type = ZFS_TYPE_BOOKMARK; break; default: abort(); } /* test the source exists */ zfs_handle_t *zhp; zhp = zfs_open(g_zfs, source, source_type); if (zhp == NULL) goto usage; zfs_close(zhp); nvl = fnvlist_alloc(); fnvlist_add_string(nvl, bookname, source); ret = lzc_bookmark(nvl, NULL); fnvlist_free(nvl); if (ret != 0) { const char *err_msg = NULL; char errbuf[1024]; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot create bookmark '%s'"), bookname); switch (ret) { case EXDEV: err_msg = "bookmark is in a different pool"; break; case ZFS_ERR_BOOKMARK_SOURCE_NOT_ANCESTOR: err_msg = "source is not an ancestor of the " "new bookmark's dataset"; break; case EEXIST: err_msg = "bookmark exists"; break; case EINVAL: err_msg = "invalid argument"; break; case ENOTSUP: err_msg = "bookmark feature not enabled"; break; case ENOSPC: err_msg = "out of space"; break; case ENOENT: err_msg = "dataset does not exist"; break; default: (void) zfs_standard_error(g_zfs, ret, errbuf); break; } if (err_msg != NULL) { (void) fprintf(stderr, "%s: %s\n", errbuf, dgettext(TEXT_DOMAIN, err_msg)); } } return (ret != 0); usage: usage(B_FALSE); return (-1); } static int zfs_do_channel_program(int argc, char **argv) { int ret, fd, c; char *progbuf, *filename, *poolname; size_t progsize, progread; nvlist_t *outnvl = NULL; uint64_t instrlimit = ZCP_DEFAULT_INSTRLIMIT; uint64_t memlimit = ZCP_DEFAULT_MEMLIMIT; boolean_t sync_flag = B_TRUE, json_output = B_FALSE; zpool_handle_t *zhp; /* check options */ while ((c = getopt(argc, argv, "nt:m:j")) != -1) { switch (c) { case 't': case 'm': { uint64_t arg; char *endp; errno = 0; arg = strtoull(optarg, &endp, 0); if (errno != 0 || *endp != '\0') { (void) fprintf(stderr, gettext( "invalid argument " "'%s': expected integer\n"), optarg); goto usage; } if (c == 't') { instrlimit = arg; } else { ASSERT3U(c, ==, 'm'); memlimit = arg; } break; } case 'n': { sync_flag = B_FALSE; break; } case 'j': { json_output = B_TRUE; break; } case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); goto usage; } } argc -= optind; argv += optind; if (argc < 2) { (void) fprintf(stderr, gettext("invalid number of arguments\n")); goto usage; } poolname = argv[0]; filename = argv[1]; if (strcmp(filename, "-") == 0) { fd = 0; filename = "standard input"; } else if ((fd = open(filename, O_RDONLY)) < 0) { (void) fprintf(stderr, gettext("cannot open '%s': %s\n"), filename, strerror(errno)); return (1); } if ((zhp = zpool_open(g_zfs, poolname)) == NULL) { (void) fprintf(stderr, gettext("cannot open pool '%s'\n"), poolname); if (fd != 0) (void) close(fd); return (1); } zpool_close(zhp); /* * Read in the channel program, expanding the program buffer as * necessary. */ progread = 0; progsize = 1024; progbuf = safe_malloc(progsize); do { ret = read(fd, progbuf + progread, progsize - progread); progread += ret; if (progread == progsize && ret > 0) { progsize *= 2; progbuf = safe_realloc(progbuf, progsize); } } while (ret > 0); if (fd != 0) (void) close(fd); if (ret < 0) { free(progbuf); (void) fprintf(stderr, gettext("cannot read '%s': %s\n"), filename, strerror(errno)); return (1); } progbuf[progread] = '\0'; /* * Any remaining arguments are passed as arguments to the lua script as * a string array: * { * "argv" -> [ "arg 1", ... "arg n" ], * } */ nvlist_t *argnvl = fnvlist_alloc(); fnvlist_add_string_array(argnvl, ZCP_ARG_CLIARGV, argv + 2, argc - 2); if (sync_flag) { ret = lzc_channel_program(poolname, progbuf, instrlimit, memlimit, argnvl, &outnvl); } else { ret = lzc_channel_program_nosync(poolname, progbuf, instrlimit, memlimit, argnvl, &outnvl); } if (ret != 0) { /* * On error, report the error message handed back by lua if one * exists. Otherwise, generate an appropriate error message, * falling back on strerror() for an unexpected return code. */ char *errstring = NULL; const char *msg = gettext("Channel program execution failed"); uint64_t instructions = 0; if (outnvl != NULL && nvlist_exists(outnvl, ZCP_RET_ERROR)) { (void) nvlist_lookup_string(outnvl, ZCP_RET_ERROR, &errstring); if (errstring == NULL) errstring = strerror(ret); if (ret == ETIME) { (void) nvlist_lookup_uint64(outnvl, ZCP_ARG_INSTRLIMIT, &instructions); } } else { switch (ret) { case EINVAL: errstring = "Invalid instruction or memory limit."; break; case ENOMEM: errstring = "Return value too large."; break; case ENOSPC: errstring = "Memory limit exhausted."; break; case ETIME: errstring = "Timed out."; break; case EPERM: errstring = "Permission denied. Channel " "programs must be run as root."; break; default: (void) zfs_standard_error(g_zfs, ret, msg); } } if (errstring != NULL) (void) fprintf(stderr, "%s:\n%s\n", msg, errstring); if (ret == ETIME && instructions != 0) (void) fprintf(stderr, gettext("%llu Lua instructions\n"), (u_longlong_t)instructions); } else { if (json_output) { (void) nvlist_print_json(stdout, outnvl); } else if (nvlist_empty(outnvl)) { (void) fprintf(stdout, gettext("Channel program fully " "executed and did not produce output.\n")); } else { (void) fprintf(stdout, gettext("Channel program fully " "executed and produced output:\n")); dump_nvlist(outnvl, 4); } } free(progbuf); fnvlist_free(outnvl); fnvlist_free(argnvl); return (ret != 0); usage: usage(B_FALSE); return (-1); } typedef struct loadkey_cbdata { boolean_t cb_loadkey; boolean_t cb_recursive; boolean_t cb_noop; char *cb_keylocation; uint64_t cb_numfailed; uint64_t cb_numattempted; } loadkey_cbdata_t; static int load_key_callback(zfs_handle_t *zhp, void *data) { int ret; boolean_t is_encroot; loadkey_cbdata_t *cb = data; uint64_t keystatus = zfs_prop_get_int(zhp, ZFS_PROP_KEYSTATUS); /* * If we are working recursively, we want to skip loading / unloading * keys for non-encryption roots and datasets whose keys are already * in the desired end-state. */ if (cb->cb_recursive) { ret = zfs_crypto_get_encryption_root(zhp, &is_encroot, NULL); if (ret != 0) return (ret); if (!is_encroot) return (0); if ((cb->cb_loadkey && keystatus == ZFS_KEYSTATUS_AVAILABLE) || (!cb->cb_loadkey && keystatus == ZFS_KEYSTATUS_UNAVAILABLE)) return (0); } cb->cb_numattempted++; if (cb->cb_loadkey) ret = zfs_crypto_load_key(zhp, cb->cb_noop, cb->cb_keylocation); else ret = zfs_crypto_unload_key(zhp); if (ret != 0) { cb->cb_numfailed++; return (ret); } return (0); } static int load_unload_keys(int argc, char **argv, boolean_t loadkey) { int c, ret = 0, flags = 0; boolean_t do_all = B_FALSE; loadkey_cbdata_t cb = { 0 }; cb.cb_loadkey = loadkey; while ((c = getopt(argc, argv, "anrL:")) != -1) { /* noop and alternate keylocations only apply to zfs load-key */ if (loadkey) { switch (c) { case 'n': cb.cb_noop = B_TRUE; continue; case 'L': cb.cb_keylocation = optarg; continue; default: break; } } switch (c) { case 'a': do_all = B_TRUE; cb.cb_recursive = B_TRUE; break; case 'r': flags |= ZFS_ITER_RECURSE; cb.cb_recursive = B_TRUE; break; default: (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; argv += optind; if (!do_all && argc == 0) { (void) fprintf(stderr, gettext("Missing dataset argument or -a option\n")); usage(B_FALSE); } if (do_all && argc != 0) { (void) fprintf(stderr, gettext("Cannot specify dataset with -a option\n")); usage(B_FALSE); } if (cb.cb_recursive && cb.cb_keylocation != NULL && strcmp(cb.cb_keylocation, "prompt") != 0) { (void) fprintf(stderr, gettext("alternate keylocation may only " "be 'prompt' with -r or -a\n")); usage(B_FALSE); } ret = zfs_for_each(argc, argv, flags, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, NULL, NULL, 0, load_key_callback, &cb); if (cb.cb_noop || (cb.cb_recursive && cb.cb_numattempted != 0)) { (void) printf(gettext("%llu / %llu key(s) successfully %s\n"), (u_longlong_t)(cb.cb_numattempted - cb.cb_numfailed), (u_longlong_t)cb.cb_numattempted, loadkey ? (cb.cb_noop ? "verified" : "loaded") : "unloaded"); } if (cb.cb_numfailed != 0) ret = -1; return (ret); } static int zfs_do_load_key(int argc, char **argv) { return (load_unload_keys(argc, argv, B_TRUE)); } static int zfs_do_unload_key(int argc, char **argv) { return (load_unload_keys(argc, argv, B_FALSE)); } static int zfs_do_change_key(int argc, char **argv) { int c, ret; uint64_t keystatus; boolean_t loadkey = B_FALSE, inheritkey = B_FALSE; zfs_handle_t *zhp = NULL; nvlist_t *props = fnvlist_alloc(); while ((c = getopt(argc, argv, "lio:")) != -1) { switch (c) { case 'l': loadkey = B_TRUE; break; case 'i': inheritkey = B_TRUE; break; case 'o': if (!parseprop(props, optarg)) { nvlist_free(props); return (1); } break; default: (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } if (inheritkey && !nvlist_empty(props)) { (void) fprintf(stderr, gettext("Properties not allowed for inheriting\n")); usage(B_FALSE); } argc -= optind; argv += optind; if (argc < 1) { (void) fprintf(stderr, gettext("Missing dataset argument\n")); usage(B_FALSE); } if (argc > 1) { (void) fprintf(stderr, gettext("Too many arguments\n")); usage(B_FALSE); } zhp = zfs_open(g_zfs, argv[argc - 1], ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME); if (zhp == NULL) usage(B_FALSE); if (loadkey) { keystatus = zfs_prop_get_int(zhp, ZFS_PROP_KEYSTATUS); if (keystatus != ZFS_KEYSTATUS_AVAILABLE) { ret = zfs_crypto_load_key(zhp, B_FALSE, NULL); if (ret != 0) { nvlist_free(props); zfs_close(zhp); return (-1); } } /* refresh the properties so the new keystatus is visible */ zfs_refresh_properties(zhp); } ret = zfs_crypto_rewrap(zhp, props, inheritkey); if (ret != 0) { nvlist_free(props); zfs_close(zhp); return (-1); } nvlist_free(props); zfs_close(zhp); return (0); } /* * 1) zfs project [-d|-r] * List project ID and inherit flag of file(s) or directories. * -d: List the directory itself, not its children. * -r: List subdirectories recursively. * * 2) zfs project -C [-k] [-r] * Clear project inherit flag and/or ID on the file(s) or directories. * -k: Keep the project ID unchanged. If not specified, the project ID * will be reset as zero. * -r: Clear on subdirectories recursively. * * 3) zfs project -c [-0] [-d|-r] [-p id] * Check project ID and inherit flag on the file(s) or directories, * report the outliers. * -0: Print file name followed by a NUL instead of newline. * -d: Check the directory itself, not its children. * -p: Specify the referenced ID for comparing with the target file(s) * or directories' project IDs. If not specified, the target (top) * directory's project ID will be used as the referenced one. * -r: Check subdirectories recursively. * * 4) zfs project [-p id] [-r] [-s] * Set project ID and/or inherit flag on the file(s) or directories. * -p: Set the project ID as the given id. * -r: Set on subdirectories recursively. If not specify "-p" option, * it will use top-level directory's project ID as the given id, * then set both project ID and inherit flag on all descendants * of the top-level directory. * -s: Set project inherit flag. */ static int zfs_do_project(int argc, char **argv) { zfs_project_control_t zpc = { .zpc_expected_projid = ZFS_INVALID_PROJID, .zpc_op = ZFS_PROJECT_OP_DEFAULT, .zpc_dironly = B_FALSE, .zpc_keep_projid = B_FALSE, .zpc_newline = B_TRUE, .zpc_recursive = B_FALSE, .zpc_set_flag = B_FALSE, }; int ret = 0, c; if (argc < 2) usage(B_FALSE); while ((c = getopt(argc, argv, "0Ccdkp:rs")) != -1) { switch (c) { case '0': zpc.zpc_newline = B_FALSE; break; case 'C': if (zpc.zpc_op != ZFS_PROJECT_OP_DEFAULT) { (void) fprintf(stderr, gettext("cannot " "specify '-C' '-c' '-s' together\n")); usage(B_FALSE); } zpc.zpc_op = ZFS_PROJECT_OP_CLEAR; break; case 'c': if (zpc.zpc_op != ZFS_PROJECT_OP_DEFAULT) { (void) fprintf(stderr, gettext("cannot " "specify '-C' '-c' '-s' together\n")); usage(B_FALSE); } zpc.zpc_op = ZFS_PROJECT_OP_CHECK; break; case 'd': zpc.zpc_dironly = B_TRUE; /* overwrite "-r" option */ zpc.zpc_recursive = B_FALSE; break; case 'k': zpc.zpc_keep_projid = B_TRUE; break; case 'p': { char *endptr; errno = 0; zpc.zpc_expected_projid = strtoull(optarg, &endptr, 0); if (errno != 0 || *endptr != '\0') { (void) fprintf(stderr, gettext("project ID must be less than " "%u\n"), UINT32_MAX); usage(B_FALSE); } if (zpc.zpc_expected_projid >= UINT32_MAX) { (void) fprintf(stderr, gettext("invalid project ID\n")); usage(B_FALSE); } break; } case 'r': zpc.zpc_recursive = B_TRUE; /* overwrite "-d" option */ zpc.zpc_dironly = B_FALSE; break; case 's': if (zpc.zpc_op != ZFS_PROJECT_OP_DEFAULT) { (void) fprintf(stderr, gettext("cannot " "specify '-C' '-c' '-s' together\n")); usage(B_FALSE); } zpc.zpc_set_flag = B_TRUE; zpc.zpc_op = ZFS_PROJECT_OP_SET; break; default: (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } if (zpc.zpc_op == ZFS_PROJECT_OP_DEFAULT) { if (zpc.zpc_expected_projid != ZFS_INVALID_PROJID) zpc.zpc_op = ZFS_PROJECT_OP_SET; else zpc.zpc_op = ZFS_PROJECT_OP_LIST; } switch (zpc.zpc_op) { case ZFS_PROJECT_OP_LIST: if (zpc.zpc_keep_projid) { (void) fprintf(stderr, gettext("'-k' is only valid together with '-C'\n")); usage(B_FALSE); } if (!zpc.zpc_newline) { (void) fprintf(stderr, gettext("'-0' is only valid together with '-c'\n")); usage(B_FALSE); } break; case ZFS_PROJECT_OP_CHECK: if (zpc.zpc_keep_projid) { (void) fprintf(stderr, gettext("'-k' is only valid together with '-C'\n")); usage(B_FALSE); } break; case ZFS_PROJECT_OP_CLEAR: if (zpc.zpc_dironly) { (void) fprintf(stderr, gettext("'-d' is useless together with '-C'\n")); usage(B_FALSE); } if (!zpc.zpc_newline) { (void) fprintf(stderr, gettext("'-0' is only valid together with '-c'\n")); usage(B_FALSE); } if (zpc.zpc_expected_projid != ZFS_INVALID_PROJID) { (void) fprintf(stderr, gettext("'-p' is useless together with '-C'\n")); usage(B_FALSE); } break; case ZFS_PROJECT_OP_SET: if (zpc.zpc_dironly) { (void) fprintf(stderr, gettext("'-d' is useless for set project ID and/or " "inherit flag\n")); usage(B_FALSE); } if (zpc.zpc_keep_projid) { (void) fprintf(stderr, gettext("'-k' is only valid together with '-C'\n")); usage(B_FALSE); } if (!zpc.zpc_newline) { (void) fprintf(stderr, gettext("'-0' is only valid together with '-c'\n")); usage(B_FALSE); } break; default: ASSERT(0); break; } argv += optind; argc -= optind; if (argc == 0) { (void) fprintf(stderr, gettext("missing file or directory target(s)\n")); usage(B_FALSE); } for (int i = 0; i < argc; i++) { int err; err = zfs_project_handle(argv[i], &zpc); if (err && !ret) ret = err; } return (ret); } static int zfs_do_wait(int argc, char **argv) { boolean_t enabled[ZFS_WAIT_NUM_ACTIVITIES]; int error, i; char c; /* By default, wait for all types of activity. */ for (i = 0; i < ZFS_WAIT_NUM_ACTIVITIES; i++) enabled[i] = B_TRUE; while ((c = getopt(argc, argv, "t:")) != -1) { switch (c) { case 't': { static char *col_subopts[] = { "deleteq", NULL }; char *value; /* Reset activities array */ bzero(&enabled, sizeof (enabled)); while (*optarg != '\0') { int activity = getsubopt(&optarg, col_subopts, &value); if (activity < 0) { (void) fprintf(stderr, gettext("invalid activity '%s'\n"), value); usage(B_FALSE); } enabled[activity] = B_TRUE; } break; } case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argv += optind; argc -= optind; if (argc < 1) { (void) fprintf(stderr, gettext("missing 'filesystem' " "argument\n")); usage(B_FALSE); } if (argc > 1) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } zfs_handle_t *zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_FILESYSTEM); if (zhp == NULL) return (1); for (;;) { boolean_t missing = B_FALSE; boolean_t any_waited = B_FALSE; for (int i = 0; i < ZFS_WAIT_NUM_ACTIVITIES; i++) { boolean_t waited; if (!enabled[i]) continue; error = zfs_wait_status(zhp, i, &missing, &waited); if (error != 0 || missing) break; any_waited = (any_waited || waited); } if (error != 0 || missing || !any_waited) break; } zfs_close(zhp); return (error); } /* * Display version message */ static int zfs_do_version(int argc, char **argv) { if (zfs_version_print() == -1) return (1); return (0); } int main(int argc, char **argv) { int ret = 0; int i = 0; char *cmdname; char **newargv; (void) setlocale(LC_ALL, ""); (void) setlocale(LC_NUMERIC, "C"); (void) textdomain(TEXT_DOMAIN); opterr = 0; /* * Make sure the user has specified some command. */ if (argc < 2) { (void) fprintf(stderr, gettext("missing command\n")); usage(B_FALSE); } cmdname = argv[1]; /* * The 'umount' command is an alias for 'unmount' */ if (strcmp(cmdname, "umount") == 0) cmdname = "unmount"; /* * The 'recv' command is an alias for 'receive' */ if (strcmp(cmdname, "recv") == 0) cmdname = "receive"; /* * The 'snap' command is an alias for 'snapshot' */ if (strcmp(cmdname, "snap") == 0) cmdname = "snapshot"; /* * Special case '-?' */ if ((strcmp(cmdname, "-?") == 0) || (strcmp(cmdname, "--help") == 0)) usage(B_TRUE); /* * Special case '-V|--version' */ if ((strcmp(cmdname, "-V") == 0) || (strcmp(cmdname, "--version") == 0)) return (zfs_do_version(argc, argv)); if ((g_zfs = libzfs_init()) == NULL) { (void) fprintf(stderr, "%s\n", libzfs_error_init(errno)); return (1); } mnttab_file = g_zfs->libzfs_mnttab; zfs_save_arguments(argc, argv, history_str, sizeof (history_str)); libzfs_print_on_error(g_zfs, B_TRUE); /* * Many commands modify input strings for string parsing reasons. * We create a copy to protect the original argv. */ newargv = malloc((argc + 1) * sizeof (newargv[0])); for (i = 0; i < argc; i++) newargv[i] = strdup(argv[i]); newargv[argc] = NULL; /* * Run the appropriate command. */ libzfs_mnttab_cache(g_zfs, B_TRUE); if (find_command_idx(cmdname, &i) == 0) { current_command = &command_table[i]; ret = command_table[i].func(argc - 1, newargv + 1); } else if (strchr(cmdname, '=') != NULL) { verify(find_command_idx("set", &i) == 0); current_command = &command_table[i]; ret = command_table[i].func(argc, newargv); } else { (void) fprintf(stderr, gettext("unrecognized " "command '%s'\n"), cmdname); usage(B_FALSE); ret = 1; } for (i = 0; i < argc; i++) free(newargv[i]); free(newargv); if (ret == 0 && log_history) (void) zpool_log_history(g_zfs, history_str); libzfs_fini(g_zfs); /* * The 'ZFS_ABORT' environment variable causes us to dump core on exit * for the purposes of running ::findleaks. */ if (getenv("ZFS_ABORT") != NULL) { (void) printf("dumping core by request\n"); abort(); } return (ret); } #ifdef __FreeBSD__ #include #include /* * Attach/detach the given dataset to/from the given jail */ /* ARGSUSED */ static int zfs_do_jail_impl(int argc, char **argv, boolean_t attach) { zfs_handle_t *zhp; int jailid, ret; /* check number of arguments */ if (argc < 3) { (void) fprintf(stderr, gettext("missing argument(s)\n")); usage(B_FALSE); } if (argc > 3) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } jailid = jail_getid(argv[1]); if (jailid < 0) { (void) fprintf(stderr, gettext("invalid jail id or name\n")); usage(B_FALSE); } zhp = zfs_open(g_zfs, argv[2], ZFS_TYPE_FILESYSTEM); if (zhp == NULL) return (1); ret = (zfs_jail(zhp, jailid, attach) != 0); zfs_close(zhp); return (ret); } /* * zfs jail jailid filesystem * * Attach the given dataset to the given jail */ /* ARGSUSED */ static int zfs_do_jail(int argc, char **argv) { return (zfs_do_jail_impl(argc, argv, B_TRUE)); } /* * zfs unjail jailid filesystem * * Detach the given dataset from the given jail */ /* ARGSUSED */ static int zfs_do_unjail(int argc, char **argv) { return (zfs_do_jail_impl(argc, argv, B_FALSE)); } #endif diff --git a/include/libzfs.h b/include/libzfs.h index e0b2676a441f..337e4934a7b2 100644 --- a/include/libzfs.h +++ b/include/libzfs.h @@ -1,932 +1,932 @@ /* * 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) 2011, 2020 by Delphix. All rights reserved. * Copyright Joyent, Inc. * Copyright (c) 2013 Steven Hartland. All rights reserved. * Copyright (c) 2016, Intel Corporation. * Copyright 2016 Nexenta Systems, Inc. * Copyright (c) 2017 Open-E, Inc. All Rights Reserved. * Copyright (c) 2019 Datto Inc. */ #ifndef _LIBZFS_H #define _LIBZFS_H #include #include #include #include #include #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif /* * Miscellaneous ZFS constants */ #define ZFS_MAXPROPLEN MAXPATHLEN #define ZPOOL_MAXPROPLEN MAXPATHLEN /* * libzfs errors */ 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 */ EZFS_PROPTYPE, /* property does not apply to dataset type */ EZFS_PROPNONINHERIT, /* property is not inheritable */ EZFS_PROPSPACE, /* bad quota or reservation */ EZFS_BADTYPE, /* dataset is not of appropriate type */ EZFS_BUSY, /* pool or dataset is busy */ EZFS_EXISTS, /* pool or dataset already exists */ EZFS_NOENT, /* no such pool or dataset */ EZFS_BADSTREAM, /* bad backup stream */ EZFS_DSREADONLY, /* dataset is readonly */ EZFS_VOLTOOBIG, /* volume is too large for 32-bit system */ EZFS_INVALIDNAME, /* invalid dataset name */ EZFS_BADRESTORE, /* unable to restore to destination */ EZFS_BADBACKUP, /* backup failed */ EZFS_BADTARGET, /* bad attach/detach/replace target */ EZFS_NODEVICE, /* no such device in pool */ EZFS_BADDEV, /* invalid device to add */ EZFS_NOREPLICAS, /* no valid replicas */ EZFS_RESILVERING, /* resilvering (healing reconstruction) */ EZFS_BADVERSION, /* unsupported version */ EZFS_POOLUNAVAIL, /* pool is currently unavailable */ EZFS_DEVOVERFLOW, /* too many devices in one vdev */ EZFS_BADPATH, /* must be an absolute path */ EZFS_CROSSTARGET, /* rename or clone across pool or dataset */ EZFS_ZONED, /* used improperly in local zone */ EZFS_MOUNTFAILED, /* failed to mount dataset */ EZFS_UMOUNTFAILED, /* failed to unmount dataset */ - EZFS_UNSHARENFSFAILED, /* unshare(1M) failed */ - EZFS_SHARENFSFAILED, /* share(1M) failed */ + EZFS_UNSHARENFSFAILED, /* failed to unshare over nfs */ + EZFS_SHARENFSFAILED, /* failed to share over nfs */ EZFS_PERM, /* permission denied */ EZFS_NOSPC, /* out of space */ EZFS_FAULT, /* bad address */ EZFS_IO, /* I/O error */ EZFS_INTR, /* signal received */ EZFS_ISSPARE, /* device is a hot spare */ EZFS_INVALCONFIG, /* invalid vdev configuration */ EZFS_RECURSIVE, /* recursive dependency */ EZFS_NOHISTORY, /* no history object */ EZFS_POOLPROPS, /* couldn't retrieve pool props */ EZFS_POOL_NOTSUP, /* ops not supported for this type of pool */ EZFS_POOL_INVALARG, /* invalid argument for this pool operation */ EZFS_NAMETOOLONG, /* dataset name is too long */ EZFS_OPENFAILED, /* open of device failed */ EZFS_NOCAP, /* couldn't get capacity */ EZFS_LABELFAILED, /* write of label failed */ EZFS_BADWHO, /* invalid permission who */ EZFS_BADPERM, /* invalid permission */ EZFS_BADPERMSET, /* invalid permission set name */ EZFS_NODELEGATION, /* delegated administration is disabled */ EZFS_UNSHARESMBFAILED, /* failed to unshare over smb */ EZFS_SHARESMBFAILED, /* failed to share over smb */ EZFS_BADCACHE, /* bad cache file */ EZFS_ISL2CACHE, /* device is for the level 2 ARC */ EZFS_VDEVNOTSUP, /* unsupported vdev type */ EZFS_NOTSUP, /* ops not supported on this dataset */ EZFS_ACTIVE_SPARE, /* pool has active shared spare devices */ EZFS_UNPLAYED_LOGS, /* log device has unplayed logs */ EZFS_REFTAG_RELE, /* snapshot release: tag not found */ EZFS_REFTAG_HOLD, /* snapshot hold: tag already exists */ EZFS_TAGTOOLONG, /* snapshot hold/rele: tag too long */ EZFS_PIPEFAILED, /* pipe create failed */ EZFS_THREADCREATEFAILED, /* thread create failed */ EZFS_POSTSPLIT_ONLINE, /* onlining a disk after splitting it */ EZFS_SCRUBBING, /* currently scrubbing */ EZFS_NO_SCRUB, /* no active scrub */ EZFS_DIFF, /* general failure of zfs diff */ EZFS_DIFFDATA, /* bad zfs diff data */ EZFS_POOLREADONLY, /* pool is in read-only mode */ EZFS_SCRUB_PAUSED, /* scrub currently paused */ EZFS_ACTIVE_POOL, /* pool is imported on a different system */ EZFS_CRYPTOFAILED, /* failed to setup encryption */ EZFS_NO_PENDING, /* cannot cancel, no operation is pending */ EZFS_CHECKPOINT_EXISTS, /* checkpoint exists */ EZFS_DISCARDING_CHECKPOINT, /* currently discarding a checkpoint */ EZFS_NO_CHECKPOINT, /* pool has no checkpoint */ EZFS_DEVRM_IN_PROGRESS, /* a device is currently being removed */ EZFS_VDEV_TOO_BIG, /* a device is too big to be used */ EZFS_IOC_NOTSUPPORTED, /* operation not supported by zfs module */ EZFS_TOOMANY, /* argument list too long */ EZFS_INITIALIZING, /* currently initializing */ EZFS_NO_INITIALIZE, /* no active initialize */ EZFS_WRONG_PARENT, /* invalid parent dataset (e.g ZVOL) */ EZFS_TRIMMING, /* currently trimming */ EZFS_NO_TRIM, /* no active trim */ EZFS_TRIM_NOTSUP, /* device does not support trim */ EZFS_NO_RESILVER_DEFER, /* pool doesn't support resilver_defer */ EZFS_EXPORT_IN_PROGRESS, /* currently exporting the pool */ EZFS_REBUILDING, /* resilvering (sequential reconstrution) */ EZFS_UNKNOWN } zfs_error_t; /* * The following data structures are all part * of the zfs_allow_t data structure which is * used for printing 'allow' permissions. * It is a linked list of zfs_allow_t's which * then contain avl tree's for user/group/sets/... * and each one of the entries in those trees have * avl tree's for the permissions they belong to and * whether they are local,descendent or local+descendent * permissions. The AVL trees are used primarily for * sorting purposes, but also so that we can quickly find * a given user and or permission. */ typedef struct zfs_perm_node { avl_node_t z_node; char z_pname[MAXPATHLEN]; } zfs_perm_node_t; typedef struct zfs_allow_node { avl_node_t z_node; char z_key[MAXPATHLEN]; /* name, such as joe */ avl_tree_t z_localdescend; /* local+descendent perms */ avl_tree_t z_local; /* local permissions */ avl_tree_t z_descend; /* descendent permissions */ } zfs_allow_node_t; typedef struct zfs_allow { struct zfs_allow *z_next; char z_setpoint[MAXPATHLEN]; avl_tree_t z_sets; avl_tree_t z_crperms; avl_tree_t z_user; avl_tree_t z_group; avl_tree_t z_everyone; } zfs_allow_t; /* * Basic handle types */ typedef struct zfs_handle zfs_handle_t; typedef struct zpool_handle zpool_handle_t; typedef struct libzfs_handle libzfs_handle_t; extern int zpool_wait(zpool_handle_t *, zpool_wait_activity_t); extern int zpool_wait_status(zpool_handle_t *, zpool_wait_activity_t, boolean_t *, boolean_t *); /* * Library initialization */ extern libzfs_handle_t *libzfs_init(void); extern void libzfs_fini(libzfs_handle_t *); extern libzfs_handle_t *zpool_get_handle(zpool_handle_t *); extern libzfs_handle_t *zfs_get_handle(zfs_handle_t *); 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_init(int); extern const char *libzfs_error_action(libzfs_handle_t *); extern const char *libzfs_error_description(libzfs_handle_t *); extern int zfs_standard_error(libzfs_handle_t *, int, const char *); extern void libzfs_mnttab_init(libzfs_handle_t *); extern void libzfs_mnttab_fini(libzfs_handle_t *); extern void libzfs_mnttab_cache(libzfs_handle_t *, boolean_t); extern int libzfs_mnttab_find(libzfs_handle_t *, const char *, struct mnttab *); extern void libzfs_mnttab_add(libzfs_handle_t *, const char *, const char *, const char *); extern void libzfs_mnttab_remove(libzfs_handle_t *, const char *); /* * Basic handle functions */ extern zpool_handle_t *zpool_open(libzfs_handle_t *, const char *); extern zpool_handle_t *zpool_open_canfail(libzfs_handle_t *, const char *); extern void zpool_close(zpool_handle_t *); extern const char *zpool_get_name(zpool_handle_t *); extern int zpool_get_state(zpool_handle_t *); extern const char *zpool_state_to_name(vdev_state_t, vdev_aux_t); extern const char *zpool_pool_state_to_name(pool_state_t); extern void zpool_free_handles(libzfs_handle_t *); /* * Iterate over all active pools in the system. */ typedef int (*zpool_iter_f)(zpool_handle_t *, void *); extern int zpool_iter(libzfs_handle_t *, zpool_iter_f, void *); extern boolean_t zpool_skip_pool(const char *); /* * Functions to create and destroy pools */ extern int zpool_create(libzfs_handle_t *, const char *, nvlist_t *, nvlist_t *, nvlist_t *); extern int zpool_destroy(zpool_handle_t *, const char *); extern int zpool_add(zpool_handle_t *, nvlist_t *); typedef struct splitflags { /* do not split, but return the config that would be split off */ int dryrun : 1; /* after splitting, import the pool */ int import : 1; int name_flags; } splitflags_t; typedef struct trimflags { /* requested vdevs are for the entire pool */ boolean_t fullpool; /* request a secure trim, requires support from device */ boolean_t secure; /* after starting trim, block until trim completes */ boolean_t wait; /* trim at the requested rate in bytes/second */ uint64_t rate; } trimflags_t; /* * Functions to manipulate pool and vdev state */ extern int zpool_scan(zpool_handle_t *, pool_scan_func_t, pool_scrub_cmd_t); extern int zpool_initialize(zpool_handle_t *, pool_initialize_func_t, nvlist_t *); extern int zpool_initialize_wait(zpool_handle_t *, pool_initialize_func_t, nvlist_t *); extern int zpool_trim(zpool_handle_t *, pool_trim_func_t, nvlist_t *, trimflags_t *); extern int zpool_clear(zpool_handle_t *, const char *, nvlist_t *); extern int zpool_reguid(zpool_handle_t *); extern int zpool_reopen_one(zpool_handle_t *, void *); extern int zpool_sync_one(zpool_handle_t *, void *); extern int zpool_vdev_online(zpool_handle_t *, const char *, int, vdev_state_t *); extern int zpool_vdev_offline(zpool_handle_t *, const char *, boolean_t); extern int zpool_vdev_attach(zpool_handle_t *, const char *, const char *, nvlist_t *, int, boolean_t); extern int zpool_vdev_detach(zpool_handle_t *, const char *); extern int zpool_vdev_remove(zpool_handle_t *, const char *); extern int zpool_vdev_remove_cancel(zpool_handle_t *); extern int zpool_vdev_indirect_size(zpool_handle_t *, const char *, uint64_t *); extern int zpool_vdev_split(zpool_handle_t *, char *, nvlist_t **, nvlist_t *, splitflags_t); extern int zpool_vdev_fault(zpool_handle_t *, uint64_t, vdev_aux_t); extern int zpool_vdev_degrade(zpool_handle_t *, uint64_t, vdev_aux_t); extern int zpool_vdev_clear(zpool_handle_t *, uint64_t); extern nvlist_t *zpool_find_vdev(zpool_handle_t *, const char *, boolean_t *, boolean_t *, boolean_t *); extern nvlist_t *zpool_find_vdev_by_physpath(zpool_handle_t *, const char *, boolean_t *, boolean_t *, boolean_t *); extern int zpool_label_disk(libzfs_handle_t *, zpool_handle_t *, const char *); extern uint64_t zpool_vdev_path_to_guid(zpool_handle_t *zhp, const char *path); const char *zpool_get_state_str(zpool_handle_t *); /* * Functions to manage pool properties */ extern int zpool_set_prop(zpool_handle_t *, const char *, const char *); extern int zpool_get_prop(zpool_handle_t *, zpool_prop_t, char *, size_t proplen, zprop_source_t *, boolean_t literal); extern uint64_t zpool_get_prop_int(zpool_handle_t *, zpool_prop_t, zprop_source_t *); extern int zpool_props_refresh(zpool_handle_t *); extern const char *zpool_prop_to_name(zpool_prop_t); extern const char *zpool_prop_values(zpool_prop_t); /* * Pool health statistics. */ typedef enum { /* * The following correspond to faults as defined in the (fault.fs.zfs.*) * event namespace. Each is associated with a corresponding message ID. * This must be kept in sync with the zfs_msgid_table in * lib/libzfs/libzfs_status.c. */ ZPOOL_STATUS_CORRUPT_CACHE, /* corrupt /kernel/drv/zpool.cache */ ZPOOL_STATUS_MISSING_DEV_R, /* missing device with replicas */ ZPOOL_STATUS_MISSING_DEV_NR, /* missing device with no replicas */ ZPOOL_STATUS_CORRUPT_LABEL_R, /* bad device label with replicas */ ZPOOL_STATUS_CORRUPT_LABEL_NR, /* bad device label with no replicas */ ZPOOL_STATUS_BAD_GUID_SUM, /* sum of device guids didn't match */ ZPOOL_STATUS_CORRUPT_POOL, /* pool metadata is corrupted */ ZPOOL_STATUS_CORRUPT_DATA, /* data errors in user (meta)data */ ZPOOL_STATUS_FAILING_DEV, /* device experiencing errors */ ZPOOL_STATUS_VERSION_NEWER, /* newer on-disk version */ ZPOOL_STATUS_HOSTID_MISMATCH, /* last accessed by another system */ ZPOOL_STATUS_HOSTID_ACTIVE, /* currently active on another system */ ZPOOL_STATUS_HOSTID_REQUIRED, /* multihost=on and hostid=0 */ ZPOOL_STATUS_IO_FAILURE_WAIT, /* failed I/O, failmode 'wait' */ ZPOOL_STATUS_IO_FAILURE_CONTINUE, /* failed I/O, failmode 'continue' */ ZPOOL_STATUS_IO_FAILURE_MMP, /* failed MMP, failmode not 'panic' */ ZPOOL_STATUS_BAD_LOG, /* cannot read log chain(s) */ ZPOOL_STATUS_ERRATA, /* informational errata available */ /* * If the pool has unsupported features but can still be opened in * read-only mode, its status is ZPOOL_STATUS_UNSUP_FEAT_WRITE. If the * pool has unsupported features but cannot be opened at all, its * status is ZPOOL_STATUS_UNSUP_FEAT_READ. */ ZPOOL_STATUS_UNSUP_FEAT_READ, /* unsupported features for read */ ZPOOL_STATUS_UNSUP_FEAT_WRITE, /* unsupported features for write */ /* * These faults have no corresponding message ID. At the time we are * checking the status, the original reason for the FMA fault (I/O or * checksum errors) has been lost. */ ZPOOL_STATUS_FAULTED_DEV_R, /* faulted device with replicas */ ZPOOL_STATUS_FAULTED_DEV_NR, /* faulted device with no replicas */ /* * The following are not faults per se, but still an error possibly * requiring administrative attention. There is no corresponding * message ID. */ ZPOOL_STATUS_VERSION_OLDER, /* older legacy on-disk version */ ZPOOL_STATUS_FEAT_DISABLED, /* supported features are disabled */ ZPOOL_STATUS_RESILVERING, /* device being resilvered */ ZPOOL_STATUS_OFFLINE_DEV, /* device offline */ ZPOOL_STATUS_REMOVED_DEV, /* removed device */ ZPOOL_STATUS_REBUILDING, /* device being rebuilt */ ZPOOL_STATUS_REBUILD_SCRUB, /* recommend scrubbing the pool */ ZPOOL_STATUS_NON_NATIVE_ASHIFT, /* (e.g. 512e dev with ashift of 9) */ /* * Finally, the following indicates a healthy pool. */ ZPOOL_STATUS_OK } zpool_status_t; extern zpool_status_t zpool_get_status(zpool_handle_t *, char **, zpool_errata_t *); extern zpool_status_t zpool_import_status(nvlist_t *, char **, zpool_errata_t *); /* * Statistics and configuration functions. */ extern nvlist_t *zpool_get_config(zpool_handle_t *, nvlist_t **); extern nvlist_t *zpool_get_features(zpool_handle_t *); extern int zpool_refresh_stats(zpool_handle_t *, boolean_t *); extern int zpool_get_errlog(zpool_handle_t *, nvlist_t **); /* * Import and export functions */ 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 *, nvlist_t *, int); extern void zpool_print_unsup_feat(nvlist_t *config); /* * Miscellaneous pool functions */ struct zfs_cmd; extern const char *zfs_history_event_names[]; typedef enum { VDEV_NAME_PATH = 1 << 0, VDEV_NAME_GUID = 1 << 1, VDEV_NAME_FOLLOW_LINKS = 1 << 2, VDEV_NAME_TYPE_ID = 1 << 3, } vdev_name_t; extern char *zpool_vdev_name(libzfs_handle_t *, zpool_handle_t *, nvlist_t *, int name_flags); extern int zpool_upgrade(zpool_handle_t *, uint64_t); extern int zpool_get_history(zpool_handle_t *, nvlist_t **, uint64_t *, boolean_t *); extern int zpool_events_next(libzfs_handle_t *, nvlist_t **, int *, unsigned, int); extern int zpool_events_clear(libzfs_handle_t *, int *); extern int zpool_events_seek(libzfs_handle_t *, uint64_t, int); extern void zpool_obj_to_path_ds(zpool_handle_t *, uint64_t, uint64_t, char *, size_t); extern void zpool_obj_to_path(zpool_handle_t *, uint64_t, uint64_t, char *, size_t); extern int zfs_ioctl(libzfs_handle_t *, int, struct zfs_cmd *); extern int zpool_get_physpath(zpool_handle_t *, char *, size_t); extern void zpool_explain_recover(libzfs_handle_t *, const char *, int, nvlist_t *); extern int zpool_checkpoint(zpool_handle_t *); extern int zpool_discard_checkpoint(zpool_handle_t *); /* * Basic handle manipulations. These functions do not create or destroy the * underlying datasets, only the references to them. */ extern zfs_handle_t *zfs_open(libzfs_handle_t *, const char *, int); extern zfs_handle_t *zfs_handle_dup(zfs_handle_t *); extern void zfs_close(zfs_handle_t *); extern zfs_type_t zfs_get_type(const zfs_handle_t *); extern const char *zfs_get_name(const zfs_handle_t *); extern zpool_handle_t *zfs_get_pool_handle(const zfs_handle_t *); extern const char *zfs_get_pool_name(const zfs_handle_t *); /* * Property management functions. Some functions are shared with the kernel, * and are found in sys/fs/zfs.h. */ /* * zfs dataset property management */ extern const char *zfs_prop_default_string(zfs_prop_t); extern uint64_t zfs_prop_default_numeric(zfs_prop_t); extern const char *zfs_prop_column_name(zfs_prop_t); extern boolean_t zfs_prop_align_right(zfs_prop_t); extern nvlist_t *zfs_valid_proplist(libzfs_handle_t *, zfs_type_t, nvlist_t *, uint64_t, zfs_handle_t *, zpool_handle_t *, boolean_t, const char *); extern const char *zfs_prop_to_name(zfs_prop_t); extern int zfs_prop_set(zfs_handle_t *, const char *, const char *); extern int zfs_prop_set_list(zfs_handle_t *, nvlist_t *); extern int zfs_prop_get(zfs_handle_t *, zfs_prop_t, char *, size_t, zprop_source_t *, char *, size_t, boolean_t); extern int zfs_prop_get_recvd(zfs_handle_t *, const char *, char *, size_t, boolean_t); extern int zfs_prop_get_numeric(zfs_handle_t *, zfs_prop_t, uint64_t *, zprop_source_t *, char *, size_t); extern int zfs_prop_get_userquota_int(zfs_handle_t *zhp, const char *propname, uint64_t *propvalue); extern int zfs_prop_get_userquota(zfs_handle_t *zhp, const char *propname, char *propbuf, int proplen, boolean_t literal); extern int zfs_prop_get_written_int(zfs_handle_t *zhp, const char *propname, uint64_t *propvalue); extern int zfs_prop_get_written(zfs_handle_t *zhp, const char *propname, 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 uint64_t getprop_uint64(zfs_handle_t *, zfs_prop_t, char **); 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); extern int zfs_prop_is_string(zfs_prop_t prop); extern nvlist_t *zfs_get_all_props(zfs_handle_t *); extern nvlist_t *zfs_get_user_props(zfs_handle_t *); extern nvlist_t *zfs_get_recvd_props(zfs_handle_t *); extern nvlist_t *zfs_get_clones_nvl(zfs_handle_t *); extern int zfs_wait_status(zfs_handle_t *, zfs_wait_activity_t, boolean_t *, boolean_t *); /* * zfs encryption management */ extern int zfs_crypto_get_encryption_root(zfs_handle_t *, boolean_t *, char *); extern int zfs_crypto_create(libzfs_handle_t *, char *, nvlist_t *, nvlist_t *, boolean_t stdin_available, uint8_t **, uint_t *); extern int zfs_crypto_clone_check(libzfs_handle_t *, zfs_handle_t *, char *, nvlist_t *); extern int zfs_crypto_attempt_load_keys(libzfs_handle_t *, char *); extern int zfs_crypto_load_key(zfs_handle_t *, boolean_t, char *); extern int zfs_crypto_unload_key(zfs_handle_t *); extern int zfs_crypto_rewrap(zfs_handle_t *, nvlist_t *, boolean_t); typedef struct zprop_list { int pl_prop; char *pl_user_prop; struct zprop_list *pl_next; boolean_t pl_all; size_t pl_width; size_t pl_recvd_width; boolean_t pl_fixed; } zprop_list_t; extern int zfs_expand_proplist(zfs_handle_t *, zprop_list_t **, boolean_t, boolean_t); extern void zfs_prune_proplist(zfs_handle_t *, uint8_t *); #define ZFS_MOUNTPOINT_NONE "none" #define ZFS_MOUNTPOINT_LEGACY "legacy" #define ZFS_FEATURE_DISABLED "disabled" #define ZFS_FEATURE_ENABLED "enabled" #define ZFS_FEATURE_ACTIVE "active" #define ZFS_UNSUPPORTED_INACTIVE "inactive" #define ZFS_UNSUPPORTED_READONLY "readonly" /* * zpool property management */ extern int zpool_expand_proplist(zpool_handle_t *, zprop_list_t **); extern int zpool_prop_get_feature(zpool_handle_t *, const char *, char *, size_t); extern const char *zpool_prop_default_string(zpool_prop_t); extern uint64_t zpool_prop_default_numeric(zpool_prop_t); extern const char *zpool_prop_column_name(zpool_prop_t); extern boolean_t zpool_prop_align_right(zpool_prop_t); /* * Functions shared by zfs and zpool property management. */ extern int zprop_iter(zprop_func func, void *cb, boolean_t show_all, boolean_t ordered, zfs_type_t type); extern int zprop_get_list(libzfs_handle_t *, char *, zprop_list_t **, zfs_type_t); extern void zprop_free_list(zprop_list_t *); #define ZFS_GET_NCOLS 5 typedef enum { GET_COL_NONE, GET_COL_NAME, GET_COL_PROPERTY, GET_COL_VALUE, GET_COL_RECVD, GET_COL_SOURCE } zfs_get_column_t; /* * Functions for printing zfs or zpool properties */ typedef struct zprop_get_cbdata { int cb_sources; zfs_get_column_t cb_columns[ZFS_GET_NCOLS]; int cb_colwidths[ZFS_GET_NCOLS + 1]; boolean_t cb_scripted; boolean_t cb_literal; boolean_t cb_first; zprop_list_t *cb_proplist; zfs_type_t cb_type; } zprop_get_cbdata_t; void zprop_print_one_property(const char *, zprop_get_cbdata_t *, const char *, const char *, zprop_source_t, const char *, const char *); /* * Iterator functions. */ typedef int (*zfs_iter_f)(zfs_handle_t *, void *); extern int zfs_iter_root(libzfs_handle_t *, zfs_iter_f, void *); extern int zfs_iter_children(zfs_handle_t *, zfs_iter_f, void *); extern int zfs_iter_dependents(zfs_handle_t *, boolean_t, zfs_iter_f, void *); extern int zfs_iter_filesystems(zfs_handle_t *, zfs_iter_f, void *); extern int zfs_iter_snapshots(zfs_handle_t *, boolean_t, zfs_iter_f, void *, uint64_t, uint64_t); extern int zfs_iter_snapshots_sorted(zfs_handle_t *, zfs_iter_f, void *, uint64_t, uint64_t); extern int zfs_iter_snapspec(zfs_handle_t *, const char *, zfs_iter_f, void *); extern int zfs_iter_bookmarks(zfs_handle_t *, zfs_iter_f, void *); extern int zfs_iter_mounted(zfs_handle_t *, zfs_iter_f, void *); typedef struct get_all_cb { zfs_handle_t **cb_handles; size_t cb_alloc; size_t cb_used; } get_all_cb_t; void zfs_foreach_mountpoint(libzfs_handle_t *, zfs_handle_t **, size_t, zfs_iter_f, void *, boolean_t); void libzfs_add_handle(get_all_cb_t *, zfs_handle_t *); /* * Functions to create and destroy datasets. */ extern int zfs_create(libzfs_handle_t *, const char *, zfs_type_t, nvlist_t *); extern int zfs_create_ancestors(libzfs_handle_t *, const char *); extern int zfs_destroy(zfs_handle_t *, boolean_t); extern int zfs_destroy_snaps(zfs_handle_t *, char *, boolean_t); extern int zfs_destroy_snaps_nvl(libzfs_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); typedef struct renameflags { /* recursive rename */ int recursive : 1; /* don't unmount file systems */ int nounmount : 1; /* force unmount file systems */ int forceunmount : 1; } renameflags_t; extern int zfs_rename(zfs_handle_t *, const char *, renameflags_t); typedef struct sendflags { /* Amount of extra information to print. */ int verbosity; /* recursive send (ie, -R) */ boolean_t replicate; /* for incrementals, do all intermediate snapshots */ boolean_t doall; /* if dataset is a clone, do incremental from its origin */ boolean_t fromorigin; /* field no longer used, maintained for backwards compatibility */ boolean_t pad; /* send properties (ie, -p) */ boolean_t props; /* do not send (no-op, ie. -n) */ boolean_t dryrun; /* parsable verbose output (ie. -P) */ boolean_t parsable; /* show progress (ie. -v) */ boolean_t progress; /* large blocks (>128K) are permitted */ boolean_t largeblock; /* WRITE_EMBEDDED records of type DATA are permitted */ boolean_t embed_data; /* compressed WRITE records are permitted */ boolean_t compress; /* raw encrypted records are permitted */ boolean_t raw; /* only send received properties (ie. -b) */ boolean_t backup; /* include snapshot holds in send stream */ boolean_t holds; /* stream represents a partially received dataset */ boolean_t saved; } sendflags_t; typedef boolean_t (snapfilter_cb_t)(zfs_handle_t *, void *); extern int zfs_send(zfs_handle_t *, const char *, const char *, sendflags_t *, int, snapfilter_cb_t, void *, nvlist_t **); extern int zfs_send_one(zfs_handle_t *, const char *, int, sendflags_t *, const char *); extern int zfs_send_progress(zfs_handle_t *, int, uint64_t *, uint64_t *); extern int zfs_send_resume(libzfs_handle_t *, sendflags_t *, int outfd, const char *); extern int zfs_send_saved(zfs_handle_t *, sendflags_t *, int, const char *); extern nvlist_t *zfs_send_resume_token_to_nvlist(libzfs_handle_t *hdl, const char *token); extern int zfs_promote(zfs_handle_t *); extern int zfs_hold(zfs_handle_t *, const char *, const char *, boolean_t, int); extern int zfs_hold_nvl(zfs_handle_t *, int, nvlist_t *); extern int zfs_release(zfs_handle_t *, const char *, const char *, boolean_t); extern int zfs_get_holds(zfs_handle_t *, nvlist_t **); extern uint64_t zvol_volsize_to_reservation(zpool_handle_t *, uint64_t, nvlist_t *); typedef int (*zfs_userspace_cb_t)(void *arg, const char *domain, uid_t rid, uint64_t space); extern int zfs_userspace(zfs_handle_t *, zfs_userquota_prop_t, zfs_userspace_cb_t, void *); extern int zfs_get_fsacl(zfs_handle_t *, nvlist_t **); extern int zfs_set_fsacl(zfs_handle_t *, boolean_t, nvlist_t *); typedef struct recvflags { /* print informational messages (ie, -v was specified) */ boolean_t verbose; /* the destination is a prefix, not the exact fs (ie, -d) */ boolean_t isprefix; /* * Only the tail of the sent snapshot path is appended to the * destination to determine the received snapshot name (ie, -e). */ boolean_t istail; /* do not actually do the recv, just check if it would work (ie, -n) */ boolean_t dryrun; /* rollback/destroy filesystems as necessary (eg, -F) */ boolean_t force; /* set "canmount=off" on all modified filesystems */ boolean_t canmountoff; /* * Mark the file systems as "resumable" and do not destroy them if the * receive is interrupted */ boolean_t resumable; /* byteswap flag is used internally; callers need not specify */ boolean_t byteswap; /* do not mount file systems as they are extracted (private) */ boolean_t nomount; /* Was holds flag set in the compound header? */ boolean_t holds; /* skip receive of snapshot holds */ boolean_t skipholds; /* mount the filesystem unless nomount is specified */ boolean_t domount; /* force unmount while recv snapshot (private) */ boolean_t forceunmount; } recvflags_t; extern int zfs_receive(libzfs_handle_t *, const char *, nvlist_t *, recvflags_t *, int, avl_tree_t *); typedef enum diff_flags { ZFS_DIFF_PARSEABLE = 0x1, ZFS_DIFF_TIMESTAMP = 0x2, ZFS_DIFF_CLASSIFY = 0x4 } diff_flags_t; extern int zfs_show_diffs(zfs_handle_t *, int, const char *, const char *, int); /* * Miscellaneous functions. */ extern const char *zfs_type_to_name(zfs_type_t); extern void zfs_refresh_properties(zfs_handle_t *); extern int zfs_name_valid(const char *, zfs_type_t); extern zfs_handle_t *zfs_path_to_zhandle(libzfs_handle_t *, const char *, zfs_type_t); extern int zfs_parent_name(zfs_handle_t *, char *, size_t); extern boolean_t zfs_dataset_exists(libzfs_handle_t *, const char *, zfs_type_t); extern int zfs_spa_version(zfs_handle_t *, int *); extern boolean_t zfs_bookmark_exists(const char *path); /* * Mount support functions. */ extern boolean_t is_mounted(libzfs_handle_t *, const char *special, char **); extern boolean_t zfs_is_mounted(zfs_handle_t *, char **); extern int zfs_mount(zfs_handle_t *, const char *, int); extern int zfs_mount_at(zfs_handle_t *, const char *, int, const char *); extern int zfs_unmount(zfs_handle_t *, const char *, int); extern int zfs_unmountall(zfs_handle_t *, int); #if defined(__linux__) extern int zfs_parse_mount_options(char *mntopts, unsigned long *mntflags, unsigned long *zfsflags, int sloppy, char *badopt, char *mtabopt); extern void zfs_adjust_mount_options(zfs_handle_t *zhp, const char *mntpoint, char *mntopts, char *mtabopt); #endif /* * Share support functions. */ extern boolean_t zfs_is_shared(zfs_handle_t *); extern int zfs_share(zfs_handle_t *); extern int zfs_unshare(zfs_handle_t *); /* * Protocol-specific share support functions. */ extern boolean_t zfs_is_shared_nfs(zfs_handle_t *, char **); extern boolean_t zfs_is_shared_smb(zfs_handle_t *, char **); extern int zfs_share_nfs(zfs_handle_t *); extern int zfs_share_smb(zfs_handle_t *); extern int zfs_shareall(zfs_handle_t *); extern int zfs_unshare_nfs(zfs_handle_t *, const char *); extern int zfs_unshare_smb(zfs_handle_t *, const char *); extern int zfs_unshareall_nfs(zfs_handle_t *); extern int zfs_unshareall_smb(zfs_handle_t *); extern int zfs_unshareall_bypath(zfs_handle_t *, const char *); extern int zfs_unshareall_bytype(zfs_handle_t *, const char *, const char *); extern int zfs_unshareall(zfs_handle_t *); extern int zfs_deleg_share_nfs(libzfs_handle_t *, char *, char *, char *, void *, void *, int, zfs_share_op_t); extern void zfs_commit_nfs_shares(void); extern void zfs_commit_smb_shares(void); extern void zfs_commit_all_shares(void); extern void zfs_commit_shares(const char *); extern int zfs_nicestrtonum(libzfs_handle_t *, const char *, uint64_t *); /* * Utility functions to run an external process. */ #define STDOUT_VERBOSE 0x01 #define STDERR_VERBOSE 0x02 #define NO_DEFAULT_PATH 0x04 /* Don't use $PATH to lookup the command */ int libzfs_run_process(const char *, char **, int); int libzfs_run_process_get_stdout(const char *, char *[], char *[], char **[], int *); int libzfs_run_process_get_stdout_nopath(const char *, char *[], char *[], char **[], int *); void libzfs_free_str_array(char **, int); int libzfs_envvar_is_set(char *); /* * Utility functions for zfs version */ extern void zfs_version_userland(char *, int); extern int zfs_version_kernel(char *, int); extern int zfs_version_print(void); /* * Given a device or file, determine if it is part of a pool. */ extern int zpool_in_use(libzfs_handle_t *, int, pool_state_t *, char **, boolean_t *); /* * Label manipulation. */ extern int zpool_clear_label(int); extern int zpool_set_bootenv(zpool_handle_t *, const nvlist_t *); extern int zpool_get_bootenv(zpool_handle_t *, nvlist_t **); /* * Management interfaces for SMB ACL files */ int zfs_smb_acl_add(libzfs_handle_t *, char *, char *, char *); int zfs_smb_acl_remove(libzfs_handle_t *, char *, char *, char *); int zfs_smb_acl_purge(libzfs_handle_t *, char *, char *); int zfs_smb_acl_rename(libzfs_handle_t *, char *, char *, char *, char *); /* * Enable and disable datasets within a pool by mounting/unmounting and * sharing/unsharing them. */ extern int zpool_enable_datasets(zpool_handle_t *, const char *, int); extern int zpool_disable_datasets(zpool_handle_t *, boolean_t); #ifdef __FreeBSD__ /* * Attach/detach the given filesystem to/from the given jail. */ extern int zfs_jail(zfs_handle_t *zhp, int jailid, int attach); /* * Set loader options for next boot. */ extern int zpool_nextboot(libzfs_handle_t *, uint64_t, uint64_t, const char *); #endif /* __FreeBSD__ */ #ifdef __cplusplus } #endif #endif /* _LIBZFS_H */ diff --git a/include/sys/fs/zfs.h b/include/sys/fs/zfs.h index fe63d735babc..211dd6d50c8c 100644 --- a/include/sys/fs/zfs.h +++ b/include/sys/fs/zfs.h @@ -1,1589 +1,1589 @@ /* * 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) 2011, 2020 by Delphix. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2013, 2017 Joyent, Inc. All rights reserved. * Copyright (c) 2014 Integros [integros.com] * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2019 Datto Inc. */ /* Portions Copyright 2010 Robert Milkowski */ #ifndef _SYS_FS_ZFS_H #define _SYS_FS_ZFS_H #include #include #ifdef __cplusplus extern "C" { #endif /* * Types and constants shared between userland and the kernel. */ /* * Each dataset can be one of the following types. These constants can be * combined into masks that can be passed to various functions. */ typedef enum { ZFS_TYPE_FILESYSTEM = (1 << 0), ZFS_TYPE_SNAPSHOT = (1 << 1), ZFS_TYPE_VOLUME = (1 << 2), ZFS_TYPE_POOL = (1 << 3), ZFS_TYPE_BOOKMARK = (1 << 4) } zfs_type_t; /* * NB: lzc_dataset_type should be updated whenever a new objset type is added, * if it represents a real type of a dataset that can be created from userland. */ 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) /* * All of these include the terminating NUL byte. */ #define ZAP_MAXNAMELEN 256 #define ZAP_MAXVALUELEN (1024 * 8) #define ZAP_OLDMAXVALUELEN 1024 #define ZFS_MAX_DATASET_NAME_LEN 256 /* * Dataset properties are identified by these constants and must be added to * the end of this list to ensure that external consumers are not affected * by the change. If you make any changes to this list, be sure to update * the property table in module/zcommon/zfs_prop.c. */ typedef enum { ZPROP_CONT = -2, ZPROP_INVAL = -1, ZFS_PROP_TYPE = 0, ZFS_PROP_CREATION, ZFS_PROP_USED, ZFS_PROP_AVAILABLE, ZFS_PROP_REFERENCED, ZFS_PROP_COMPRESSRATIO, ZFS_PROP_MOUNTED, ZFS_PROP_ORIGIN, ZFS_PROP_QUOTA, ZFS_PROP_RESERVATION, ZFS_PROP_VOLSIZE, ZFS_PROP_VOLBLOCKSIZE, ZFS_PROP_RECORDSIZE, ZFS_PROP_MOUNTPOINT, ZFS_PROP_SHARENFS, ZFS_PROP_CHECKSUM, ZFS_PROP_COMPRESSION, ZFS_PROP_ATIME, ZFS_PROP_DEVICES, ZFS_PROP_EXEC, ZFS_PROP_SETUID, ZFS_PROP_READONLY, ZFS_PROP_ZONED, ZFS_PROP_SNAPDIR, ZFS_PROP_ACLMODE, ZFS_PROP_ACLINHERIT, ZFS_PROP_CREATETXG, ZFS_PROP_NAME, /* not exposed to the user */ ZFS_PROP_CANMOUNT, ZFS_PROP_ISCSIOPTIONS, /* not exposed to the user */ ZFS_PROP_XATTR, ZFS_PROP_NUMCLONES, /* not exposed to the user */ ZFS_PROP_COPIES, ZFS_PROP_VERSION, ZFS_PROP_UTF8ONLY, ZFS_PROP_NORMALIZE, ZFS_PROP_CASE, ZFS_PROP_VSCAN, ZFS_PROP_NBMAND, ZFS_PROP_SHARESMB, ZFS_PROP_REFQUOTA, ZFS_PROP_REFRESERVATION, ZFS_PROP_GUID, ZFS_PROP_PRIMARYCACHE, ZFS_PROP_SECONDARYCACHE, ZFS_PROP_USEDSNAP, ZFS_PROP_USEDDS, ZFS_PROP_USEDCHILD, ZFS_PROP_USEDREFRESERV, ZFS_PROP_USERACCOUNTING, /* not exposed to the user */ ZFS_PROP_STMF_SHAREINFO, /* not exposed to the user */ ZFS_PROP_DEFER_DESTROY, ZFS_PROP_USERREFS, ZFS_PROP_LOGBIAS, ZFS_PROP_UNIQUE, /* not exposed to the user */ ZFS_PROP_OBJSETID, ZFS_PROP_DEDUP, ZFS_PROP_MLSLABEL, ZFS_PROP_SYNC, ZFS_PROP_DNODESIZE, ZFS_PROP_REFRATIO, ZFS_PROP_WRITTEN, ZFS_PROP_CLONES, ZFS_PROP_LOGICALUSED, ZFS_PROP_LOGICALREFERENCED, ZFS_PROP_INCONSISTENT, /* not exposed to the user */ ZFS_PROP_VOLMODE, ZFS_PROP_FILESYSTEM_LIMIT, ZFS_PROP_SNAPSHOT_LIMIT, ZFS_PROP_FILESYSTEM_COUNT, ZFS_PROP_SNAPSHOT_COUNT, ZFS_PROP_SNAPDEV, ZFS_PROP_ACLTYPE, ZFS_PROP_SELINUX_CONTEXT, ZFS_PROP_SELINUX_FSCONTEXT, ZFS_PROP_SELINUX_DEFCONTEXT, ZFS_PROP_SELINUX_ROOTCONTEXT, ZFS_PROP_RELATIME, ZFS_PROP_REDUNDANT_METADATA, ZFS_PROP_OVERLAY, ZFS_PROP_PREV_SNAP, ZFS_PROP_RECEIVE_RESUME_TOKEN, ZFS_PROP_ENCRYPTION, ZFS_PROP_KEYLOCATION, ZFS_PROP_KEYFORMAT, ZFS_PROP_PBKDF2_SALT, ZFS_PROP_PBKDF2_ITERS, ZFS_PROP_ENCRYPTION_ROOT, ZFS_PROP_KEY_GUID, ZFS_PROP_KEYSTATUS, ZFS_PROP_REMAPTXG, /* obsolete - no longer used */ ZFS_PROP_SPECIAL_SMALL_BLOCKS, ZFS_PROP_IVSET_GUID, /* not exposed to the user */ ZFS_PROP_REDACTED, ZFS_PROP_REDACT_SNAPS, ZFS_NUM_PROPS } zfs_prop_t; typedef enum { ZFS_PROP_USERUSED, ZFS_PROP_USERQUOTA, ZFS_PROP_GROUPUSED, ZFS_PROP_GROUPQUOTA, ZFS_PROP_USEROBJUSED, ZFS_PROP_USEROBJQUOTA, ZFS_PROP_GROUPOBJUSED, ZFS_PROP_GROUPOBJQUOTA, ZFS_PROP_PROJECTUSED, ZFS_PROP_PROJECTQUOTA, ZFS_PROP_PROJECTOBJUSED, ZFS_PROP_PROJECTOBJQUOTA, ZFS_NUM_USERQUOTA_PROPS } zfs_userquota_prop_t; extern const char *zfs_userquota_prop_prefixes[ZFS_NUM_USERQUOTA_PROPS]; /* * Pool properties are identified by these constants and must be added to the * end of this list to ensure that external consumers are not affected * by the change. Properties must be registered in zfs_prop_init(). */ typedef enum { ZPOOL_PROP_INVAL = -1, ZPOOL_PROP_NAME, ZPOOL_PROP_SIZE, ZPOOL_PROP_CAPACITY, ZPOOL_PROP_ALTROOT, ZPOOL_PROP_HEALTH, ZPOOL_PROP_GUID, ZPOOL_PROP_VERSION, ZPOOL_PROP_BOOTFS, ZPOOL_PROP_DELEGATION, ZPOOL_PROP_AUTOREPLACE, ZPOOL_PROP_CACHEFILE, ZPOOL_PROP_FAILUREMODE, ZPOOL_PROP_LISTSNAPS, ZPOOL_PROP_AUTOEXPAND, ZPOOL_PROP_DEDUPDITTO, ZPOOL_PROP_DEDUPRATIO, ZPOOL_PROP_FREE, ZPOOL_PROP_ALLOCATED, ZPOOL_PROP_READONLY, ZPOOL_PROP_ASHIFT, ZPOOL_PROP_COMMENT, ZPOOL_PROP_EXPANDSZ, ZPOOL_PROP_FREEING, ZPOOL_PROP_FRAGMENTATION, ZPOOL_PROP_LEAKED, ZPOOL_PROP_MAXBLOCKSIZE, ZPOOL_PROP_TNAME, ZPOOL_PROP_MAXDNODESIZE, ZPOOL_PROP_MULTIHOST, ZPOOL_PROP_CHECKPOINT, ZPOOL_PROP_LOAD_GUID, ZPOOL_PROP_AUTOTRIM, ZPOOL_NUM_PROPS } zpool_prop_t; -/* Small enough to not hog a whole line of printout in zpool(1M). */ +/* Small enough to not hog a whole line of printout in zpool(8). */ #define ZPROP_MAX_COMMENT 32 #define ZPROP_VALUE "value" #define ZPROP_SOURCE "source" typedef enum { ZPROP_SRC_NONE = 0x1, ZPROP_SRC_DEFAULT = 0x2, ZPROP_SRC_TEMPORARY = 0x4, ZPROP_SRC_LOCAL = 0x8, ZPROP_SRC_INHERITED = 0x10, ZPROP_SRC_RECEIVED = 0x20 } zprop_source_t; #define ZPROP_SRC_ALL 0x3f #define ZPROP_SOURCE_VAL_RECVD "$recvd" #define ZPROP_N_MORE_ERRORS "N_MORE_ERRORS" /* * Dataset flag implemented as a special entry in the props zap object * indicating that the dataset has received properties on or after * SPA_VERSION_RECVD_PROPS. The first such receive blows away local properties * just as it did in earlier versions, and thereafter, local properties are * preserved. */ #define ZPROP_HAS_RECVD "$hasrecvd" typedef enum { ZPROP_ERR_NOCLEAR = 0x1, /* failure to clear existing props */ ZPROP_ERR_NORESTORE = 0x2 /* failure to restore props on error */ } zprop_errflags_t; typedef int (*zprop_func)(int, void *); /* * Properties to be set on the root file system of a new pool * are stuffed into their own nvlist, which is then included in * the properties nvlist with the pool properties. */ #define ZPOOL_ROOTFS_PROPS "root-props-nvl" /* * Length of 'written@' and 'written#' */ #define ZFS_WRITTEN_PROP_PREFIX_LEN 8 /* * Dataset property functions shared between libzfs and kernel. */ const char *zfs_prop_default_string(zfs_prop_t); uint64_t zfs_prop_default_numeric(zfs_prop_t); boolean_t zfs_prop_readonly(zfs_prop_t); boolean_t zfs_prop_visible(zfs_prop_t prop); boolean_t zfs_prop_inheritable(zfs_prop_t); boolean_t zfs_prop_setonce(zfs_prop_t); boolean_t zfs_prop_encryption_key_param(zfs_prop_t); boolean_t zfs_prop_valid_keylocation(const char *, boolean_t); const char *zfs_prop_to_name(zfs_prop_t); zfs_prop_t zfs_name_to_prop(const char *); boolean_t zfs_prop_user(const char *); boolean_t zfs_prop_userquota(const char *); boolean_t zfs_prop_written(const char *); int zfs_prop_index_to_string(zfs_prop_t, uint64_t, const char **); int zfs_prop_string_to_index(zfs_prop_t, const char *, uint64_t *); uint64_t zfs_prop_random_value(zfs_prop_t, uint64_t seed); boolean_t zfs_prop_valid_for_type(int, zfs_type_t, boolean_t); /* * Pool property functions shared between libzfs and kernel. */ zpool_prop_t zpool_name_to_prop(const char *); const char *zpool_prop_to_name(zpool_prop_t); const char *zpool_prop_default_string(zpool_prop_t); uint64_t zpool_prop_default_numeric(zpool_prop_t); boolean_t zpool_prop_readonly(zpool_prop_t); boolean_t zpool_prop_setonce(zpool_prop_t); boolean_t zpool_prop_feature(const char *); boolean_t zpool_prop_unsupported(const char *); int zpool_prop_index_to_string(zpool_prop_t, uint64_t, const char **); int zpool_prop_string_to_index(zpool_prop_t, const char *, uint64_t *); uint64_t zpool_prop_random_value(zpool_prop_t, uint64_t seed); /* * Definitions for the Delegation. */ typedef enum { ZFS_DELEG_WHO_UNKNOWN = 0, ZFS_DELEG_USER = 'u', ZFS_DELEG_USER_SETS = 'U', ZFS_DELEG_GROUP = 'g', ZFS_DELEG_GROUP_SETS = 'G', ZFS_DELEG_EVERYONE = 'e', ZFS_DELEG_EVERYONE_SETS = 'E', ZFS_DELEG_CREATE = 'c', ZFS_DELEG_CREATE_SETS = 'C', ZFS_DELEG_NAMED_SET = 's', ZFS_DELEG_NAMED_SET_SETS = 'S' } zfs_deleg_who_type_t; typedef enum { ZFS_DELEG_NONE = 0, ZFS_DELEG_PERM_LOCAL = 1, ZFS_DELEG_PERM_DESCENDENT = 2, ZFS_DELEG_PERM_LOCALDESCENDENT = 3, ZFS_DELEG_PERM_CREATE = 4 } zfs_deleg_inherit_t; #define ZFS_DELEG_PERM_UID "uid" #define ZFS_DELEG_PERM_GID "gid" #define ZFS_DELEG_PERM_GROUPS "groups" #define ZFS_MLSLABEL_DEFAULT "none" #define ZFS_SMB_ACL_SRC "src" #define ZFS_SMB_ACL_TARGET "target" typedef enum { ZFS_CANMOUNT_OFF = 0, ZFS_CANMOUNT_ON = 1, ZFS_CANMOUNT_NOAUTO = 2 } zfs_canmount_type_t; typedef enum { ZFS_LOGBIAS_LATENCY = 0, ZFS_LOGBIAS_THROUGHPUT = 1 } zfs_logbias_op_t; typedef enum zfs_share_op { ZFS_SHARE_NFS = 0, ZFS_UNSHARE_NFS = 1, ZFS_SHARE_SMB = 2, ZFS_UNSHARE_SMB = 3 } zfs_share_op_t; typedef enum zfs_smb_acl_op { ZFS_SMB_ACL_ADD, ZFS_SMB_ACL_REMOVE, ZFS_SMB_ACL_RENAME, ZFS_SMB_ACL_PURGE } zfs_smb_acl_op_t; typedef enum zfs_cache_type { ZFS_CACHE_NONE = 0, ZFS_CACHE_METADATA = 1, ZFS_CACHE_ALL = 2 } zfs_cache_type_t; typedef enum { ZFS_SYNC_STANDARD = 0, ZFS_SYNC_ALWAYS = 1, ZFS_SYNC_DISABLED = 2 } zfs_sync_type_t; typedef enum { ZFS_XATTR_OFF = 0, ZFS_XATTR_DIR = 1, ZFS_XATTR_SA = 2 } zfs_xattr_type_t; typedef enum { ZFS_DNSIZE_LEGACY = 0, ZFS_DNSIZE_AUTO = 1, ZFS_DNSIZE_1K = 1024, ZFS_DNSIZE_2K = 2048, ZFS_DNSIZE_4K = 4096, ZFS_DNSIZE_8K = 8192, ZFS_DNSIZE_16K = 16384 } zfs_dnsize_type_t; typedef enum { ZFS_REDUNDANT_METADATA_ALL, ZFS_REDUNDANT_METADATA_MOST } zfs_redundant_metadata_type_t; typedef enum { ZFS_VOLMODE_DEFAULT = 0, ZFS_VOLMODE_GEOM = 1, ZFS_VOLMODE_DEV = 2, ZFS_VOLMODE_NONE = 3 } zfs_volmode_t; typedef enum zfs_keystatus { ZFS_KEYSTATUS_NONE = 0, ZFS_KEYSTATUS_UNAVAILABLE, ZFS_KEYSTATUS_AVAILABLE, } zfs_keystatus_t; typedef enum zfs_keyformat { ZFS_KEYFORMAT_NONE = 0, ZFS_KEYFORMAT_RAW, ZFS_KEYFORMAT_HEX, ZFS_KEYFORMAT_PASSPHRASE, ZFS_KEYFORMAT_FORMATS } zfs_keyformat_t; typedef enum zfs_key_location { ZFS_KEYLOCATION_NONE = 0, ZFS_KEYLOCATION_PROMPT, ZFS_KEYLOCATION_URI, ZFS_KEYLOCATION_LOCATIONS } zfs_keylocation_t; #define DEFAULT_PBKDF2_ITERATIONS 350000 #define MIN_PBKDF2_ITERATIONS 100000 /* * On-disk version number. */ #define SPA_VERSION_1 1ULL #define SPA_VERSION_2 2ULL #define SPA_VERSION_3 3ULL #define SPA_VERSION_4 4ULL #define SPA_VERSION_5 5ULL #define SPA_VERSION_6 6ULL #define SPA_VERSION_7 7ULL #define SPA_VERSION_8 8ULL #define SPA_VERSION_9 9ULL #define SPA_VERSION_10 10ULL #define SPA_VERSION_11 11ULL #define SPA_VERSION_12 12ULL #define SPA_VERSION_13 13ULL #define SPA_VERSION_14 14ULL #define SPA_VERSION_15 15ULL #define SPA_VERSION_16 16ULL #define SPA_VERSION_17 17ULL #define SPA_VERSION_18 18ULL #define SPA_VERSION_19 19ULL #define SPA_VERSION_20 20ULL #define SPA_VERSION_21 21ULL #define SPA_VERSION_22 22ULL #define SPA_VERSION_23 23ULL #define SPA_VERSION_24 24ULL #define SPA_VERSION_25 25ULL #define SPA_VERSION_26 26ULL #define SPA_VERSION_27 27ULL #define SPA_VERSION_28 28ULL #define SPA_VERSION_5000 5000ULL /* * The incrementing pool version number has been replaced by pool feature * flags. For more details, see zfeature.c. */ #define SPA_VERSION SPA_VERSION_5000 #define SPA_VERSION_STRING "5000" /* * Symbolic names for the changes that caused a SPA_VERSION switch. * Used in the code when checking for presence or absence of a feature. * Feel free to define multiple symbolic names for each version if there * were multiple changes to on-disk structures during that version. * * NOTE: When checking the current SPA_VERSION in your code, be sure * to use spa_version() since it reports the version of the * last synced uberblock. Checking the in-flight version can * be dangerous in some cases. */ #define SPA_VERSION_INITIAL SPA_VERSION_1 #define SPA_VERSION_DITTO_BLOCKS SPA_VERSION_2 #define SPA_VERSION_SPARES SPA_VERSION_3 #define SPA_VERSION_RAIDZ2 SPA_VERSION_3 #define SPA_VERSION_BPOBJ_ACCOUNT SPA_VERSION_3 #define SPA_VERSION_RAIDZ_DEFLATE SPA_VERSION_3 #define SPA_VERSION_DNODE_BYTES SPA_VERSION_3 #define SPA_VERSION_ZPOOL_HISTORY SPA_VERSION_4 #define SPA_VERSION_GZIP_COMPRESSION SPA_VERSION_5 #define SPA_VERSION_BOOTFS SPA_VERSION_6 #define SPA_VERSION_SLOGS SPA_VERSION_7 #define SPA_VERSION_DELEGATED_PERMS SPA_VERSION_8 #define SPA_VERSION_FUID SPA_VERSION_9 #define SPA_VERSION_REFRESERVATION SPA_VERSION_9 #define SPA_VERSION_REFQUOTA SPA_VERSION_9 #define SPA_VERSION_UNIQUE_ACCURATE SPA_VERSION_9 #define SPA_VERSION_L2CACHE SPA_VERSION_10 #define SPA_VERSION_NEXT_CLONES SPA_VERSION_11 #define SPA_VERSION_ORIGIN SPA_VERSION_11 #define SPA_VERSION_DSL_SCRUB SPA_VERSION_11 #define SPA_VERSION_SNAP_PROPS SPA_VERSION_12 #define SPA_VERSION_USED_BREAKDOWN SPA_VERSION_13 #define SPA_VERSION_PASSTHROUGH_X SPA_VERSION_14 #define SPA_VERSION_USERSPACE SPA_VERSION_15 #define SPA_VERSION_STMF_PROP SPA_VERSION_16 #define SPA_VERSION_RAIDZ3 SPA_VERSION_17 #define SPA_VERSION_USERREFS SPA_VERSION_18 #define SPA_VERSION_HOLES SPA_VERSION_19 #define SPA_VERSION_ZLE_COMPRESSION SPA_VERSION_20 #define SPA_VERSION_DEDUP SPA_VERSION_21 #define SPA_VERSION_RECVD_PROPS SPA_VERSION_22 #define SPA_VERSION_SLIM_ZIL SPA_VERSION_23 #define SPA_VERSION_SA SPA_VERSION_24 #define SPA_VERSION_SCAN SPA_VERSION_25 #define SPA_VERSION_DIR_CLONES SPA_VERSION_26 #define SPA_VERSION_DEADLISTS SPA_VERSION_26 #define SPA_VERSION_FAST_SNAP SPA_VERSION_27 #define SPA_VERSION_MULTI_REPLACE SPA_VERSION_28 #define SPA_VERSION_BEFORE_FEATURES SPA_VERSION_28 #define SPA_VERSION_FEATURES SPA_VERSION_5000 #define SPA_VERSION_IS_SUPPORTED(v) \ (((v) >= SPA_VERSION_INITIAL && (v) <= SPA_VERSION_BEFORE_FEATURES) || \ ((v) >= SPA_VERSION_FEATURES && (v) <= SPA_VERSION)) /* * ZPL version - rev'd whenever an incompatible on-disk format change * occurs. This is independent of SPA/DMU/ZAP versioning. You must * also update the version_table[] and help message in zfs_prop.c. */ #define ZPL_VERSION_1 1ULL #define ZPL_VERSION_2 2ULL #define ZPL_VERSION_3 3ULL #define ZPL_VERSION_4 4ULL #define ZPL_VERSION_5 5ULL #define ZPL_VERSION ZPL_VERSION_5 #define ZPL_VERSION_STRING "5" #define ZPL_VERSION_INITIAL ZPL_VERSION_1 #define ZPL_VERSION_DIRENT_TYPE ZPL_VERSION_2 #define ZPL_VERSION_FUID ZPL_VERSION_3 #define ZPL_VERSION_NORMALIZATION ZPL_VERSION_3 #define ZPL_VERSION_SYSATTR ZPL_VERSION_3 #define ZPL_VERSION_USERSPACE ZPL_VERSION_4 #define ZPL_VERSION_SA ZPL_VERSION_5 /* Persistent L2ARC version */ #define L2ARC_PERSISTENT_VERSION_1 1ULL #define L2ARC_PERSISTENT_VERSION L2ARC_PERSISTENT_VERSION_1 #define L2ARC_PERSISTENT_VERSION_STRING "1" /* Rewind policy information */ #define ZPOOL_NO_REWIND 1 /* No policy - default behavior */ #define ZPOOL_NEVER_REWIND 2 /* Do not search for best txg or rewind */ #define ZPOOL_TRY_REWIND 4 /* Search for best txg, but do not rewind */ #define ZPOOL_DO_REWIND 8 /* Rewind to best txg w/in deferred frees */ #define ZPOOL_EXTREME_REWIND 16 /* Allow extreme measures to find best txg */ #define ZPOOL_REWIND_MASK 28 /* All the possible rewind bits */ #define ZPOOL_REWIND_POLICIES 31 /* All the possible policy bits */ typedef struct zpool_load_policy { uint32_t zlp_rewind; /* rewind policy requested */ uint64_t zlp_maxmeta; /* max acceptable meta-data errors */ uint64_t zlp_maxdata; /* max acceptable data errors */ uint64_t zlp_txg; /* specific txg to load */ } zpool_load_policy_t; /* * The following are configuration names used in the nvlist describing a pool's * configuration. New on-disk names should be prefixed with ":" * (e.g. "org.openzfs:") to avoid conflicting names being developed * independently. */ #define ZPOOL_CONFIG_VERSION "version" #define ZPOOL_CONFIG_POOL_NAME "name" #define ZPOOL_CONFIG_POOL_STATE "state" #define ZPOOL_CONFIG_POOL_TXG "txg" #define ZPOOL_CONFIG_POOL_GUID "pool_guid" #define ZPOOL_CONFIG_CREATE_TXG "create_txg" #define ZPOOL_CONFIG_TOP_GUID "top_guid" #define ZPOOL_CONFIG_VDEV_TREE "vdev_tree" #define ZPOOL_CONFIG_TYPE "type" #define ZPOOL_CONFIG_CHILDREN "children" #define ZPOOL_CONFIG_ID "id" #define ZPOOL_CONFIG_GUID "guid" #define ZPOOL_CONFIG_INDIRECT_OBJECT "com.delphix:indirect_object" #define ZPOOL_CONFIG_INDIRECT_BIRTHS "com.delphix:indirect_births" #define ZPOOL_CONFIG_PREV_INDIRECT_VDEV "com.delphix:prev_indirect_vdev" #define ZPOOL_CONFIG_PATH "path" #define ZPOOL_CONFIG_DEVID "devid" #define ZPOOL_CONFIG_METASLAB_ARRAY "metaslab_array" #define ZPOOL_CONFIG_METASLAB_SHIFT "metaslab_shift" #define ZPOOL_CONFIG_ASHIFT "ashift" #define ZPOOL_CONFIG_ASIZE "asize" #define ZPOOL_CONFIG_DTL "DTL" #define ZPOOL_CONFIG_SCAN_STATS "scan_stats" /* not stored on disk */ #define ZPOOL_CONFIG_REMOVAL_STATS "removal_stats" /* not stored on disk */ #define ZPOOL_CONFIG_CHECKPOINT_STATS "checkpoint_stats" /* not on disk */ #define ZPOOL_CONFIG_VDEV_STATS "vdev_stats" /* not stored on disk */ #define ZPOOL_CONFIG_INDIRECT_SIZE "indirect_size" /* not stored on disk */ /* container nvlist of extended stats */ #define ZPOOL_CONFIG_VDEV_STATS_EX "vdev_stats_ex" /* Active queue read/write stats */ #define ZPOOL_CONFIG_VDEV_SYNC_R_ACTIVE_QUEUE "vdev_sync_r_active_queue" #define ZPOOL_CONFIG_VDEV_SYNC_W_ACTIVE_QUEUE "vdev_sync_w_active_queue" #define ZPOOL_CONFIG_VDEV_ASYNC_R_ACTIVE_QUEUE "vdev_async_r_active_queue" #define ZPOOL_CONFIG_VDEV_ASYNC_W_ACTIVE_QUEUE "vdev_async_w_active_queue" #define ZPOOL_CONFIG_VDEV_SCRUB_ACTIVE_QUEUE "vdev_async_scrub_active_queue" #define ZPOOL_CONFIG_VDEV_TRIM_ACTIVE_QUEUE "vdev_async_trim_active_queue" /* Queue sizes */ #define ZPOOL_CONFIG_VDEV_SYNC_R_PEND_QUEUE "vdev_sync_r_pend_queue" #define ZPOOL_CONFIG_VDEV_SYNC_W_PEND_QUEUE "vdev_sync_w_pend_queue" #define ZPOOL_CONFIG_VDEV_ASYNC_R_PEND_QUEUE "vdev_async_r_pend_queue" #define ZPOOL_CONFIG_VDEV_ASYNC_W_PEND_QUEUE "vdev_async_w_pend_queue" #define ZPOOL_CONFIG_VDEV_SCRUB_PEND_QUEUE "vdev_async_scrub_pend_queue" #define ZPOOL_CONFIG_VDEV_TRIM_PEND_QUEUE "vdev_async_trim_pend_queue" /* Latency read/write histogram stats */ #define ZPOOL_CONFIG_VDEV_TOT_R_LAT_HISTO "vdev_tot_r_lat_histo" #define ZPOOL_CONFIG_VDEV_TOT_W_LAT_HISTO "vdev_tot_w_lat_histo" #define ZPOOL_CONFIG_VDEV_DISK_R_LAT_HISTO "vdev_disk_r_lat_histo" #define ZPOOL_CONFIG_VDEV_DISK_W_LAT_HISTO "vdev_disk_w_lat_histo" #define ZPOOL_CONFIG_VDEV_SYNC_R_LAT_HISTO "vdev_sync_r_lat_histo" #define ZPOOL_CONFIG_VDEV_SYNC_W_LAT_HISTO "vdev_sync_w_lat_histo" #define ZPOOL_CONFIG_VDEV_ASYNC_R_LAT_HISTO "vdev_async_r_lat_histo" #define ZPOOL_CONFIG_VDEV_ASYNC_W_LAT_HISTO "vdev_async_w_lat_histo" #define ZPOOL_CONFIG_VDEV_SCRUB_LAT_HISTO "vdev_scrub_histo" #define ZPOOL_CONFIG_VDEV_TRIM_LAT_HISTO "vdev_trim_histo" /* Request size histograms */ #define ZPOOL_CONFIG_VDEV_SYNC_IND_R_HISTO "vdev_sync_ind_r_histo" #define ZPOOL_CONFIG_VDEV_SYNC_IND_W_HISTO "vdev_sync_ind_w_histo" #define ZPOOL_CONFIG_VDEV_ASYNC_IND_R_HISTO "vdev_async_ind_r_histo" #define ZPOOL_CONFIG_VDEV_ASYNC_IND_W_HISTO "vdev_async_ind_w_histo" #define ZPOOL_CONFIG_VDEV_IND_SCRUB_HISTO "vdev_ind_scrub_histo" #define ZPOOL_CONFIG_VDEV_IND_TRIM_HISTO "vdev_ind_trim_histo" #define ZPOOL_CONFIG_VDEV_SYNC_AGG_R_HISTO "vdev_sync_agg_r_histo" #define ZPOOL_CONFIG_VDEV_SYNC_AGG_W_HISTO "vdev_sync_agg_w_histo" #define ZPOOL_CONFIG_VDEV_ASYNC_AGG_R_HISTO "vdev_async_agg_r_histo" #define ZPOOL_CONFIG_VDEV_ASYNC_AGG_W_HISTO "vdev_async_agg_w_histo" #define ZPOOL_CONFIG_VDEV_AGG_SCRUB_HISTO "vdev_agg_scrub_histo" #define ZPOOL_CONFIG_VDEV_AGG_TRIM_HISTO "vdev_agg_trim_histo" /* Number of slow IOs */ #define ZPOOL_CONFIG_VDEV_SLOW_IOS "vdev_slow_ios" /* vdev enclosure sysfs path */ #define ZPOOL_CONFIG_VDEV_ENC_SYSFS_PATH "vdev_enc_sysfs_path" #define ZPOOL_CONFIG_WHOLE_DISK "whole_disk" #define ZPOOL_CONFIG_ERRCOUNT "error_count" #define ZPOOL_CONFIG_NOT_PRESENT "not_present" #define ZPOOL_CONFIG_SPARES "spares" #define ZPOOL_CONFIG_IS_SPARE "is_spare" #define ZPOOL_CONFIG_NPARITY "nparity" #define ZPOOL_CONFIG_HOSTID "hostid" #define ZPOOL_CONFIG_HOSTNAME "hostname" #define ZPOOL_CONFIG_LOADED_TIME "initial_load_time" #define ZPOOL_CONFIG_UNSPARE "unspare" #define ZPOOL_CONFIG_PHYS_PATH "phys_path" #define ZPOOL_CONFIG_IS_LOG "is_log" #define ZPOOL_CONFIG_L2CACHE "l2cache" #define ZPOOL_CONFIG_HOLE_ARRAY "hole_array" #define ZPOOL_CONFIG_VDEV_CHILDREN "vdev_children" #define ZPOOL_CONFIG_IS_HOLE "is_hole" #define ZPOOL_CONFIG_DDT_HISTOGRAM "ddt_histogram" #define ZPOOL_CONFIG_DDT_OBJ_STATS "ddt_object_stats" #define ZPOOL_CONFIG_DDT_STATS "ddt_stats" #define ZPOOL_CONFIG_SPLIT "splitcfg" #define ZPOOL_CONFIG_ORIG_GUID "orig_guid" #define ZPOOL_CONFIG_SPLIT_GUID "split_guid" #define ZPOOL_CONFIG_SPLIT_LIST "guid_list" #define ZPOOL_CONFIG_REMOVING "removing" #define ZPOOL_CONFIG_RESILVER_TXG "resilver_txg" #define ZPOOL_CONFIG_REBUILD_TXG "rebuild_txg" #define ZPOOL_CONFIG_COMMENT "comment" #define ZPOOL_CONFIG_SUSPENDED "suspended" /* not stored on disk */ #define ZPOOL_CONFIG_SUSPENDED_REASON "suspended_reason" /* not stored */ #define ZPOOL_CONFIG_TIMESTAMP "timestamp" /* not stored on disk */ #define ZPOOL_CONFIG_BOOTFS "bootfs" /* not stored on disk */ #define ZPOOL_CONFIG_MISSING_DEVICES "missing_vdevs" /* not stored on disk */ #define ZPOOL_CONFIG_LOAD_INFO "load_info" /* not stored on disk */ #define ZPOOL_CONFIG_REWIND_INFO "rewind_info" /* not stored on disk */ #define ZPOOL_CONFIG_UNSUP_FEAT "unsup_feat" /* not stored on disk */ #define ZPOOL_CONFIG_ENABLED_FEAT "enabled_feat" /* not stored on disk */ #define ZPOOL_CONFIG_CAN_RDONLY "can_rdonly" /* not stored on disk */ #define ZPOOL_CONFIG_FEATURES_FOR_READ "features_for_read" #define ZPOOL_CONFIG_FEATURE_STATS "feature_stats" /* not stored on disk */ #define ZPOOL_CONFIG_ERRATA "errata" /* not stored on disk */ #define ZPOOL_CONFIG_VDEV_TOP_ZAP "com.delphix:vdev_zap_top" #define ZPOOL_CONFIG_VDEV_LEAF_ZAP "com.delphix:vdev_zap_leaf" #define ZPOOL_CONFIG_HAS_PER_VDEV_ZAPS "com.delphix:has_per_vdev_zaps" #define ZPOOL_CONFIG_RESILVER_DEFER "com.datto:resilver_defer" #define ZPOOL_CONFIG_CACHEFILE "cachefile" /* not stored on disk */ #define ZPOOL_CONFIG_MMP_STATE "mmp_state" /* not stored on disk */ #define ZPOOL_CONFIG_MMP_TXG "mmp_txg" /* not stored on disk */ #define ZPOOL_CONFIG_MMP_SEQ "mmp_seq" /* not stored on disk */ #define ZPOOL_CONFIG_MMP_HOSTNAME "mmp_hostname" /* not stored on disk */ #define ZPOOL_CONFIG_MMP_HOSTID "mmp_hostid" /* not stored on disk */ #define ZPOOL_CONFIG_ALLOCATION_BIAS "alloc_bias" /* not stored on disk */ #define ZPOOL_CONFIG_EXPANSION_TIME "expansion_time" /* not stored */ #define ZPOOL_CONFIG_REBUILD_STATS "org.openzfs:rebuild_stats" /* * The persistent vdev state is stored as separate values rather than a single * 'vdev_state' entry. This is because a device can be in multiple states, such * as offline and degraded. */ #define ZPOOL_CONFIG_OFFLINE "offline" #define ZPOOL_CONFIG_FAULTED "faulted" #define ZPOOL_CONFIG_DEGRADED "degraded" #define ZPOOL_CONFIG_REMOVED "removed" #define ZPOOL_CONFIG_FRU "fru" #define ZPOOL_CONFIG_AUX_STATE "aux_state" /* Pool load policy parameters */ #define ZPOOL_LOAD_POLICY "load-policy" #define ZPOOL_LOAD_REWIND_POLICY "load-rewind-policy" #define ZPOOL_LOAD_REQUEST_TXG "load-request-txg" #define ZPOOL_LOAD_META_THRESH "load-meta-thresh" #define ZPOOL_LOAD_DATA_THRESH "load-data-thresh" /* Rewind data discovered */ #define ZPOOL_CONFIG_LOAD_TIME "rewind_txg_ts" #define ZPOOL_CONFIG_LOAD_DATA_ERRORS "verify_data_errors" #define ZPOOL_CONFIG_REWIND_TIME "seconds_of_rewind" #define VDEV_TYPE_ROOT "root" #define VDEV_TYPE_MIRROR "mirror" #define VDEV_TYPE_REPLACING "replacing" #define VDEV_TYPE_RAIDZ "raidz" #define VDEV_TYPE_DISK "disk" #define VDEV_TYPE_FILE "file" #define VDEV_TYPE_MISSING "missing" #define VDEV_TYPE_HOLE "hole" #define VDEV_TYPE_SPARE "spare" #define VDEV_TYPE_LOG "log" #define VDEV_TYPE_L2CACHE "l2cache" #define VDEV_TYPE_INDIRECT "indirect" /* VDEV_TOP_ZAP_* are used in top-level vdev ZAP objects. */ #define VDEV_TOP_ZAP_INDIRECT_OBSOLETE_SM \ "com.delphix:indirect_obsolete_sm" #define VDEV_TOP_ZAP_OBSOLETE_COUNTS_ARE_PRECISE \ "com.delphix:obsolete_counts_are_precise" #define VDEV_TOP_ZAP_POOL_CHECKPOINT_SM \ "com.delphix:pool_checkpoint_sm" #define VDEV_TOP_ZAP_MS_UNFLUSHED_PHYS_TXGS \ "com.delphix:ms_unflushed_phys_txgs" #define VDEV_TOP_ZAP_VDEV_REBUILD_PHYS \ "org.openzfs:vdev_rebuild" #define VDEV_TOP_ZAP_ALLOCATION_BIAS \ "org.zfsonlinux:allocation_bias" /* vdev metaslab allocation bias */ #define VDEV_ALLOC_BIAS_LOG "log" #define VDEV_ALLOC_BIAS_SPECIAL "special" #define VDEV_ALLOC_BIAS_DEDUP "dedup" /* vdev initialize state */ #define VDEV_LEAF_ZAP_INITIALIZE_LAST_OFFSET \ "com.delphix:next_offset_to_initialize" #define VDEV_LEAF_ZAP_INITIALIZE_STATE \ "com.delphix:vdev_initialize_state" #define VDEV_LEAF_ZAP_INITIALIZE_ACTION_TIME \ "com.delphix:vdev_initialize_action_time" /* vdev TRIM state */ #define VDEV_LEAF_ZAP_TRIM_LAST_OFFSET \ "org.zfsonlinux:next_offset_to_trim" #define VDEV_LEAF_ZAP_TRIM_STATE \ "org.zfsonlinux:vdev_trim_state" #define VDEV_LEAF_ZAP_TRIM_ACTION_TIME \ "org.zfsonlinux:vdev_trim_action_time" #define VDEV_LEAF_ZAP_TRIM_RATE \ "org.zfsonlinux:vdev_trim_rate" #define VDEV_LEAF_ZAP_TRIM_PARTIAL \ "org.zfsonlinux:vdev_trim_partial" #define VDEV_LEAF_ZAP_TRIM_SECURE \ "org.zfsonlinux:vdev_trim_secure" /* * This is needed in userland to report the minimum necessary device size. */ #define SPA_MINDEVSIZE (64ULL << 20) /* * Set if the fragmentation has not yet been calculated. This can happen * because the space maps have not been upgraded or the histogram feature * is not enabled. */ #define ZFS_FRAG_INVALID UINT64_MAX /* * The location of the pool configuration repository, shared between kernel and * userland. */ #define ZPOOL_CACHE_BOOT "/boot/zfs/zpool.cache" #define ZPOOL_CACHE "/etc/zfs/zpool.cache" /* * vdev states are ordered from least to most healthy. * A vdev that's CANT_OPEN or below is considered unusable. */ typedef enum vdev_state { VDEV_STATE_UNKNOWN = 0, /* Uninitialized vdev */ VDEV_STATE_CLOSED, /* Not currently open */ VDEV_STATE_OFFLINE, /* Not allowed to open */ VDEV_STATE_REMOVED, /* Explicitly removed from system */ VDEV_STATE_CANT_OPEN, /* Tried to open, but failed */ VDEV_STATE_FAULTED, /* External request to fault device */ VDEV_STATE_DEGRADED, /* Replicated vdev with unhealthy kids */ VDEV_STATE_HEALTHY /* Presumed good */ } vdev_state_t; #define VDEV_STATE_ONLINE VDEV_STATE_HEALTHY /* * vdev aux states. When a vdev is in the CANT_OPEN state, the aux field * of the vdev stats structure uses these constants to distinguish why. */ typedef enum vdev_aux { VDEV_AUX_NONE, /* no error */ VDEV_AUX_OPEN_FAILED, /* ldi_open_*() or vn_open() failed */ VDEV_AUX_CORRUPT_DATA, /* bad label or disk contents */ VDEV_AUX_NO_REPLICAS, /* insufficient number of replicas */ VDEV_AUX_BAD_GUID_SUM, /* vdev guid sum doesn't match */ VDEV_AUX_TOO_SMALL, /* vdev size is too small */ VDEV_AUX_BAD_LABEL, /* the label is OK but invalid */ VDEV_AUX_VERSION_NEWER, /* on-disk version is too new */ VDEV_AUX_VERSION_OLDER, /* on-disk version is too old */ VDEV_AUX_UNSUP_FEAT, /* unsupported features */ VDEV_AUX_SPARED, /* hot spare used in another pool */ VDEV_AUX_ERR_EXCEEDED, /* too many errors */ VDEV_AUX_IO_FAILURE, /* experienced I/O failure */ VDEV_AUX_BAD_LOG, /* cannot read log chain(s) */ VDEV_AUX_EXTERNAL, /* external diagnosis or forced fault */ VDEV_AUX_SPLIT_POOL, /* vdev was split off into another pool */ VDEV_AUX_BAD_ASHIFT, /* vdev ashift is invalid */ VDEV_AUX_EXTERNAL_PERSIST, /* persistent forced fault */ VDEV_AUX_ACTIVE, /* vdev active on a different host */ VDEV_AUX_CHILDREN_OFFLINE, /* all children are offline */ VDEV_AUX_ASHIFT_TOO_BIG, /* vdev's min block size is too large */ } vdev_aux_t; /* * pool state. The following states are written to disk as part of the normal * SPA lifecycle: ACTIVE, EXPORTED, DESTROYED, SPARE, L2CACHE. The remaining * states are software abstractions used at various levels to communicate * pool state. */ typedef enum pool_state { POOL_STATE_ACTIVE = 0, /* In active use */ POOL_STATE_EXPORTED, /* Explicitly exported */ POOL_STATE_DESTROYED, /* Explicitly destroyed */ POOL_STATE_SPARE, /* Reserved for hot spare use */ POOL_STATE_L2CACHE, /* Level 2 ARC device */ POOL_STATE_UNINITIALIZED, /* Internal spa_t state */ POOL_STATE_UNAVAIL, /* Internal libzfs state */ POOL_STATE_POTENTIALLY_ACTIVE /* Internal libzfs state */ } pool_state_t; /* * mmp state. The following states provide additional detail describing * why a pool couldn't be safely imported. */ typedef enum mmp_state { MMP_STATE_ACTIVE = 0, /* In active use */ MMP_STATE_INACTIVE, /* Inactive and safe to import */ MMP_STATE_NO_HOSTID /* System hostid is not set */ } mmp_state_t; /* * Scan Functions. */ typedef enum pool_scan_func { POOL_SCAN_NONE, POOL_SCAN_SCRUB, POOL_SCAN_RESILVER, POOL_SCAN_FUNCS } pool_scan_func_t; /* * Used to control scrub pause and resume. */ typedef enum pool_scrub_cmd { POOL_SCRUB_NORMAL = 0, POOL_SCRUB_PAUSE, POOL_SCRUB_FLAGS_END } pool_scrub_cmd_t; typedef enum { CS_NONE, CS_CHECKPOINT_EXISTS, CS_CHECKPOINT_DISCARDING, CS_NUM_STATES } checkpoint_state_t; typedef struct pool_checkpoint_stat { uint64_t pcs_state; /* checkpoint_state_t */ uint64_t pcs_start_time; /* time checkpoint/discard started */ uint64_t pcs_space; /* checkpointed space */ } pool_checkpoint_stat_t; /* * ZIO types. Needed to interpret vdev statistics below. */ typedef enum zio_type { ZIO_TYPE_NULL = 0, ZIO_TYPE_READ, ZIO_TYPE_WRITE, ZIO_TYPE_FREE, ZIO_TYPE_CLAIM, ZIO_TYPE_IOCTL, ZIO_TYPE_TRIM, ZIO_TYPES } zio_type_t; /* * Pool statistics. Note: all fields should be 64-bit because this * is passed between kernel and userland as an nvlist uint64 array. */ typedef struct pool_scan_stat { /* values stored on disk */ uint64_t pss_func; /* pool_scan_func_t */ uint64_t pss_state; /* dsl_scan_state_t */ uint64_t pss_start_time; /* scan start time */ uint64_t pss_end_time; /* scan end time */ uint64_t pss_to_examine; /* total bytes to scan */ uint64_t pss_examined; /* total bytes located by scanner */ uint64_t pss_to_process; /* total bytes to process */ uint64_t pss_processed; /* total processed bytes */ uint64_t pss_errors; /* scan errors */ /* values not stored on disk */ uint64_t pss_pass_exam; /* examined bytes per scan pass */ uint64_t pss_pass_start; /* start time of a scan pass */ uint64_t pss_pass_scrub_pause; /* pause time of a scrub pass */ /* cumulative time scrub spent paused, needed for rate calculation */ uint64_t pss_pass_scrub_spent_paused; uint64_t pss_pass_issued; /* issued bytes per scan pass */ uint64_t pss_issued; /* total bytes checked by scanner */ } pool_scan_stat_t; typedef struct pool_removal_stat { uint64_t prs_state; /* dsl_scan_state_t */ uint64_t prs_removing_vdev; uint64_t prs_start_time; uint64_t prs_end_time; uint64_t prs_to_copy; /* bytes that need to be copied */ uint64_t prs_copied; /* bytes copied so far */ /* * bytes of memory used for indirect mappings. * This includes all removed vdevs. */ uint64_t prs_mapping_memory; } pool_removal_stat_t; typedef enum dsl_scan_state { DSS_NONE, DSS_SCANNING, DSS_FINISHED, DSS_CANCELED, DSS_NUM_STATES } dsl_scan_state_t; typedef struct vdev_rebuild_stat { uint64_t vrs_state; /* vdev_rebuild_state_t */ uint64_t vrs_start_time; /* time_t */ uint64_t vrs_end_time; /* time_t */ uint64_t vrs_scan_time_ms; /* total run time (millisecs) */ uint64_t vrs_bytes_scanned; /* allocated bytes scanned */ uint64_t vrs_bytes_issued; /* read bytes issued */ uint64_t vrs_bytes_rebuilt; /* rebuilt bytes */ uint64_t vrs_bytes_est; /* total bytes to scan */ uint64_t vrs_errors; /* scanning errors */ uint64_t vrs_pass_time_ms; /* pass run time (millisecs) */ uint64_t vrs_pass_bytes_scanned; /* bytes scanned since start/resume */ uint64_t vrs_pass_bytes_issued; /* bytes rebuilt since start/resume */ } vdev_rebuild_stat_t; /* * Errata described by https://openzfs.github.io/openzfs-docs/msg/ZFS-8000-ER. * The ordering of this enum must be maintained to ensure the errata identifiers * map to the correct documentation. New errata may only be appended to the * list and must contain corresponding documentation at the above link. */ typedef enum zpool_errata { ZPOOL_ERRATA_NONE, ZPOOL_ERRATA_ZOL_2094_SCRUB, ZPOOL_ERRATA_ZOL_2094_ASYNC_DESTROY, ZPOOL_ERRATA_ZOL_6845_ENCRYPTION, ZPOOL_ERRATA_ZOL_8308_ENCRYPTION, } zpool_errata_t; /* * Vdev statistics. Note: all fields should be 64-bit because this * is passed between kernel and user land as an nvlist uint64 array. * * The vs_ops[] and vs_bytes[] arrays must always be an array size of 6 in * order to keep subsequent members at their known fixed offsets. When * adding a new field it must be added to the end the structure. */ #define VS_ZIO_TYPES 6 typedef struct vdev_stat { hrtime_t vs_timestamp; /* time since vdev load */ uint64_t vs_state; /* vdev state */ uint64_t vs_aux; /* see vdev_aux_t */ uint64_t vs_alloc; /* space allocated */ uint64_t vs_space; /* total capacity */ uint64_t vs_dspace; /* deflated capacity */ uint64_t vs_rsize; /* replaceable dev size */ uint64_t vs_esize; /* expandable dev size */ uint64_t vs_ops[VS_ZIO_TYPES]; /* operation count */ uint64_t vs_bytes[VS_ZIO_TYPES]; /* bytes read/written */ uint64_t vs_read_errors; /* read errors */ uint64_t vs_write_errors; /* write errors */ uint64_t vs_checksum_errors; /* checksum errors */ uint64_t vs_initialize_errors; /* initializing errors */ uint64_t vs_self_healed; /* self-healed bytes */ uint64_t vs_scan_removing; /* removing? */ uint64_t vs_scan_processed; /* scan processed bytes */ uint64_t vs_fragmentation; /* device fragmentation */ uint64_t vs_initialize_bytes_done; /* bytes initialized */ uint64_t vs_initialize_bytes_est; /* total bytes to initialize */ uint64_t vs_initialize_state; /* vdev_initializing_state_t */ uint64_t vs_initialize_action_time; /* time_t */ uint64_t vs_checkpoint_space; /* checkpoint-consumed space */ uint64_t vs_resilver_deferred; /* resilver deferred */ uint64_t vs_slow_ios; /* slow IOs */ uint64_t vs_trim_errors; /* trimming errors */ uint64_t vs_trim_notsup; /* supported by device */ uint64_t vs_trim_bytes_done; /* bytes trimmed */ uint64_t vs_trim_bytes_est; /* total bytes to trim */ uint64_t vs_trim_state; /* vdev_trim_state_t */ uint64_t vs_trim_action_time; /* time_t */ uint64_t vs_rebuild_processed; /* bytes rebuilt */ uint64_t vs_configured_ashift; /* TLV vdev_ashift */ uint64_t vs_logical_ashift; /* vdev_logical_ashift */ uint64_t vs_physical_ashift; /* vdev_physical_ashift */ } vdev_stat_t; /* BEGIN CSTYLED */ #define VDEV_STAT_VALID(field, uint64_t_field_count) \ ((uint64_t_field_count * sizeof (uint64_t)) >= \ (offsetof(vdev_stat_t, field) + sizeof (((vdev_stat_t *)NULL)->field))) /* END CSTYLED */ /* * Extended stats * * These are stats which aren't included in the original iostat output. For * convenience, they are grouped together in vdev_stat_ex, although each stat * is individually exported as an nvlist. */ typedef struct vdev_stat_ex { /* Number of ZIOs issued to disk and waiting to finish */ uint64_t vsx_active_queue[ZIO_PRIORITY_NUM_QUEUEABLE]; /* Number of ZIOs pending to be issued to disk */ uint64_t vsx_pend_queue[ZIO_PRIORITY_NUM_QUEUEABLE]; /* * Below are the histograms for various latencies. Buckets are in * units of nanoseconds. */ /* * 2^37 nanoseconds = 134s. Timeouts will probably start kicking in * before this. */ #define VDEV_L_HISTO_BUCKETS 37 /* Latency histo buckets */ #define VDEV_RQ_HISTO_BUCKETS 25 /* Request size histo buckets */ /* Amount of time in ZIO queue (ns) */ uint64_t vsx_queue_histo[ZIO_PRIORITY_NUM_QUEUEABLE] [VDEV_L_HISTO_BUCKETS]; /* Total ZIO latency (ns). Includes queuing and disk access time */ uint64_t vsx_total_histo[ZIO_TYPES][VDEV_L_HISTO_BUCKETS]; /* Amount of time to read/write the disk (ns) */ uint64_t vsx_disk_histo[ZIO_TYPES][VDEV_L_HISTO_BUCKETS]; /* "lookup the bucket for a value" histogram macros */ #define HISTO(val, buckets) (val != 0 ? MIN(highbit64(val) - 1, \ buckets - 1) : 0) #define L_HISTO(a) HISTO(a, VDEV_L_HISTO_BUCKETS) #define RQ_HISTO(a) HISTO(a, VDEV_RQ_HISTO_BUCKETS) /* Physical IO histogram */ uint64_t vsx_ind_histo[ZIO_PRIORITY_NUM_QUEUEABLE] [VDEV_RQ_HISTO_BUCKETS]; /* Delegated (aggregated) physical IO histogram */ uint64_t vsx_agg_histo[ZIO_PRIORITY_NUM_QUEUEABLE] [VDEV_RQ_HISTO_BUCKETS]; } vdev_stat_ex_t; /* * Initialize functions. */ typedef enum pool_initialize_func { POOL_INITIALIZE_START, POOL_INITIALIZE_CANCEL, POOL_INITIALIZE_SUSPEND, POOL_INITIALIZE_FUNCS } pool_initialize_func_t; /* * TRIM functions. */ typedef enum pool_trim_func { POOL_TRIM_START, POOL_TRIM_CANCEL, POOL_TRIM_SUSPEND, POOL_TRIM_FUNCS } pool_trim_func_t; /* * DDT statistics. Note: all fields should be 64-bit because this * is passed between kernel and userland as an nvlist uint64 array. */ typedef struct ddt_object { uint64_t ddo_count; /* number of elements in ddt */ uint64_t ddo_dspace; /* size of ddt on disk */ uint64_t ddo_mspace; /* size of ddt in-core */ } ddt_object_t; typedef struct ddt_stat { uint64_t dds_blocks; /* blocks */ uint64_t dds_lsize; /* logical size */ uint64_t dds_psize; /* physical size */ uint64_t dds_dsize; /* deflated allocated size */ uint64_t dds_ref_blocks; /* referenced blocks */ uint64_t dds_ref_lsize; /* referenced lsize * refcnt */ uint64_t dds_ref_psize; /* referenced psize * refcnt */ uint64_t dds_ref_dsize; /* referenced dsize * refcnt */ } ddt_stat_t; typedef struct ddt_histogram { ddt_stat_t ddh_stat[64]; /* power-of-two histogram buckets */ } ddt_histogram_t; #define ZVOL_DRIVER "zvol" #define ZFS_DRIVER "zfs" #define ZFS_DEV "/dev/zfs" #define ZFS_SUPER_MAGIC 0x2fc12fc1 /* general zvol path */ #define ZVOL_DIR "/dev/zvol/" #define ZVOL_MAJOR 230 #define ZVOL_MINOR_BITS 4 #define ZVOL_MINOR_MASK ((1U << ZVOL_MINOR_BITS) - 1) #define ZVOL_MINORS (1 << 4) #define ZVOL_DEV_NAME "zd" #define ZVOL_PROP_NAME "name" #define ZVOL_DEFAULT_BLOCKSIZE 8192 typedef enum { VDEV_INITIALIZE_NONE, VDEV_INITIALIZE_ACTIVE, VDEV_INITIALIZE_CANCELED, VDEV_INITIALIZE_SUSPENDED, VDEV_INITIALIZE_COMPLETE } vdev_initializing_state_t; typedef enum { VDEV_TRIM_NONE, VDEV_TRIM_ACTIVE, VDEV_TRIM_CANCELED, VDEV_TRIM_SUSPENDED, VDEV_TRIM_COMPLETE, } vdev_trim_state_t; typedef enum { VDEV_REBUILD_NONE, VDEV_REBUILD_ACTIVE, VDEV_REBUILD_CANCELED, VDEV_REBUILD_COMPLETE, } vdev_rebuild_state_t; /* * nvlist name constants. Facilitate restricting snapshot iteration range for * the "list next snapshot" ioctl */ #define SNAP_ITER_MIN_TXG "snap_iter_min_txg" #define SNAP_ITER_MAX_TXG "snap_iter_max_txg" /* * /dev/zfs ioctl numbers. * * These numbers cannot change over time. New ioctl numbers must be appended. */ typedef enum zfs_ioc { /* * Core features - 81/128 numbers reserved. */ #ifdef __FreeBSD__ ZFS_IOC_FIRST = 0, #else ZFS_IOC_FIRST = ('Z' << 8), #endif ZFS_IOC = ZFS_IOC_FIRST, ZFS_IOC_POOL_CREATE = ZFS_IOC_FIRST, /* 0x5a00 */ ZFS_IOC_POOL_DESTROY, /* 0x5a01 */ ZFS_IOC_POOL_IMPORT, /* 0x5a02 */ ZFS_IOC_POOL_EXPORT, /* 0x5a03 */ ZFS_IOC_POOL_CONFIGS, /* 0x5a04 */ ZFS_IOC_POOL_STATS, /* 0x5a05 */ ZFS_IOC_POOL_TRYIMPORT, /* 0x5a06 */ ZFS_IOC_POOL_SCAN, /* 0x5a07 */ ZFS_IOC_POOL_FREEZE, /* 0x5a08 */ ZFS_IOC_POOL_UPGRADE, /* 0x5a09 */ ZFS_IOC_POOL_GET_HISTORY, /* 0x5a0a */ ZFS_IOC_VDEV_ADD, /* 0x5a0b */ ZFS_IOC_VDEV_REMOVE, /* 0x5a0c */ ZFS_IOC_VDEV_SET_STATE, /* 0x5a0d */ ZFS_IOC_VDEV_ATTACH, /* 0x5a0e */ ZFS_IOC_VDEV_DETACH, /* 0x5a0f */ ZFS_IOC_VDEV_SETPATH, /* 0x5a10 */ ZFS_IOC_VDEV_SETFRU, /* 0x5a11 */ ZFS_IOC_OBJSET_STATS, /* 0x5a12 */ ZFS_IOC_OBJSET_ZPLPROPS, /* 0x5a13 */ ZFS_IOC_DATASET_LIST_NEXT, /* 0x5a14 */ ZFS_IOC_SNAPSHOT_LIST_NEXT, /* 0x5a15 */ ZFS_IOC_SET_PROP, /* 0x5a16 */ ZFS_IOC_CREATE, /* 0x5a17 */ ZFS_IOC_DESTROY, /* 0x5a18 */ ZFS_IOC_ROLLBACK, /* 0x5a19 */ ZFS_IOC_RENAME, /* 0x5a1a */ ZFS_IOC_RECV, /* 0x5a1b */ ZFS_IOC_SEND, /* 0x5a1c */ ZFS_IOC_INJECT_FAULT, /* 0x5a1d */ ZFS_IOC_CLEAR_FAULT, /* 0x5a1e */ ZFS_IOC_INJECT_LIST_NEXT, /* 0x5a1f */ ZFS_IOC_ERROR_LOG, /* 0x5a20 */ ZFS_IOC_CLEAR, /* 0x5a21 */ ZFS_IOC_PROMOTE, /* 0x5a22 */ ZFS_IOC_SNAPSHOT, /* 0x5a23 */ ZFS_IOC_DSOBJ_TO_DSNAME, /* 0x5a24 */ ZFS_IOC_OBJ_TO_PATH, /* 0x5a25 */ ZFS_IOC_POOL_SET_PROPS, /* 0x5a26 */ ZFS_IOC_POOL_GET_PROPS, /* 0x5a27 */ ZFS_IOC_SET_FSACL, /* 0x5a28 */ ZFS_IOC_GET_FSACL, /* 0x5a29 */ ZFS_IOC_SHARE, /* 0x5a2a */ ZFS_IOC_INHERIT_PROP, /* 0x5a2b */ ZFS_IOC_SMB_ACL, /* 0x5a2c */ ZFS_IOC_USERSPACE_ONE, /* 0x5a2d */ ZFS_IOC_USERSPACE_MANY, /* 0x5a2e */ ZFS_IOC_USERSPACE_UPGRADE, /* 0x5a2f */ ZFS_IOC_HOLD, /* 0x5a30 */ ZFS_IOC_RELEASE, /* 0x5a31 */ ZFS_IOC_GET_HOLDS, /* 0x5a32 */ ZFS_IOC_OBJSET_RECVD_PROPS, /* 0x5a33 */ ZFS_IOC_VDEV_SPLIT, /* 0x5a34 */ ZFS_IOC_NEXT_OBJ, /* 0x5a35 */ ZFS_IOC_DIFF, /* 0x5a36 */ ZFS_IOC_TMP_SNAPSHOT, /* 0x5a37 */ ZFS_IOC_OBJ_TO_STATS, /* 0x5a38 */ ZFS_IOC_SPACE_WRITTEN, /* 0x5a39 */ ZFS_IOC_SPACE_SNAPS, /* 0x5a3a */ ZFS_IOC_DESTROY_SNAPS, /* 0x5a3b */ ZFS_IOC_POOL_REGUID, /* 0x5a3c */ ZFS_IOC_POOL_REOPEN, /* 0x5a3d */ ZFS_IOC_SEND_PROGRESS, /* 0x5a3e */ ZFS_IOC_LOG_HISTORY, /* 0x5a3f */ ZFS_IOC_SEND_NEW, /* 0x5a40 */ ZFS_IOC_SEND_SPACE, /* 0x5a41 */ ZFS_IOC_CLONE, /* 0x5a42 */ ZFS_IOC_BOOKMARK, /* 0x5a43 */ ZFS_IOC_GET_BOOKMARKS, /* 0x5a44 */ ZFS_IOC_DESTROY_BOOKMARKS, /* 0x5a45 */ ZFS_IOC_RECV_NEW, /* 0x5a46 */ ZFS_IOC_POOL_SYNC, /* 0x5a47 */ ZFS_IOC_CHANNEL_PROGRAM, /* 0x5a48 */ ZFS_IOC_LOAD_KEY, /* 0x5a49 */ ZFS_IOC_UNLOAD_KEY, /* 0x5a4a */ ZFS_IOC_CHANGE_KEY, /* 0x5a4b */ ZFS_IOC_REMAP, /* 0x5a4c */ ZFS_IOC_POOL_CHECKPOINT, /* 0x5a4d */ ZFS_IOC_POOL_DISCARD_CHECKPOINT, /* 0x5a4e */ ZFS_IOC_POOL_INITIALIZE, /* 0x5a4f */ ZFS_IOC_POOL_TRIM, /* 0x5a50 */ ZFS_IOC_REDACT, /* 0x5a51 */ ZFS_IOC_GET_BOOKMARK_PROPS, /* 0x5a52 */ ZFS_IOC_WAIT, /* 0x5a53 */ ZFS_IOC_WAIT_FS, /* 0x5a54 */ /* * Per-platform (Optional) - 8/128 numbers reserved. */ ZFS_IOC_PLATFORM = ZFS_IOC_FIRST + 0x80, ZFS_IOC_EVENTS_NEXT, /* 0x81 (Linux) */ ZFS_IOC_EVENTS_CLEAR, /* 0x82 (Linux) */ ZFS_IOC_EVENTS_SEEK, /* 0x83 (Linux) */ ZFS_IOC_NEXTBOOT, /* 0x84 (FreeBSD) */ ZFS_IOC_JAIL, /* 0x85 (FreeBSD) */ ZFS_IOC_UNJAIL, /* 0x86 (FreeBSD) */ ZFS_IOC_SET_BOOTENV, /* 0x87 */ ZFS_IOC_GET_BOOTENV, /* 0x88 */ ZFS_IOC_LAST } zfs_ioc_t; /* * zvol ioctl to get dataset name */ #define BLKZNAME _IOR(0x12, 125, char[ZFS_MAX_DATASET_NAME_LEN]) /* * ZFS-specific error codes used for returning descriptive errors * to the userland through zfs ioctls. * * The enum implicitly includes all the error codes from errno.h. * New code should use and extend this enum for errors that are * not described precisely by generic errno codes. * * These numbers should not change over time. New entries should be appended. * * (Keep in sync with contrib/pyzfs/libzfs_core/_constants.py) */ typedef enum { ZFS_ERR_CHECKPOINT_EXISTS = 1024, ZFS_ERR_DISCARDING_CHECKPOINT, ZFS_ERR_NO_CHECKPOINT, ZFS_ERR_DEVRM_IN_PROGRESS, ZFS_ERR_VDEV_TOO_BIG, ZFS_ERR_IOC_CMD_UNAVAIL, ZFS_ERR_IOC_ARG_UNAVAIL, ZFS_ERR_IOC_ARG_REQUIRED, ZFS_ERR_IOC_ARG_BADTYPE, ZFS_ERR_WRONG_PARENT, ZFS_ERR_FROM_IVSET_GUID_MISSING, ZFS_ERR_FROM_IVSET_GUID_MISMATCH, ZFS_ERR_SPILL_BLOCK_FLAG_MISSING, ZFS_ERR_UNKNOWN_SEND_STREAM_FEATURE, ZFS_ERR_EXPORT_IN_PROGRESS, ZFS_ERR_BOOKMARK_SOURCE_NOT_ANCESTOR, ZFS_ERR_STREAM_TRUNCATED, ZFS_ERR_STREAM_LARGE_BLOCK_MISMATCH, ZFS_ERR_RESILVER_IN_PROGRESS, ZFS_ERR_REBUILD_IN_PROGRESS, ZFS_ERR_BADPROP, } zfs_errno_t; /* * Internal SPA load state. Used by FMA diagnosis engine. */ typedef enum { SPA_LOAD_NONE, /* no load in progress */ SPA_LOAD_OPEN, /* normal open */ SPA_LOAD_IMPORT, /* import in progress */ SPA_LOAD_TRYIMPORT, /* tryimport in progress */ SPA_LOAD_RECOVER, /* recovery requested */ SPA_LOAD_ERROR, /* load failed */ SPA_LOAD_CREATE /* creation in progress */ } spa_load_state_t; typedef enum { ZPOOL_WAIT_CKPT_DISCARD, ZPOOL_WAIT_FREE, ZPOOL_WAIT_INITIALIZE, ZPOOL_WAIT_REPLACE, ZPOOL_WAIT_REMOVE, ZPOOL_WAIT_RESILVER, ZPOOL_WAIT_SCRUB, ZPOOL_WAIT_TRIM, ZPOOL_WAIT_NUM_ACTIVITIES } zpool_wait_activity_t; typedef enum { ZFS_WAIT_DELETEQ, ZFS_WAIT_NUM_ACTIVITIES } zfs_wait_activity_t; /* * Bookmark name values. */ #define ZPOOL_ERR_LIST "error list" #define ZPOOL_ERR_DATASET "dataset" #define ZPOOL_ERR_OBJECT "object" #define HIS_MAX_RECORD_LEN (MAXPATHLEN + MAXPATHLEN + 1) /* * The following are names used in the nvlist describing * the pool's history log. */ #define ZPOOL_HIST_RECORD "history record" #define ZPOOL_HIST_TIME "history time" #define ZPOOL_HIST_CMD "history command" #define ZPOOL_HIST_WHO "history who" #define ZPOOL_HIST_ZONE "history zone" #define ZPOOL_HIST_HOST "history hostname" #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" #define ZPOOL_HIST_ERRNO "errno" /* * Special nvlist name that will not have its args recorded in the pool's * history log. */ #define ZPOOL_HIDDEN_ARGS "hidden_args" /* * The following are names used when invoking ZFS_IOC_POOL_INITIALIZE. */ #define ZPOOL_INITIALIZE_COMMAND "initialize_command" #define ZPOOL_INITIALIZE_VDEVS "initialize_vdevs" /* * The following are names used when invoking ZFS_IOC_POOL_TRIM. */ #define ZPOOL_TRIM_COMMAND "trim_command" #define ZPOOL_TRIM_VDEVS "trim_vdevs" #define ZPOOL_TRIM_RATE "trim_rate" #define ZPOOL_TRIM_SECURE "trim_secure" /* * The following are names used when invoking ZFS_IOC_POOL_WAIT. */ #define ZPOOL_WAIT_ACTIVITY "wait_activity" #define ZPOOL_WAIT_TAG "wait_tag" #define ZPOOL_WAIT_WAITED "wait_waited" /* * The following are names used when invoking ZFS_IOC_WAIT_FS. */ #define ZFS_WAIT_ACTIVITY "wait_activity" #define ZFS_WAIT_WAITED "wait_waited" /* * Flags for ZFS_IOC_VDEV_SET_STATE */ #define ZFS_ONLINE_CHECKREMOVE 0x1 #define ZFS_ONLINE_UNSPARE 0x2 #define ZFS_ONLINE_FORCEFAULT 0x4 #define ZFS_ONLINE_EXPAND 0x8 #define ZFS_OFFLINE_TEMPORARY 0x1 /* * Flags for ZFS_IOC_POOL_IMPORT */ #define ZFS_IMPORT_NORMAL 0x0 #define ZFS_IMPORT_VERBATIM 0x1 #define ZFS_IMPORT_ANY_HOST 0x2 #define ZFS_IMPORT_MISSING_LOG 0x4 #define ZFS_IMPORT_ONLY 0x8 #define ZFS_IMPORT_TEMP_NAME 0x10 #define ZFS_IMPORT_SKIP_MMP 0x20 #define ZFS_IMPORT_LOAD_KEYS 0x40 #define ZFS_IMPORT_CHECKPOINT 0x80 /* * Channel program argument/return nvlist keys and defaults. */ #define ZCP_ARG_PROGRAM "program" #define ZCP_ARG_ARGLIST "arg" #define ZCP_ARG_SYNC "sync" #define ZCP_ARG_INSTRLIMIT "instrlimit" #define ZCP_ARG_MEMLIMIT "memlimit" #define ZCP_ARG_CLIARGV "argv" #define ZCP_RET_ERROR "error" #define ZCP_RET_RETURN "return" #define ZCP_DEFAULT_INSTRLIMIT (10 * 1000 * 1000) #define ZCP_MAX_INSTRLIMIT (10 * ZCP_DEFAULT_INSTRLIMIT) #define ZCP_DEFAULT_MEMLIMIT (10 * 1024 * 1024) #define ZCP_MAX_MEMLIMIT (10 * ZCP_DEFAULT_MEMLIMIT) /* * Sysevent payload members. ZFS will generate the following sysevents with the * given payloads: * * ESC_ZFS_RESILVER_START * ESC_ZFS_RESILVER_FINISH * * ZFS_EV_POOL_NAME DATA_TYPE_STRING * ZFS_EV_POOL_GUID DATA_TYPE_UINT64 * ZFS_EV_RESILVER_TYPE DATA_TYPE_STRING * * ESC_ZFS_POOL_DESTROY * ESC_ZFS_POOL_REGUID * * ZFS_EV_POOL_NAME DATA_TYPE_STRING * ZFS_EV_POOL_GUID DATA_TYPE_UINT64 * * ESC_ZFS_VDEV_REMOVE * ESC_ZFS_VDEV_CLEAR * ESC_ZFS_VDEV_CHECK * * ZFS_EV_POOL_NAME DATA_TYPE_STRING * ZFS_EV_POOL_GUID DATA_TYPE_UINT64 * ZFS_EV_VDEV_PATH DATA_TYPE_STRING (optional) * ZFS_EV_VDEV_GUID DATA_TYPE_UINT64 * * ESC_ZFS_HISTORY_EVENT * * ZFS_EV_POOL_NAME DATA_TYPE_STRING * ZFS_EV_POOL_GUID DATA_TYPE_UINT64 * ZFS_EV_HIST_TIME DATA_TYPE_UINT64 (optional) * ZFS_EV_HIST_CMD DATA_TYPE_STRING (optional) * ZFS_EV_HIST_WHO DATA_TYPE_UINT64 (optional) * ZFS_EV_HIST_ZONE DATA_TYPE_STRING (optional) * ZFS_EV_HIST_HOST DATA_TYPE_STRING (optional) * ZFS_EV_HIST_TXG DATA_TYPE_UINT64 (optional) * ZFS_EV_HIST_INT_EVENT DATA_TYPE_UINT64 (optional) * ZFS_EV_HIST_INT_STR DATA_TYPE_STRING (optional) * ZFS_EV_HIST_INT_NAME DATA_TYPE_STRING (optional) * ZFS_EV_HIST_IOCTL DATA_TYPE_STRING (optional) * ZFS_EV_HIST_DSNAME DATA_TYPE_STRING (optional) * ZFS_EV_HIST_DSID DATA_TYPE_UINT64 (optional) * * The ZFS_EV_HIST_* members will correspond to the ZPOOL_HIST_* members in the * history log nvlist. The keynames will be free of any spaces or other * characters that could be potentially unexpected to consumers of the * sysevents. */ #define ZFS_EV_POOL_NAME "pool_name" #define ZFS_EV_POOL_GUID "pool_guid" #define ZFS_EV_VDEV_PATH "vdev_path" #define ZFS_EV_VDEV_GUID "vdev_guid" #define ZFS_EV_HIST_TIME "history_time" #define ZFS_EV_HIST_CMD "history_command" #define ZFS_EV_HIST_WHO "history_who" #define ZFS_EV_HIST_ZONE "history_zone" #define ZFS_EV_HIST_HOST "history_hostname" #define ZFS_EV_HIST_TXG "history_txg" #define ZFS_EV_HIST_INT_EVENT "history_internal_event" #define ZFS_EV_HIST_INT_STR "history_internal_str" #define ZFS_EV_HIST_INT_NAME "history_internal_name" #define ZFS_EV_HIST_IOCTL "history_ioctl" #define ZFS_EV_HIST_DSNAME "history_dsname" #define ZFS_EV_HIST_DSID "history_dsid" #define ZFS_EV_RESILVER_TYPE "resilver_type" #ifdef __cplusplus } #endif #endif /* _SYS_FS_ZFS_H */ diff --git a/lib/libzfs/libzfs_util.c b/lib/libzfs/libzfs_util.c index a457fbfd0639..95cb32957218 100644 --- a/lib/libzfs/libzfs_util.c +++ b/lib/libzfs/libzfs_util.c @@ -1,2104 +1,2104 @@ /* * 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 2020 Joyent, Inc. All rights reserved. * Copyright (c) 2011, 2020 by Delphix. All rights reserved. * Copyright 2016 Igor Kozhukhov * Copyright (c) 2017 Datto Inc. * Copyright (c) 2020 The FreeBSD Foundation * * Portions of this software were developed by Allan Jude * under sponsorship from the FreeBSD Foundation. */ /* * Internal utility routines for the ZFS library. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "libzfs_impl.h" #include "zfs_prop.h" #include "zfeature_common.h" #include #include /* * We only care about the scheme in order to match the scheme * with the handler. Each handler should validate the full URI * as necessary. */ #define URI_REGEX "^\\([A-Za-z][A-Za-z0-9+.\\-]*\\):" int libzfs_errno(libzfs_handle_t *hdl) { return (hdl->libzfs_error); } const char * libzfs_error_action(libzfs_handle_t *hdl) { return (hdl->libzfs_action); } const char * libzfs_error_description(libzfs_handle_t *hdl) { if (hdl->libzfs_desc[0] != '\0') return (hdl->libzfs_desc); switch (hdl->libzfs_error) { case EZFS_NOMEM: return (dgettext(TEXT_DOMAIN, "out of memory")); case EZFS_BADPROP: return (dgettext(TEXT_DOMAIN, "invalid property value")); case EZFS_PROPREADONLY: return (dgettext(TEXT_DOMAIN, "read-only property")); case EZFS_PROPTYPE: return (dgettext(TEXT_DOMAIN, "property doesn't apply to " "datasets of this type")); case EZFS_PROPNONINHERIT: return (dgettext(TEXT_DOMAIN, "property cannot be inherited")); case EZFS_PROPSPACE: return (dgettext(TEXT_DOMAIN, "invalid quota or reservation")); case EZFS_BADTYPE: return (dgettext(TEXT_DOMAIN, "operation not applicable to " "datasets of this type")); case EZFS_BUSY: return (dgettext(TEXT_DOMAIN, "pool or dataset is busy")); case EZFS_EXISTS: return (dgettext(TEXT_DOMAIN, "pool or dataset exists")); case EZFS_NOENT: return (dgettext(TEXT_DOMAIN, "no such pool or dataset")); case EZFS_BADSTREAM: return (dgettext(TEXT_DOMAIN, "invalid backup stream")); case EZFS_DSREADONLY: return (dgettext(TEXT_DOMAIN, "dataset is read-only")); case EZFS_VOLTOOBIG: return (dgettext(TEXT_DOMAIN, "volume size exceeds limit for " "this system")); case EZFS_INVALIDNAME: return (dgettext(TEXT_DOMAIN, "invalid name")); case EZFS_BADRESTORE: return (dgettext(TEXT_DOMAIN, "unable to restore to " "destination")); case EZFS_BADBACKUP: return (dgettext(TEXT_DOMAIN, "backup failed")); case EZFS_BADTARGET: return (dgettext(TEXT_DOMAIN, "invalid target vdev")); case EZFS_NODEVICE: return (dgettext(TEXT_DOMAIN, "no such device in pool")); case EZFS_BADDEV: return (dgettext(TEXT_DOMAIN, "invalid device")); case EZFS_NOREPLICAS: return (dgettext(TEXT_DOMAIN, "no valid replicas")); case EZFS_RESILVERING: return (dgettext(TEXT_DOMAIN, "currently resilvering")); case EZFS_BADVERSION: return (dgettext(TEXT_DOMAIN, "unsupported version or " "feature")); case EZFS_POOLUNAVAIL: return (dgettext(TEXT_DOMAIN, "pool is unavailable")); case EZFS_DEVOVERFLOW: return (dgettext(TEXT_DOMAIN, "too many devices in one vdev")); case EZFS_BADPATH: return (dgettext(TEXT_DOMAIN, "must be an absolute path")); case EZFS_CROSSTARGET: return (dgettext(TEXT_DOMAIN, "operation crosses datasets or " "pools")); case EZFS_ZONED: return (dgettext(TEXT_DOMAIN, "dataset in use by local zone")); case EZFS_MOUNTFAILED: return (dgettext(TEXT_DOMAIN, "mount failed")); case EZFS_UMOUNTFAILED: - return (dgettext(TEXT_DOMAIN, "umount failed")); + return (dgettext(TEXT_DOMAIN, "unmount failed")); case EZFS_UNSHARENFSFAILED: - return (dgettext(TEXT_DOMAIN, "unshare(1M) failed")); + return (dgettext(TEXT_DOMAIN, "NFS share removal failed")); case EZFS_SHARENFSFAILED: - return (dgettext(TEXT_DOMAIN, "share(1M) failed")); + return (dgettext(TEXT_DOMAIN, "NFS share creation failed")); case EZFS_UNSHARESMBFAILED: - return (dgettext(TEXT_DOMAIN, "smb remove share failed")); + return (dgettext(TEXT_DOMAIN, "SMB share removal failed")); case EZFS_SHARESMBFAILED: - return (dgettext(TEXT_DOMAIN, "smb add share failed")); + return (dgettext(TEXT_DOMAIN, "SMB share creation failed")); case EZFS_PERM: return (dgettext(TEXT_DOMAIN, "permission denied")); case EZFS_NOSPC: return (dgettext(TEXT_DOMAIN, "out of space")); case EZFS_FAULT: return (dgettext(TEXT_DOMAIN, "bad address")); case EZFS_IO: return (dgettext(TEXT_DOMAIN, "I/O error")); case EZFS_INTR: return (dgettext(TEXT_DOMAIN, "signal received")); case EZFS_ISSPARE: return (dgettext(TEXT_DOMAIN, "device is reserved as a hot " "spare")); case EZFS_INVALCONFIG: return (dgettext(TEXT_DOMAIN, "invalid vdev configuration")); case EZFS_RECURSIVE: return (dgettext(TEXT_DOMAIN, "recursive dataset dependency")); case EZFS_NOHISTORY: return (dgettext(TEXT_DOMAIN, "no history available")); case EZFS_POOLPROPS: return (dgettext(TEXT_DOMAIN, "failed to retrieve " "pool properties")); case EZFS_POOL_NOTSUP: return (dgettext(TEXT_DOMAIN, "operation not supported " "on this type of pool")); case EZFS_POOL_INVALARG: return (dgettext(TEXT_DOMAIN, "invalid argument for " "this pool operation")); case EZFS_NAMETOOLONG: return (dgettext(TEXT_DOMAIN, "dataset name is too long")); case EZFS_OPENFAILED: return (dgettext(TEXT_DOMAIN, "open failed")); case EZFS_NOCAP: return (dgettext(TEXT_DOMAIN, "disk capacity information could not be retrieved")); case EZFS_LABELFAILED: return (dgettext(TEXT_DOMAIN, "write of label failed")); case EZFS_BADWHO: return (dgettext(TEXT_DOMAIN, "invalid user/group")); case EZFS_BADPERM: return (dgettext(TEXT_DOMAIN, "invalid permission")); case EZFS_BADPERMSET: return (dgettext(TEXT_DOMAIN, "invalid permission set name")); case EZFS_NODELEGATION: return (dgettext(TEXT_DOMAIN, "delegated administration is " "disabled on pool")); case EZFS_BADCACHE: return (dgettext(TEXT_DOMAIN, "invalid or missing cache file")); case EZFS_ISL2CACHE: return (dgettext(TEXT_DOMAIN, "device is in use as a cache")); case EZFS_VDEVNOTSUP: return (dgettext(TEXT_DOMAIN, "vdev specification is not " "supported")); case EZFS_NOTSUP: return (dgettext(TEXT_DOMAIN, "operation not supported " "on this dataset")); case EZFS_IOC_NOTSUPPORTED: return (dgettext(TEXT_DOMAIN, "operation not supported by " "zfs kernel module")); case EZFS_ACTIVE_SPARE: return (dgettext(TEXT_DOMAIN, "pool has active shared spare " "device")); case EZFS_UNPLAYED_LOGS: return (dgettext(TEXT_DOMAIN, "log device has unplayed intent " "logs")); case EZFS_REFTAG_RELE: return (dgettext(TEXT_DOMAIN, "no such tag on this dataset")); case EZFS_REFTAG_HOLD: return (dgettext(TEXT_DOMAIN, "tag already exists on this " "dataset")); case EZFS_TAGTOOLONG: return (dgettext(TEXT_DOMAIN, "tag too long")); case EZFS_PIPEFAILED: return (dgettext(TEXT_DOMAIN, "pipe create failed")); case EZFS_THREADCREATEFAILED: return (dgettext(TEXT_DOMAIN, "thread create failed")); case EZFS_POSTSPLIT_ONLINE: return (dgettext(TEXT_DOMAIN, "disk was split from this pool " "into a new one")); case EZFS_SCRUB_PAUSED: return (dgettext(TEXT_DOMAIN, "scrub is paused; " "use 'zpool scrub' to resume")); case EZFS_SCRUBBING: return (dgettext(TEXT_DOMAIN, "currently scrubbing; " "use 'zpool scrub -s' to cancel current scrub")); case EZFS_NO_SCRUB: return (dgettext(TEXT_DOMAIN, "there is no active scrub")); case EZFS_DIFF: return (dgettext(TEXT_DOMAIN, "unable to generate diffs")); case EZFS_DIFFDATA: return (dgettext(TEXT_DOMAIN, "invalid diff data")); case EZFS_POOLREADONLY: return (dgettext(TEXT_DOMAIN, "pool is read-only")); case EZFS_NO_PENDING: return (dgettext(TEXT_DOMAIN, "operation is not " "in progress")); case EZFS_CHECKPOINT_EXISTS: return (dgettext(TEXT_DOMAIN, "checkpoint exists")); case EZFS_DISCARDING_CHECKPOINT: return (dgettext(TEXT_DOMAIN, "currently discarding " "checkpoint")); case EZFS_NO_CHECKPOINT: return (dgettext(TEXT_DOMAIN, "checkpoint does not exist")); case EZFS_DEVRM_IN_PROGRESS: return (dgettext(TEXT_DOMAIN, "device removal in progress")); case EZFS_VDEV_TOO_BIG: return (dgettext(TEXT_DOMAIN, "device exceeds supported size")); case EZFS_ACTIVE_POOL: return (dgettext(TEXT_DOMAIN, "pool is imported on a " "different host")); case EZFS_CRYPTOFAILED: return (dgettext(TEXT_DOMAIN, "encryption failure")); case EZFS_TOOMANY: return (dgettext(TEXT_DOMAIN, "argument list too long")); case EZFS_INITIALIZING: return (dgettext(TEXT_DOMAIN, "currently initializing")); case EZFS_NO_INITIALIZE: return (dgettext(TEXT_DOMAIN, "there is no active " "initialization")); case EZFS_WRONG_PARENT: return (dgettext(TEXT_DOMAIN, "invalid parent dataset")); case EZFS_TRIMMING: return (dgettext(TEXT_DOMAIN, "currently trimming")); case EZFS_NO_TRIM: return (dgettext(TEXT_DOMAIN, "there is no active trim")); case EZFS_TRIM_NOTSUP: return (dgettext(TEXT_DOMAIN, "trim operations are not " "supported by this device")); case EZFS_NO_RESILVER_DEFER: return (dgettext(TEXT_DOMAIN, "this action requires the " "resilver_defer feature")); case EZFS_EXPORT_IN_PROGRESS: return (dgettext(TEXT_DOMAIN, "pool export in progress")); case EZFS_REBUILDING: return (dgettext(TEXT_DOMAIN, "currently sequentially " "resilvering")); case EZFS_UNKNOWN: return (dgettext(TEXT_DOMAIN, "unknown error")); default: assert(hdl->libzfs_error == 0); return (dgettext(TEXT_DOMAIN, "no error")); } } /*PRINTFLIKE2*/ void zfs_error_aux(libzfs_handle_t *hdl, const char *fmt, ...) { va_list ap; va_start(ap, fmt); (void) vsnprintf(hdl->libzfs_desc, sizeof (hdl->libzfs_desc), fmt, ap); hdl->libzfs_desc_active = 1; va_end(ap); } static void zfs_verror(libzfs_handle_t *hdl, int error, const char *fmt, va_list ap) { (void) vsnprintf(hdl->libzfs_action, sizeof (hdl->libzfs_action), fmt, ap); hdl->libzfs_error = error; if (hdl->libzfs_desc_active) hdl->libzfs_desc_active = 0; else hdl->libzfs_desc[0] = '\0'; if (hdl->libzfs_printerr) { if (error == EZFS_UNKNOWN) { (void) fprintf(stderr, dgettext(TEXT_DOMAIN, "internal " "error: %s: %s\n"), hdl->libzfs_action, libzfs_error_description(hdl)); abort(); } (void) fprintf(stderr, "%s: %s\n", hdl->libzfs_action, libzfs_error_description(hdl)); if (error == EZFS_NOMEM) exit(1); } } int zfs_error(libzfs_handle_t *hdl, int error, const char *msg) { return (zfs_error_fmt(hdl, error, "%s", msg)); } /*PRINTFLIKE3*/ int zfs_error_fmt(libzfs_handle_t *hdl, int error, const char *fmt, ...) { va_list ap; va_start(ap, fmt); zfs_verror(hdl, error, fmt, ap); va_end(ap); return (-1); } static int zfs_common_error(libzfs_handle_t *hdl, int error, const char *fmt, va_list ap) { switch (error) { case EPERM: case EACCES: zfs_verror(hdl, EZFS_PERM, fmt, ap); return (-1); case ECANCELED: zfs_verror(hdl, EZFS_NODELEGATION, fmt, ap); return (-1); case EIO: zfs_verror(hdl, EZFS_IO, fmt, ap); return (-1); case EFAULT: zfs_verror(hdl, EZFS_FAULT, fmt, ap); return (-1); case EINTR: zfs_verror(hdl, EZFS_INTR, fmt, ap); return (-1); } return (0); } int zfs_standard_error(libzfs_handle_t *hdl, int error, const char *msg) { return (zfs_standard_error_fmt(hdl, error, "%s", msg)); } /*PRINTFLIKE3*/ int zfs_standard_error_fmt(libzfs_handle_t *hdl, int error, const char *fmt, ...) { va_list ap; va_start(ap, fmt); if (zfs_common_error(hdl, error, fmt, ap) != 0) { va_end(ap); return (-1); } switch (error) { case ENXIO: case ENODEV: case EPIPE: zfs_verror(hdl, EZFS_IO, fmt, ap); break; case ENOENT: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "dataset does not exist")); zfs_verror(hdl, EZFS_NOENT, fmt, ap); break; case ENOSPC: case EDQUOT: zfs_verror(hdl, EZFS_NOSPC, fmt, ap); break; case EEXIST: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "dataset already exists")); zfs_verror(hdl, EZFS_EXISTS, fmt, ap); break; case EBUSY: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "dataset is busy")); zfs_verror(hdl, EZFS_BUSY, fmt, ap); break; case EROFS: zfs_verror(hdl, EZFS_POOLREADONLY, fmt, ap); break; case ENAMETOOLONG: zfs_verror(hdl, EZFS_NAMETOOLONG, fmt, ap); break; case ENOTSUP: zfs_verror(hdl, EZFS_BADVERSION, fmt, ap); break; case EAGAIN: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool I/O is currently suspended")); zfs_verror(hdl, EZFS_POOLUNAVAIL, fmt, ap); break; case EREMOTEIO: zfs_verror(hdl, EZFS_ACTIVE_POOL, fmt, ap); break; case ZFS_ERR_UNKNOWN_SEND_STREAM_FEATURE: case ZFS_ERR_IOC_CMD_UNAVAIL: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "the loaded zfs " "module does not support this operation. A reboot may " "be required to enable this operation.")); zfs_verror(hdl, EZFS_IOC_NOTSUPPORTED, fmt, ap); break; case ZFS_ERR_IOC_ARG_UNAVAIL: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "the loaded zfs " "module does not support an option for this operation. " "A reboot may be required to enable this option.")); zfs_verror(hdl, EZFS_IOC_NOTSUPPORTED, fmt, ap); break; case ZFS_ERR_IOC_ARG_REQUIRED: case ZFS_ERR_IOC_ARG_BADTYPE: zfs_verror(hdl, EZFS_IOC_NOTSUPPORTED, fmt, ap); break; case ZFS_ERR_WRONG_PARENT: zfs_verror(hdl, EZFS_WRONG_PARENT, fmt, ap); break; case ZFS_ERR_BADPROP: zfs_verror(hdl, EZFS_BADPROP, fmt, ap); break; default: zfs_error_aux(hdl, strerror(error)); zfs_verror(hdl, EZFS_UNKNOWN, fmt, ap); break; } va_end(ap); return (-1); } void zfs_setprop_error(libzfs_handle_t *hdl, zfs_prop_t prop, int err, char *errbuf) { switch (err) { case ENOSPC: /* * For quotas and reservations, ENOSPC indicates * something different; setting a quota or reservation * doesn't use any disk space. */ switch (prop) { case ZFS_PROP_QUOTA: case ZFS_PROP_REFQUOTA: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "size is less than current used or " "reserved space")); (void) zfs_error(hdl, EZFS_PROPSPACE, errbuf); break; case ZFS_PROP_RESERVATION: case ZFS_PROP_REFRESERVATION: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "size is greater than available space")); (void) zfs_error(hdl, EZFS_PROPSPACE, errbuf); break; default: (void) zfs_standard_error(hdl, err, errbuf); break; } break; case EBUSY: (void) zfs_standard_error(hdl, EBUSY, errbuf); break; case EROFS: (void) zfs_error(hdl, EZFS_DSREADONLY, errbuf); break; case E2BIG: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property value too long")); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); break; case ENOTSUP: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool and or dataset must be upgraded to set this " "property or value")); (void) zfs_error(hdl, EZFS_BADVERSION, errbuf); break; case ERANGE: if (prop == ZFS_PROP_COMPRESSION || prop == ZFS_PROP_DNODESIZE || prop == ZFS_PROP_RECORDSIZE) { (void) zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property setting is not allowed on " "bootable datasets")); (void) zfs_error(hdl, EZFS_NOTSUP, errbuf); } else if (prop == ZFS_PROP_CHECKSUM || prop == ZFS_PROP_DEDUP) { (void) zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property setting is not allowed on " "root pools")); (void) zfs_error(hdl, EZFS_NOTSUP, errbuf); } else { (void) zfs_standard_error(hdl, err, errbuf); } break; case EINVAL: if (prop == ZPROP_INVAL) { (void) zfs_error(hdl, EZFS_BADPROP, errbuf); } else { (void) zfs_standard_error(hdl, err, errbuf); } break; case ZFS_ERR_BADPROP: (void) zfs_error(hdl, EZFS_BADPROP, errbuf); break; case EACCES: if (prop == ZFS_PROP_KEYLOCATION) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "keylocation may only be set on encryption roots")); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); } else { (void) zfs_standard_error(hdl, err, errbuf); } break; case EOVERFLOW: /* * This platform can't address a volume this big. */ #ifdef _ILP32 if (prop == ZFS_PROP_VOLSIZE) { (void) zfs_error(hdl, EZFS_VOLTOOBIG, errbuf); break; } #endif /* FALLTHROUGH */ default: (void) zfs_standard_error(hdl, err, errbuf); } } int zpool_standard_error(libzfs_handle_t *hdl, int error, const char *msg) { return (zpool_standard_error_fmt(hdl, error, "%s", msg)); } /*PRINTFLIKE3*/ int zpool_standard_error_fmt(libzfs_handle_t *hdl, int error, const char *fmt, ...) { va_list ap; va_start(ap, fmt); if (zfs_common_error(hdl, error, fmt, ap) != 0) { va_end(ap); return (-1); } switch (error) { case ENODEV: zfs_verror(hdl, EZFS_NODEVICE, fmt, ap); break; case ENOENT: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "no such pool or dataset")); zfs_verror(hdl, EZFS_NOENT, fmt, ap); break; case EEXIST: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool already exists")); zfs_verror(hdl, EZFS_EXISTS, fmt, ap); break; case EBUSY: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool is busy")); zfs_verror(hdl, EZFS_BUSY, fmt, ap); break; /* There is no pending operation to cancel */ case ENOTACTIVE: zfs_verror(hdl, EZFS_NO_PENDING, fmt, ap); break; case ENXIO: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more devices is currently unavailable")); zfs_verror(hdl, EZFS_BADDEV, fmt, ap); break; case ENAMETOOLONG: zfs_verror(hdl, EZFS_DEVOVERFLOW, fmt, ap); break; case ENOTSUP: zfs_verror(hdl, EZFS_POOL_NOTSUP, fmt, ap); break; case EINVAL: zfs_verror(hdl, EZFS_POOL_INVALARG, fmt, ap); break; case ENOSPC: case EDQUOT: zfs_verror(hdl, EZFS_NOSPC, fmt, ap); return (-1); case EAGAIN: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool I/O is currently suspended")); zfs_verror(hdl, EZFS_POOLUNAVAIL, fmt, ap); break; case EROFS: zfs_verror(hdl, EZFS_POOLREADONLY, fmt, ap); break; case EDOM: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "block size out of range or does not match")); zfs_verror(hdl, EZFS_BADPROP, fmt, ap); break; case EREMOTEIO: zfs_verror(hdl, EZFS_ACTIVE_POOL, fmt, ap); break; case ZFS_ERR_CHECKPOINT_EXISTS: zfs_verror(hdl, EZFS_CHECKPOINT_EXISTS, fmt, ap); break; case ZFS_ERR_DISCARDING_CHECKPOINT: zfs_verror(hdl, EZFS_DISCARDING_CHECKPOINT, fmt, ap); break; case ZFS_ERR_NO_CHECKPOINT: zfs_verror(hdl, EZFS_NO_CHECKPOINT, fmt, ap); break; case ZFS_ERR_DEVRM_IN_PROGRESS: zfs_verror(hdl, EZFS_DEVRM_IN_PROGRESS, fmt, ap); break; case ZFS_ERR_VDEV_TOO_BIG: zfs_verror(hdl, EZFS_VDEV_TOO_BIG, fmt, ap); break; case ZFS_ERR_EXPORT_IN_PROGRESS: zfs_verror(hdl, EZFS_EXPORT_IN_PROGRESS, fmt, ap); break; case ZFS_ERR_RESILVER_IN_PROGRESS: zfs_verror(hdl, EZFS_RESILVERING, fmt, ap); break; case ZFS_ERR_REBUILD_IN_PROGRESS: zfs_verror(hdl, EZFS_REBUILDING, fmt, ap); break; case ZFS_ERR_BADPROP: zfs_verror(hdl, EZFS_BADPROP, fmt, ap); break; case ZFS_ERR_IOC_CMD_UNAVAIL: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "the loaded zfs " "module does not support this operation. A reboot may " "be required to enable this operation.")); zfs_verror(hdl, EZFS_IOC_NOTSUPPORTED, fmt, ap); break; case ZFS_ERR_IOC_ARG_UNAVAIL: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "the loaded zfs " "module does not support an option for this operation. " "A reboot may be required to enable this option.")); zfs_verror(hdl, EZFS_IOC_NOTSUPPORTED, fmt, ap); break; case ZFS_ERR_IOC_ARG_REQUIRED: case ZFS_ERR_IOC_ARG_BADTYPE: zfs_verror(hdl, EZFS_IOC_NOTSUPPORTED, fmt, ap); break; default: zfs_error_aux(hdl, strerror(error)); zfs_verror(hdl, EZFS_UNKNOWN, fmt, ap); } va_end(ap); return (-1); } /* * Display an out of memory error message and abort the current program. */ int no_memory(libzfs_handle_t *hdl) { return (zfs_error(hdl, EZFS_NOMEM, "internal error")); } /* * A safe form of malloc() which will die if the allocation fails. */ void * zfs_alloc(libzfs_handle_t *hdl, size_t size) { void *data; if ((data = calloc(1, size)) == NULL) (void) no_memory(hdl); return (data); } /* * A safe form of asprintf() which will die if the allocation fails. */ /*PRINTFLIKE2*/ char * zfs_asprintf(libzfs_handle_t *hdl, const char *fmt, ...) { va_list ap; char *ret; int err; va_start(ap, fmt); err = vasprintf(&ret, fmt, ap); va_end(ap); if (err < 0) (void) no_memory(hdl); return (ret); } /* * A safe form of realloc(), which also zeroes newly allocated space. */ void * zfs_realloc(libzfs_handle_t *hdl, void *ptr, size_t oldsize, size_t newsize) { void *ret; if ((ret = realloc(ptr, newsize)) == NULL) { (void) no_memory(hdl); return (NULL); } bzero((char *)ret + oldsize, (newsize - oldsize)); return (ret); } /* * A safe form of strdup() which will die if the allocation fails. */ char * zfs_strdup(libzfs_handle_t *hdl, const char *str) { char *ret; if ((ret = strdup(str)) == NULL) (void) no_memory(hdl); return (ret); } void libzfs_print_on_error(libzfs_handle_t *hdl, boolean_t printerr) { hdl->libzfs_printerr = printerr; } /* * Read lines from an open file descriptor and store them in an array of * strings until EOF. lines[] will be allocated and populated with all the * lines read. All newlines are replaced with NULL terminators for * convenience. lines[] must be freed after use with libzfs_free_str_array(). * * Returns the number of lines read. */ static int libzfs_read_stdout_from_fd(int fd, char **lines[]) { FILE *fp; int lines_cnt = 0; size_t len = 0; char *line = NULL; char **tmp_lines = NULL, **tmp; char *nl = NULL; int rc; fp = fdopen(fd, "r"); if (fp == NULL) return (0); while (1) { rc = getline(&line, &len, fp); if (rc == -1) break; tmp = realloc(tmp_lines, sizeof (*tmp_lines) * (lines_cnt + 1)); if (tmp == NULL) { /* Return the lines we were able to process */ break; } tmp_lines = tmp; /* Terminate newlines */ if ((nl = strchr(line, '\n')) != NULL) *nl = '\0'; tmp_lines[lines_cnt] = line; lines_cnt++; line = NULL; } fclose(fp); *lines = tmp_lines; return (lines_cnt); } static int libzfs_run_process_impl(const char *path, char *argv[], char *env[], int flags, char **lines[], int *lines_cnt) { pid_t pid; int error, devnull_fd; int link[2]; /* * Setup a pipe between our child and parent process if we're * reading stdout. */ if ((lines != NULL) && pipe(link) == -1) return (-EPIPE); pid = vfork(); if (pid == 0) { /* Child process */ devnull_fd = open("/dev/null", O_WRONLY); if (devnull_fd < 0) _exit(-1); if (!(flags & STDOUT_VERBOSE) && (lines == NULL)) (void) dup2(devnull_fd, STDOUT_FILENO); else if (lines != NULL) { /* Save the output to lines[] */ dup2(link[1], STDOUT_FILENO); close(link[0]); close(link[1]); } if (!(flags & STDERR_VERBOSE)) (void) dup2(devnull_fd, STDERR_FILENO); close(devnull_fd); if (flags & NO_DEFAULT_PATH) { if (env == NULL) execv(path, argv); else execve(path, argv, env); } else { if (env == NULL) execvp(path, argv); else execvpe(path, argv, env); } _exit(-1); } else if (pid > 0) { /* Parent process */ int status; while ((error = waitpid(pid, &status, 0)) == -1 && errno == EINTR) { } if (error < 0 || !WIFEXITED(status)) return (-1); if (lines != NULL) { close(link[1]); *lines_cnt = libzfs_read_stdout_from_fd(link[0], lines); } return (WEXITSTATUS(status)); } return (-1); } int libzfs_run_process(const char *path, char *argv[], int flags) { return (libzfs_run_process_impl(path, argv, NULL, flags, NULL, NULL)); } /* * Run a command and store its stdout lines in an array of strings (lines[]). * lines[] is allocated and populated for you, and the number of lines is set in * lines_cnt. lines[] must be freed after use with libzfs_free_str_array(). * All newlines (\n) in lines[] are terminated for convenience. */ int libzfs_run_process_get_stdout(const char *path, char *argv[], char *env[], char **lines[], int *lines_cnt) { return (libzfs_run_process_impl(path, argv, env, 0, lines, lines_cnt)); } /* * Same as libzfs_run_process_get_stdout(), but run without $PATH set. This * means that *path needs to be the full path to the executable. */ int libzfs_run_process_get_stdout_nopath(const char *path, char *argv[], char *env[], char **lines[], int *lines_cnt) { return (libzfs_run_process_impl(path, argv, env, NO_DEFAULT_PATH, lines, lines_cnt)); } /* * Free an array of strings. Free both the strings contained in the array and * the array itself. */ void libzfs_free_str_array(char **strs, int count) { while (--count >= 0) free(strs[count]); free(strs); } /* * Returns 1 if environment variable is set to "YES", "yes", "ON", "on", or * a non-zero number. * * Returns 0 otherwise. */ int libzfs_envvar_is_set(char *envvar) { char *env = getenv(envvar); if (env && (strtoul(env, NULL, 0) > 0 || (!strncasecmp(env, "YES", 3) && strnlen(env, 4) == 3) || (!strncasecmp(env, "ON", 2) && strnlen(env, 3) == 2))) return (1); return (0); } libzfs_handle_t * libzfs_init(void) { libzfs_handle_t *hdl; int error; char *env; if ((error = libzfs_load_module()) != 0) { errno = error; return (NULL); } if ((hdl = calloc(1, sizeof (libzfs_handle_t))) == NULL) { return (NULL); } if (regcomp(&hdl->libzfs_urire, URI_REGEX, 0) != 0) { free(hdl); return (NULL); } if ((hdl->libzfs_fd = open(ZFS_DEV, O_RDWR|O_EXCL)) < 0) { free(hdl); return (NULL); } #ifdef HAVE_SETMNTENT if ((hdl->libzfs_mnttab = setmntent(MNTTAB, "r")) == NULL) { #else if ((hdl->libzfs_mnttab = fopen(MNTTAB, "r")) == NULL) { #endif (void) close(hdl->libzfs_fd); free(hdl); return (NULL); } if (libzfs_core_init() != 0) { (void) close(hdl->libzfs_fd); (void) fclose(hdl->libzfs_mnttab); free(hdl); return (NULL); } zfs_prop_init(); zpool_prop_init(); zpool_feature_init(); libzfs_mnttab_init(hdl); fletcher_4_init(); if (getenv("ZFS_PROP_DEBUG") != NULL) { hdl->libzfs_prop_debug = B_TRUE; } if ((env = getenv("ZFS_SENDRECV_MAX_NVLIST")) != NULL) { if ((error = zfs_nicestrtonum(hdl, env, &hdl->libzfs_max_nvlist))) { errno = error; (void) close(hdl->libzfs_fd); (void) fclose(hdl->libzfs_mnttab); free(hdl); return (NULL); } } else { hdl->libzfs_max_nvlist = (SPA_MAXBLOCKSIZE * 4); } /* * For testing, remove some settable properties and features */ if (libzfs_envvar_is_set("ZFS_SYSFS_PROP_SUPPORT_TEST")) { zprop_desc_t *proptbl; proptbl = zpool_prop_get_table(); proptbl[ZPOOL_PROP_COMMENT].pd_zfs_mod_supported = B_FALSE; proptbl = zfs_prop_get_table(); proptbl[ZFS_PROP_DNODESIZE].pd_zfs_mod_supported = B_FALSE; zfeature_info_t *ftbl = spa_feature_table; ftbl[SPA_FEATURE_LARGE_BLOCKS].fi_zfs_mod_supported = B_FALSE; } return (hdl); } void libzfs_fini(libzfs_handle_t *hdl) { (void) close(hdl->libzfs_fd); if (hdl->libzfs_mnttab) #ifdef HAVE_SETMNTENT (void) endmntent(hdl->libzfs_mnttab); #else (void) fclose(hdl->libzfs_mnttab); #endif zpool_free_handles(hdl); namespace_clear(hdl); libzfs_mnttab_fini(hdl); libzfs_core_fini(); regfree(&hdl->libzfs_urire); fletcher_4_fini(); free(hdl); } libzfs_handle_t * zpool_get_handle(zpool_handle_t *zhp) { return (zhp->zpool_hdl); } libzfs_handle_t * zfs_get_handle(zfs_handle_t *zhp) { return (zhp->zfs_hdl); } zpool_handle_t * zfs_get_pool_handle(const zfs_handle_t *zhp) { return (zhp->zpool_hdl); } /* * Given a name, determine whether or not it's a valid path * (starts with '/' or "./"). If so, walk the mnttab trying * to match the device number. If not, treat the path as an * fs/vol/snap/bkmark name. */ zfs_handle_t * zfs_path_to_zhandle(libzfs_handle_t *hdl, const char *path, zfs_type_t argtype) { struct stat64 statbuf; struct extmnttab entry; if (path[0] != '/' && strncmp(path, "./", strlen("./")) != 0) { /* * It's not a valid path, assume it's a name of type 'argtype'. */ return (zfs_open(hdl, path, argtype)); } /* Reopen MNTTAB to prevent reading stale data from open file */ if (freopen(MNTTAB, "r", hdl->libzfs_mnttab) == NULL) return (NULL); if (getextmntent(path, &entry, &statbuf) != 0) return (NULL); if (strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0) { (void) fprintf(stderr, gettext("'%s': not a ZFS filesystem\n"), path); return (NULL); } return (zfs_open(hdl, entry.mnt_special, ZFS_TYPE_FILESYSTEM)); } /* * Initialize the zc_nvlist_dst member to prepare for receiving an nvlist from * an ioctl(). */ int zcmd_alloc_dst_nvlist(libzfs_handle_t *hdl, zfs_cmd_t *zc, size_t len) { if (len == 0) len = 256 * 1024; zc->zc_nvlist_dst_size = len; zc->zc_nvlist_dst = (uint64_t)(uintptr_t)zfs_alloc(hdl, zc->zc_nvlist_dst_size); if (zc->zc_nvlist_dst == 0) return (-1); return (0); } /* * Called when an ioctl() which returns an nvlist fails with ENOMEM. This will * expand the nvlist to the size specified in 'zc_nvlist_dst_size', which was * filled in by the kernel to indicate the actual required size. */ int zcmd_expand_dst_nvlist(libzfs_handle_t *hdl, zfs_cmd_t *zc) { free((void *)(uintptr_t)zc->zc_nvlist_dst); zc->zc_nvlist_dst = (uint64_t)(uintptr_t)zfs_alloc(hdl, zc->zc_nvlist_dst_size); if (zc->zc_nvlist_dst == 0) return (-1); return (0); } /* * Called to free the src and dst nvlists stored in the command structure. */ void zcmd_free_nvlists(zfs_cmd_t *zc) { free((void *)(uintptr_t)zc->zc_nvlist_conf); free((void *)(uintptr_t)zc->zc_nvlist_src); free((void *)(uintptr_t)zc->zc_nvlist_dst); zc->zc_nvlist_conf = 0; zc->zc_nvlist_src = 0; zc->zc_nvlist_dst = 0; } static int zcmd_write_nvlist_com(libzfs_handle_t *hdl, uint64_t *outnv, uint64_t *outlen, nvlist_t *nvl) { char *packed; size_t len; verify(nvlist_size(nvl, &len, NV_ENCODE_NATIVE) == 0); if ((packed = zfs_alloc(hdl, len)) == NULL) return (-1); verify(nvlist_pack(nvl, &packed, &len, NV_ENCODE_NATIVE, 0) == 0); *outnv = (uint64_t)(uintptr_t)packed; *outlen = len; return (0); } int zcmd_write_conf_nvlist(libzfs_handle_t *hdl, zfs_cmd_t *zc, nvlist_t *nvl) { return (zcmd_write_nvlist_com(hdl, &zc->zc_nvlist_conf, &zc->zc_nvlist_conf_size, nvl)); } int zcmd_write_src_nvlist(libzfs_handle_t *hdl, zfs_cmd_t *zc, nvlist_t *nvl) { return (zcmd_write_nvlist_com(hdl, &zc->zc_nvlist_src, &zc->zc_nvlist_src_size, nvl)); } /* * Unpacks an nvlist from the ZFS ioctl command structure. */ int zcmd_read_dst_nvlist(libzfs_handle_t *hdl, zfs_cmd_t *zc, nvlist_t **nvlp) { if (nvlist_unpack((void *)(uintptr_t)zc->zc_nvlist_dst, zc->zc_nvlist_dst_size, nvlp, 0) != 0) return (no_memory(hdl)); return (0); } /* * ================================================================ * API shared by zfs and zpool property management * ================================================================ */ static void zprop_print_headers(zprop_get_cbdata_t *cbp, zfs_type_t type) { zprop_list_t *pl = cbp->cb_proplist; int i; char *title; size_t len; cbp->cb_first = B_FALSE; if (cbp->cb_scripted) return; /* * Start with the length of the column headers. */ cbp->cb_colwidths[GET_COL_NAME] = strlen(dgettext(TEXT_DOMAIN, "NAME")); cbp->cb_colwidths[GET_COL_PROPERTY] = strlen(dgettext(TEXT_DOMAIN, "PROPERTY")); cbp->cb_colwidths[GET_COL_VALUE] = strlen(dgettext(TEXT_DOMAIN, "VALUE")); cbp->cb_colwidths[GET_COL_RECVD] = strlen(dgettext(TEXT_DOMAIN, "RECEIVED")); cbp->cb_colwidths[GET_COL_SOURCE] = strlen(dgettext(TEXT_DOMAIN, "SOURCE")); /* first property is always NAME */ assert(cbp->cb_proplist->pl_prop == ((type == ZFS_TYPE_POOL) ? ZPOOL_PROP_NAME : ZFS_PROP_NAME)); /* * Go through and calculate the widths for each column. For the * 'source' column, we kludge it up by taking the worst-case scenario of * inheriting from the longest name. This is acceptable because in the * majority of cases 'SOURCE' is the last column displayed, and we don't * use the width anyway. Note that the 'VALUE' column can be oversized, * if the name of the property is much longer than any values we find. */ for (pl = cbp->cb_proplist; pl != NULL; pl = pl->pl_next) { /* * 'PROPERTY' column */ if (pl->pl_prop != ZPROP_INVAL) { const char *propname = (type == ZFS_TYPE_POOL) ? zpool_prop_to_name(pl->pl_prop) : zfs_prop_to_name(pl->pl_prop); len = strlen(propname); if (len > cbp->cb_colwidths[GET_COL_PROPERTY]) cbp->cb_colwidths[GET_COL_PROPERTY] = len; } else { len = strlen(pl->pl_user_prop); if (len > cbp->cb_colwidths[GET_COL_PROPERTY]) cbp->cb_colwidths[GET_COL_PROPERTY] = len; } /* * 'VALUE' column. The first property is always the 'name' * property that was tacked on either by /sbin/zfs's * zfs_do_get() or when calling zprop_expand_list(), so we * ignore its width. If the user specified the name property * to display, then it will be later in the list in any case. */ if (pl != cbp->cb_proplist && pl->pl_width > cbp->cb_colwidths[GET_COL_VALUE]) cbp->cb_colwidths[GET_COL_VALUE] = pl->pl_width; /* 'RECEIVED' column. */ if (pl != cbp->cb_proplist && pl->pl_recvd_width > cbp->cb_colwidths[GET_COL_RECVD]) cbp->cb_colwidths[GET_COL_RECVD] = pl->pl_recvd_width; /* * 'NAME' and 'SOURCE' columns */ if (pl->pl_prop == (type == ZFS_TYPE_POOL ? ZPOOL_PROP_NAME : ZFS_PROP_NAME) && pl->pl_width > cbp->cb_colwidths[GET_COL_NAME]) { cbp->cb_colwidths[GET_COL_NAME] = pl->pl_width; cbp->cb_colwidths[GET_COL_SOURCE] = pl->pl_width + strlen(dgettext(TEXT_DOMAIN, "inherited from")); } } /* * Now go through and print the headers. */ for (i = 0; i < ZFS_GET_NCOLS; i++) { switch (cbp->cb_columns[i]) { case GET_COL_NAME: title = dgettext(TEXT_DOMAIN, "NAME"); break; case GET_COL_PROPERTY: title = dgettext(TEXT_DOMAIN, "PROPERTY"); break; case GET_COL_VALUE: title = dgettext(TEXT_DOMAIN, "VALUE"); break; case GET_COL_RECVD: title = dgettext(TEXT_DOMAIN, "RECEIVED"); break; case GET_COL_SOURCE: title = dgettext(TEXT_DOMAIN, "SOURCE"); break; default: title = NULL; } if (title != NULL) { if (i == (ZFS_GET_NCOLS - 1) || cbp->cb_columns[i + 1] == GET_COL_NONE) (void) printf("%s", title); else (void) printf("%-*s ", cbp->cb_colwidths[cbp->cb_columns[i]], title); } } (void) printf("\n"); } /* * Display a single line of output, according to the settings in the callback * structure. */ void zprop_print_one_property(const char *name, zprop_get_cbdata_t *cbp, const char *propname, const char *value, zprop_source_t sourcetype, const char *source, const char *recvd_value) { int i; const char *str = NULL; char buf[128]; /* * Ignore those source types that the user has chosen to ignore. */ if ((sourcetype & cbp->cb_sources) == 0) return; if (cbp->cb_first) zprop_print_headers(cbp, cbp->cb_type); for (i = 0; i < ZFS_GET_NCOLS; i++) { switch (cbp->cb_columns[i]) { case GET_COL_NAME: str = name; break; case GET_COL_PROPERTY: str = propname; break; case GET_COL_VALUE: str = value; break; case GET_COL_SOURCE: switch (sourcetype) { case ZPROP_SRC_NONE: str = "-"; break; case ZPROP_SRC_DEFAULT: str = "default"; break; case ZPROP_SRC_LOCAL: str = "local"; break; case ZPROP_SRC_TEMPORARY: str = "temporary"; break; case ZPROP_SRC_INHERITED: (void) snprintf(buf, sizeof (buf), "inherited from %s", source); str = buf; break; case ZPROP_SRC_RECEIVED: str = "received"; break; default: str = NULL; assert(!"unhandled zprop_source_t"); } break; case GET_COL_RECVD: str = (recvd_value == NULL ? "-" : recvd_value); break; default: continue; } if (i == (ZFS_GET_NCOLS - 1) || cbp->cb_columns[i + 1] == GET_COL_NONE) (void) printf("%s", str); else if (cbp->cb_scripted) (void) printf("%s\t", str); else (void) printf("%-*s ", cbp->cb_colwidths[cbp->cb_columns[i]], str); } (void) printf("\n"); } /* * Given a numeric suffix, convert the value into a number of bits that the * resulting value must be shifted. */ static int str2shift(libzfs_handle_t *hdl, const char *buf) { const char *ends = "BKMGTPEZ"; int i; if (buf[0] == '\0') return (0); for (i = 0; i < strlen(ends); i++) { if (toupper(buf[0]) == ends[i]) break; } if (i == strlen(ends)) { if (hdl) zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "invalid numeric suffix '%s'"), buf); return (-1); } /* * Allow 'G' = 'GB' = 'GiB', case-insensitively. * However, 'BB' and 'BiB' are disallowed. */ if (buf[1] == '\0' || (toupper(buf[0]) != 'B' && ((toupper(buf[1]) == 'B' && buf[2] == '\0') || (toupper(buf[1]) == 'I' && toupper(buf[2]) == 'B' && buf[3] == '\0')))) return (10 * i); if (hdl) zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "invalid numeric suffix '%s'"), buf); return (-1); } /* * Convert a string of the form '100G' into a real number. Used when setting * properties or creating a volume. 'buf' is used to place an extended error * message for the caller to use. */ int zfs_nicestrtonum(libzfs_handle_t *hdl, const char *value, uint64_t *num) { char *end; int shift; *num = 0; /* Check to see if this looks like a number. */ if ((value[0] < '0' || value[0] > '9') && value[0] != '.') { if (hdl) zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "bad numeric value '%s'"), value); return (-1); } /* Rely on strtoull() to process the numeric portion. */ errno = 0; *num = strtoull(value, &end, 10); /* * Check for ERANGE, which indicates that the value is too large to fit * in a 64-bit value. */ if (errno == ERANGE) { if (hdl) zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "numeric value is too large")); return (-1); } /* * If we have a decimal value, then do the computation with floating * point arithmetic. Otherwise, use standard arithmetic. */ if (*end == '.') { double fval = strtod(value, &end); if ((shift = str2shift(hdl, end)) == -1) return (-1); fval *= pow(2, shift); /* * UINT64_MAX is not exactly representable as a double. * The closest representation is UINT64_MAX + 1, so we * use a >= comparison instead of > for the bounds check. */ if (fval >= (double)UINT64_MAX) { if (hdl) zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "numeric value is too large")); return (-1); } *num = (uint64_t)fval; } else { if ((shift = str2shift(hdl, end)) == -1) return (-1); /* Check for overflow */ if (shift >= 64 || (*num << shift) >> shift != *num) { if (hdl) zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "numeric value is too large")); return (-1); } *num <<= shift; } return (0); } /* * Given a propname=value nvpair to set, parse any numeric properties * (index, boolean, etc) if they are specified as strings and add the * resulting nvpair to the returned nvlist. * * At the DSL layer, all properties are either 64-bit numbers or strings. * We want the user to be able to ignore this fact and specify properties * as native values (numbers, for example) or as strings (to simplify * command line utilities). This also handles converting index types * (compression, checksum, etc) from strings to their on-disk index. */ int zprop_parse_value(libzfs_handle_t *hdl, nvpair_t *elem, int prop, zfs_type_t type, nvlist_t *ret, char **svalp, uint64_t *ivalp, const char *errbuf) { data_type_t datatype = nvpair_type(elem); zprop_type_t proptype; const char *propname; char *value; boolean_t isnone = B_FALSE; boolean_t isauto = B_FALSE; int err = 0; if (type == ZFS_TYPE_POOL) { proptype = zpool_prop_get_type(prop); propname = zpool_prop_to_name(prop); } else { proptype = zfs_prop_get_type(prop); propname = zfs_prop_to_name(prop); } /* * Convert any properties to the internal DSL value types. */ *svalp = NULL; *ivalp = 0; switch (proptype) { case PROP_TYPE_STRING: if (datatype != DATA_TYPE_STRING) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' must be a string"), nvpair_name(elem)); goto error; } err = nvpair_value_string(elem, svalp); if (err != 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' is invalid"), nvpair_name(elem)); goto error; } if (strlen(*svalp) >= ZFS_MAXPROPLEN) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' is too long"), nvpair_name(elem)); goto error; } break; case PROP_TYPE_NUMBER: if (datatype == DATA_TYPE_STRING) { (void) nvpair_value_string(elem, &value); if (strcmp(value, "none") == 0) { isnone = B_TRUE; } else if (strcmp(value, "auto") == 0) { isauto = B_TRUE; } else if (zfs_nicestrtonum(hdl, value, ivalp) != 0) { goto error; } } else if (datatype == DATA_TYPE_UINT64) { (void) nvpair_value_uint64(elem, ivalp); } else { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' must be a number"), nvpair_name(elem)); goto error; } /* * Quota special: force 'none' and don't allow 0. */ if ((type & ZFS_TYPE_DATASET) && *ivalp == 0 && !isnone && (prop == ZFS_PROP_QUOTA || prop == ZFS_PROP_REFQUOTA)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "use 'none' to disable quota/refquota")); goto error; } /* * Special handling for "*_limit=none". In this case it's not * 0 but UINT64_MAX. */ if ((type & ZFS_TYPE_DATASET) && isnone && (prop == ZFS_PROP_FILESYSTEM_LIMIT || prop == ZFS_PROP_SNAPSHOT_LIMIT)) { *ivalp = UINT64_MAX; } /* * Special handling for setting 'refreservation' to 'auto'. Use * UINT64_MAX to tell the caller to use zfs_fix_auto_resv(). * 'auto' is only allowed on volumes. */ if (isauto) { switch (prop) { case ZFS_PROP_REFRESERVATION: if ((type & ZFS_TYPE_VOLUME) == 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s=auto' only allowed on " "volumes"), nvpair_name(elem)); goto error; } *ivalp = UINT64_MAX; break; default: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'auto' is invalid value for '%s'"), nvpair_name(elem)); goto error; } } break; case PROP_TYPE_INDEX: if (datatype != DATA_TYPE_STRING) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' must be a string"), nvpair_name(elem)); goto error; } (void) nvpair_value_string(elem, &value); if (zprop_string_to_index(prop, value, ivalp, type) != 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' must be one of '%s'"), propname, zprop_values(prop, type)); goto error; } break; default: abort(); } /* * Add the result to our return set of properties. */ if (*svalp != NULL) { if (nvlist_add_string(ret, propname, *svalp) != 0) { (void) no_memory(hdl); return (-1); } } else { if (nvlist_add_uint64(ret, propname, *ivalp) != 0) { (void) no_memory(hdl); return (-1); } } return (0); error: (void) zfs_error(hdl, EZFS_BADPROP, errbuf); return (-1); } static int addlist(libzfs_handle_t *hdl, char *propname, zprop_list_t **listp, zfs_type_t type) { int prop; zprop_list_t *entry; prop = zprop_name_to_prop(propname, type); if (prop != ZPROP_INVAL && !zprop_valid_for_type(prop, type, B_FALSE)) prop = ZPROP_INVAL; /* * When no property table entry can be found, return failure if * this is a pool property or if this isn't a user-defined * dataset property, */ if (prop == ZPROP_INVAL && ((type == ZFS_TYPE_POOL && !zpool_prop_feature(propname) && !zpool_prop_unsupported(propname)) || (type == ZFS_TYPE_DATASET && !zfs_prop_user(propname) && !zfs_prop_userquota(propname) && !zfs_prop_written(propname)))) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "invalid property '%s'"), propname); return (zfs_error(hdl, EZFS_BADPROP, dgettext(TEXT_DOMAIN, "bad property list"))); } if ((entry = zfs_alloc(hdl, sizeof (zprop_list_t))) == NULL) return (-1); entry->pl_prop = prop; if (prop == ZPROP_INVAL) { if ((entry->pl_user_prop = zfs_strdup(hdl, propname)) == NULL) { free(entry); return (-1); } entry->pl_width = strlen(propname); } else { entry->pl_width = zprop_width(prop, &entry->pl_fixed, type); } *listp = entry; return (0); } /* * Given a comma-separated list of properties, construct a property list * containing both user-defined and native properties. This function will * return a NULL list if 'all' is specified, which can later be expanded * by zprop_expand_list(). */ int zprop_get_list(libzfs_handle_t *hdl, char *props, zprop_list_t **listp, zfs_type_t type) { *listp = NULL; /* * If 'all' is specified, return a NULL list. */ if (strcmp(props, "all") == 0) return (0); /* * If no props were specified, return an error. */ if (props[0] == '\0') { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "no properties specified")); return (zfs_error(hdl, EZFS_BADPROP, dgettext(TEXT_DOMAIN, "bad property list"))); } /* * It would be nice to use getsubopt() here, but the inclusion of column * aliases makes this more effort than it's worth. */ while (*props != '\0') { size_t len; char *p; char c; if ((p = strchr(props, ',')) == NULL) { len = strlen(props); p = props + len; } else { len = p - props; } /* * Check for empty options. */ if (len == 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "empty property name")); return (zfs_error(hdl, EZFS_BADPROP, dgettext(TEXT_DOMAIN, "bad property list"))); } /* * Check all regular property names. */ c = props[len]; props[len] = '\0'; if (strcmp(props, "space") == 0) { static char *spaceprops[] = { "name", "avail", "used", "usedbysnapshots", "usedbydataset", "usedbyrefreservation", "usedbychildren", NULL }; int i; for (i = 0; spaceprops[i]; i++) { if (addlist(hdl, spaceprops[i], listp, type)) return (-1); listp = &(*listp)->pl_next; } } else { if (addlist(hdl, props, listp, type)) return (-1); listp = &(*listp)->pl_next; } props = p; if (c == ',') props++; } return (0); } void zprop_free_list(zprop_list_t *pl) { zprop_list_t *next; while (pl != NULL) { next = pl->pl_next; free(pl->pl_user_prop); free(pl); pl = next; } } typedef struct expand_data { zprop_list_t **last; libzfs_handle_t *hdl; zfs_type_t type; } expand_data_t; static int zprop_expand_list_cb(int prop, void *cb) { zprop_list_t *entry; expand_data_t *edp = cb; if ((entry = zfs_alloc(edp->hdl, sizeof (zprop_list_t))) == NULL) return (ZPROP_INVAL); entry->pl_prop = prop; entry->pl_width = zprop_width(prop, &entry->pl_fixed, edp->type); entry->pl_all = B_TRUE; *(edp->last) = entry; edp->last = &entry->pl_next; return (ZPROP_CONT); } int zprop_expand_list(libzfs_handle_t *hdl, zprop_list_t **plp, zfs_type_t type) { zprop_list_t *entry; zprop_list_t **last; expand_data_t exp; if (*plp == NULL) { /* * If this is the very first time we've been called for an 'all' * specification, expand the list to include all native * properties. */ last = plp; exp.last = last; exp.hdl = hdl; exp.type = type; if (zprop_iter_common(zprop_expand_list_cb, &exp, B_FALSE, B_FALSE, type) == ZPROP_INVAL) return (-1); /* * Add 'name' to the beginning of the list, which is handled * specially. */ if ((entry = zfs_alloc(hdl, sizeof (zprop_list_t))) == NULL) return (-1); entry->pl_prop = (type == ZFS_TYPE_POOL) ? ZPOOL_PROP_NAME : ZFS_PROP_NAME; entry->pl_width = zprop_width(entry->pl_prop, &entry->pl_fixed, type); entry->pl_all = B_TRUE; entry->pl_next = *plp; *plp = entry; } return (0); } int zprop_iter(zprop_func func, void *cb, boolean_t show_all, boolean_t ordered, zfs_type_t type) { return (zprop_iter_common(func, cb, show_all, ordered, type)); } /* * Fill given version buffer with zfs userland version */ void zfs_version_userland(char *version, int len) { (void) strlcpy(version, ZFS_META_ALIAS, len); } /* * Prints both zfs userland and kernel versions * Returns 0 on success, and -1 on error (with errno set) */ int zfs_version_print(void) { char zver_userland[128]; char zver_kernel[128]; zfs_version_userland(zver_userland, sizeof (zver_userland)); (void) printf("%s\n", zver_userland); if (zfs_version_kernel(zver_kernel, sizeof (zver_kernel)) == -1) { fprintf(stderr, "zfs_version_kernel() failed: %s\n", strerror(errno)); return (-1); } (void) printf("zfs-kmod-%s\n", zver_kernel); return (0); } /* * Return 1 if the user requested ANSI color output, and our terminal supports * it. Return 0 for no color. */ static int use_color(void) { static int use_color = -1; char *term; /* * Optimization: * * For each zpool invocation, we do a single check to see if we should * be using color or not, and cache that value for the lifetime of the * the zpool command. That makes it cheap to call use_color() when * we're printing with color. We assume that the settings are not going * to change during the invocation of a zpool command (the user isn't * going to change the ZFS_COLOR value while zpool is running, for * example). */ if (use_color != -1) { /* * We've already figured out if we should be using color or * not. Return the cached value. */ return (use_color); } term = getenv("TERM"); /* * The user sets the ZFS_COLOR env var set to enable zpool ANSI color * output. However if NO_COLOR is set (https://no-color.org/) then * don't use it. Also, don't use color if terminal doesn't support * it. */ if (libzfs_envvar_is_set("ZFS_COLOR") && !libzfs_envvar_is_set("NO_COLOR") && isatty(STDOUT_FILENO) && term && strcmp("dumb", term) != 0 && strcmp("unknown", term) != 0) { /* Color supported */ use_color = 1; } else { use_color = 0; } return (use_color); } /* * color_start() and color_end() are used for when you want to colorize a block * of text. For example: * * color_start(ANSI_RED_FG) * printf("hello"); * printf("world"); * color_end(); */ void color_start(char *color) { if (use_color()) printf("%s", color); } void color_end(void) { if (use_color()) printf(ANSI_RESET); } /* printf() with a color. If color is NULL, then do a normal printf. */ int printf_color(char *color, char *format, ...) { va_list aptr; int rc; if (color) color_start(color); va_start(aptr, format); rc = vprintf(format, aptr); va_end(aptr); if (color) color_end(); return (rc); } diff --git a/man/man5/zpool-features.5 b/man/man5/zpool-features.5 index 36c4343a1388..08a84ece205b 100644 --- a/man/man5/zpool-features.5 +++ b/man/man5/zpool-features.5 @@ -1,985 +1,985 @@ '\" te .\" Copyright (c) 2012, 2018 by Delphix. All rights reserved. .\" Copyright (c) 2013 by Saso Kiselkov. All rights reserved. .\" Copyright (c) 2014, Joyent, Inc. All rights reserved. .\" 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] .\" Copyright (c) 2019, Klara Inc. .\" Copyright (c) 2019, Allan Jude .TH ZPOOL-FEATURES 5 "Aug 24, 2020" OpenZFS .SH NAME zpool\-features \- ZFS pool feature descriptions .SH DESCRIPTION .sp .LP ZFS pool on\-disk format versions are specified via "features" which replace the old on\-disk format numbers (the last supported on\-disk format number is 28). To enable a feature on a pool use the \fBupgrade\fR subcommand of the zpool(8) command, or set the \fBfeature@\fR\fIfeature_name\fR property to \fBenabled\fR. .sp .LP The pool format does not affect file system version compatibility or the ability to send file systems between pools. .sp .LP Since most features can be enabled independently of each other the on\-disk format of the pool is specified by the set of all features marked as \fBactive\fR on the pool. If the pool was created by another software version this set may include unsupported features. .SS "Identifying features" .sp .LP Every feature has a GUID of the form \fIcom.example:feature_name\fR. The reversed DNS name ensures that the feature's GUID is unique across all ZFS implementations. When unsupported features are encountered on a pool they will be identified by their GUIDs. Refer to the documentation for the ZFS implementation that created the pool for information about those features. .sp .LP Each supported feature also has a short name. By convention a feature's short name is the portion of its GUID which follows the ':' (e.g. \fIcom.example:feature_name\fR would have the short name \fIfeature_name\fR), however a feature's short name may differ across ZFS implementations if following the convention would result in name conflicts. .SS "Feature states" .sp .LP Features can be in one of three states: .sp .ne 2 .na \fBactive\fR .ad .RS 12n This feature's on\-disk format changes are in effect on the pool. Support for this feature is required to import the pool in read\-write mode. If this feature is not read-only compatible, support is also required to import the pool in read\-only mode (see "Read\-only compatibility"). .RE .sp .ne 2 .na \fBenabled\fR .ad .RS 12n An administrator has marked this feature as enabled on the pool, but the feature's on\-disk format changes have not been made yet. The pool can still be imported by software that does not support this feature, but changes may be made to the on\-disk format at any time which will move the feature to the \fBactive\fR state. Some features may support returning to the \fBenabled\fR state after becoming \fBactive\fR. See feature\-specific documentation for details. .RE .sp .ne 2 .na \fBdisabled\fR .ad .RS 12n This feature's on\-disk format changes have not been made and will not be made unless an administrator moves the feature to the \fBenabled\fR state. Features cannot be disabled once they have been enabled. .RE .sp .LP The state of supported features is exposed through pool properties of the form \fIfeature@short_name\fR. .SS "Read\-only compatibility" .sp .LP Some features may make on\-disk format changes that do not interfere with other software's ability to read from the pool. These features are referred to as "read\-only compatible". If all unsupported features on a pool are read\-only compatible, the pool can be imported in read\-only mode by setting the \fBreadonly\fR property during import (see zpool(8) for details on importing pools). .SS "Unsupported features" .sp .LP For each unsupported feature enabled on an imported pool a pool property named \fIunsupported@feature_name\fR will indicate why the import was allowed despite the unsupported feature. Possible values for this property are: .sp .ne 2 .na \fBinactive\fR .ad .RS 12n The feature is in the \fBenabled\fR state and therefore the pool's on\-disk format is still compatible with software that does not support this feature. .RE .sp .ne 2 .na \fBreadonly\fR .ad .RS 12n The feature is read\-only compatible and the pool has been imported in read\-only mode. .RE .SS "Feature dependencies" .sp .LP Some features depend on other features being enabled in order to function properly. Enabling a feature will automatically enable any features it depends on. .SH FEATURES .sp .LP The following features are supported on this system: .sp .ne 2 .na \fBallocation_classes\fR .ad .RS 4n .TS l l . GUID org.zfsonlinux:allocation_classes READ\-ONLY COMPATIBLE yes DEPENDENCIES none .TE This feature enables support for separate allocation classes. This feature becomes \fBactive\fR when a dedicated allocation class vdev (dedup or special) is created with the \fBzpool create\fR or \fBzpool add\fR subcommands. With device removal, it can be returned to the \fBenabled\fR state if all the dedicated allocation class vdevs are removed. .RE .sp .ne 2 .na \fBasync_destroy\fR .ad .RS 4n .TS l l . GUID com.delphix:async_destroy READ\-ONLY COMPATIBLE yes DEPENDENCIES none .TE Destroying a file system requires traversing all of its data in order to return its used space to the pool. Without \fBasync_destroy\fR the file system is not fully removed until all space has been reclaimed. If the destroy operation is interrupted by a reboot or power outage the next attempt to open the pool will need to complete the destroy operation synchronously. When \fBasync_destroy\fR is enabled the file system's data will be reclaimed by a background process, allowing the destroy operation to complete without traversing the entire file system. The background process is able to resume interrupted destroys after the pool has been opened, eliminating the need to finish interrupted destroys as part of the open operation. The amount of space remaining to be reclaimed by the background process is available through the \fBfreeing\fR property. This feature is only \fBactive\fR while \fBfreeing\fR is non\-zero. .RE .sp .ne 2 .na \fBbookmarks\fR .ad .RS 4n .TS l l . GUID com.delphix:bookmarks READ\-ONLY COMPATIBLE yes DEPENDENCIES extensible_dataset .TE This feature enables use of the \fBzfs bookmark\fR subcommand. This feature is \fBactive\fR while any bookmarks exist in the pool. All bookmarks in the pool can be listed by running \fBzfs list -t bookmark -r \fIpoolname\fR\fR. .RE .sp .ne 2 .na \fBbookmark_v2\fR .ad .RS 4n .TS l l . GUID com.datto:bookmark_v2 READ\-ONLY COMPATIBLE no DEPENDENCIES bookmark, extensible_dataset .TE This feature enables the creation and management of larger bookmarks which are needed for other features in ZFS. This feature becomes \fBactive\fR when a v2 bookmark is created and will be returned to the \fBenabled\fR state when all v2 bookmarks are destroyed. .RE .sp .ne 2 .na \fBbookmark_written\fR .ad .RS 4n .TS l l . GUID com.delphix:bookmark_written READ\-ONLY COMPATIBLE no DEPENDENCIES bookmark, extensible_dataset, bookmark_v2 .TE This feature enables additional bookmark accounting fields, enabling the written# property (space written since a bookmark) and estimates of send stream sizes for incrementals from bookmarks. This feature becomes \fBactive\fR when a bookmark is created and will be returned to the \fBenabled\fR state when all bookmarks with these fields are destroyed. .RE .sp .ne 2 .na \fBdevice_rebuild\fR .ad .RS 4n .TS l l . GUID org.openzfs:device_rebuild READ\-ONLY COMPATIBLE yes DEPENDENCIES none .TE This feature enables the ability for the \fBzpool attach\fR and \fBzpool replace\fR subcommands to perform sequential reconstruction (instead of healing reconstruction) when resilvering. Sequential reconstruction resilvers a device in LBA order without immediately verifying the checksums. Once complete a scrub is started which then verifies the checksums. This approach allows full redundancy to be restored to the pool in the minimum amount of time. This two phase approach will take longer than a healing resilver when the time to verify the checksums is included. However, unless there is additional pool damage no checksum errors should be reported by the scrub. This feature is incompatible with raidz configurations. This feature becomes \fBactive\fR while a sequential resilver is in progress, and returns to \fBenabled\fR when the resilver completes. .RE .sp .ne 2 .na \fBdevice_removal\fR .ad .RS 4n .TS l l . GUID com.delphix:device_removal READ\-ONLY COMPATIBLE no DEPENDENCIES none .TE This feature enables the \fBzpool remove\fR subcommand to remove top-level vdevs, evacuating them to reduce the total size of the pool. This feature becomes \fBactive\fR when the \fBzpool remove\fR subcommand is used on a top-level vdev, and will never return to being \fBenabled\fR. .RE .sp .ne 2 .na \fBedonr\fR .ad .RS 4n .TS l l . GUID org.illumos:edonr READ\-ONLY COMPATIBLE no DEPENDENCIES extensible_dataset .TE This feature enables the use of the Edon-R hash algorithm for checksum, including for nopwrite (if compression is also enabled, an overwrite of a block whose checksum matches the data being written will be ignored). In an abundance of caution, Edon-R requires verification when used with dedup: \fBzfs set dedup=edonr,verify\fR. See \fBzfs\fR(8). Edon-R is a very high-performance hash algorithm that was part of the NIST SHA-3 competition. It provides extremely high hash performance (over 350% faster than SHA-256), but was not selected because of its unsuitability as a general purpose secure hash algorithm. This implementation utilizes the new salted checksumming functionality in ZFS, which means that the checksum is pre-seeded with a secret 256-bit random key (stored on the pool) before being fed the data block to be checksummed. Thus the produced checksums are unique to a given pool. When the \fBedonr\fR feature is set to \fBenabled\fR, the administrator can turn on the \fBedonr\fR checksum on any dataset using the \fBzfs set checksum=edonr\fR. See zfs(8). This feature becomes \fBactive\fR once a \fBchecksum\fR property has been set to \fBedonr\fR, and will return to being \fBenabled\fR once all filesystems that have ever had their checksum set to \fBedonr\fR are destroyed. FreeBSD does not support the \fBedonr\fR feature. .RE .sp .ne 2 .na \fBembedded_data\fR .ad .RS 4n .TS l l . GUID com.delphix:embedded_data READ\-ONLY COMPATIBLE no DEPENDENCIES none .TE This feature improves the performance and compression ratio of highly-compressible blocks. Blocks whose contents can compress to 112 bytes or smaller can take advantage of this feature. When this feature is enabled, the contents of highly-compressible blocks are stored in the block "pointer" itself (a misnomer in this case, as it contains the compressed data, rather than a pointer to its location on disk). Thus the space of the block (one sector, typically 512 bytes or 4KB) is saved, and no additional i/o is needed to read and write the data block. This feature becomes \fBactive\fR as soon as it is enabled and will never return to being \fBenabled\fR. .RE .sp .ne 2 .na \fBempty_bpobj\fR .ad .RS 4n .TS l l . GUID com.delphix:empty_bpobj READ\-ONLY COMPATIBLE yes DEPENDENCIES none .TE This feature increases the performance of creating and using a large number of snapshots of a single filesystem or volume, and also reduces the disk space required. When there are many snapshots, each snapshot uses many Block Pointer Objects (bpobj's) to track blocks associated with that snapshot. However, in common use cases, most of these bpobj's are empty. This feature allows us to create each bpobj on-demand, thus eliminating the empty bpobjs. This feature is \fBactive\fR while there are any filesystems, volumes, or snapshots which were created after enabling this feature. .RE .sp .ne 2 .na \fBenabled_txg\fR .ad .RS 4n .TS l l . GUID com.delphix:enabled_txg READ\-ONLY COMPATIBLE yes DEPENDENCIES none .TE Once this feature is enabled ZFS records the transaction group number in which new features are enabled. This has no user-visible impact, but other features may depend on this feature. This feature becomes \fBactive\fR as soon as it is enabled and will never return to being \fBenabled\fB. .RE .sp .ne 2 .na \fBencryption\fR .ad .RS 4n .TS l l . GUID com.datto:encryption READ\-ONLY COMPATIBLE no DEPENDENCIES bookmark_v2, extensible_dataset .TE This feature enables the creation and management of natively encrypted datasets. This feature becomes \fBactive\fR when an encrypted dataset is created and will be returned to the \fBenabled\fR state when all datasets that use this feature are destroyed. .RE .sp .ne 2 .na \fBextensible_dataset\fR .ad .RS 4n .TS l l . GUID com.delphix:extensible_dataset READ\-ONLY COMPATIBLE no DEPENDENCIES none .TE This feature allows more flexible use of internal ZFS data structures, and exists for other features to depend on. This feature will be \fBactive\fR when the first dependent feature uses it, and will be returned to the \fBenabled\fR state when all datasets that use this feature are destroyed. .RE .sp .ne 2 .na \fBfilesystem_limits\fR .ad .RS 4n .TS l l . GUID com.joyent:filesystem_limits READ\-ONLY COMPATIBLE yes DEPENDENCIES extensible_dataset .TE This feature enables filesystem and snapshot limits. These limits can be used to control how many filesystems and/or snapshots can be created at the point in the tree on which the limits are set. This feature is \fBactive\fR once either of the limit properties has been set on a dataset. Once activated the feature is never deactivated. .RE .sp .ne 2 .na \fBhole_birth\fR .ad .RS 4n .TS l l . GUID com.delphix:hole_birth READ\-ONLY COMPATIBLE no DEPENDENCIES enabled_txg .TE This feature has/had bugs, the result of which is that, if you do a \fBzfs send -i\fR (or \fB-R\fR, since it uses \fB-i\fR) from an affected dataset, the receiver will not see any checksum or other errors, but the resulting destination snapshot will not match the source. Its use by \fBzfs send -i\fR has been disabled by default. See the \fBsend_holes_without_birth_time\fR module parameter in zfs-module-parameters(5). This feature improves performance of incremental sends (\fBzfs send -i\fR) and receives for objects with many holes. The most common case of hole-filled objects is zvols. An incremental send stream from snapshot \fBA\fR to snapshot \fBB\fR contains information about every block that changed between \fBA\fR and \fBB\fR. Blocks which did not change between those snapshots can be identified and omitted from the stream using a piece of metadata called the 'block birth time', but birth times are not recorded for holes (blocks filled only with zeroes). Since holes created after \fBA\fR cannot be distinguished from holes created before \fBA\fR, information about every hole in the entire filesystem or zvol is included in the send stream. For workloads where holes are rare this is not a problem. However, when incrementally replicating filesystems or zvols with many holes (for example a zvol formatted with another filesystem) a lot of time will be spent sending and receiving unnecessary information about holes that already exist on the receiving side. Once the \fBhole_birth\fR feature has been enabled the block birth times of all new holes will be recorded. Incremental sends between snapshots created after this feature is enabled will use this new metadata to avoid sending information about holes that already exist on the receiving side. This feature becomes \fBactive\fR as soon as it is enabled and will never return to being \fBenabled\fB. .RE .sp .ne 2 .na \fBlarge_blocks\fR .ad .RS 4n .TS l l . GUID org.open-zfs:large_blocks READ\-ONLY COMPATIBLE no DEPENDENCIES extensible_dataset .TE The \fBlarge_block\fR feature allows the record size on a dataset to be set larger than 128KB. This feature becomes \fBactive\fR once a dataset contains a file with a block size larger than 128KB, and will return to being \fBenabled\fR once all filesystems that have ever had their recordsize larger than 128KB are destroyed. .RE .sp .ne 2 .na \fBlarge_dnode\fR .ad .RS 4n .TS l l . GUID org.zfsonlinux:large_dnode READ\-ONLY COMPATIBLE no DEPENDENCIES extensible_dataset .TE The \fBlarge_dnode\fR feature allows the size of dnodes in a dataset to be set larger than 512B. This feature becomes \fBactive\fR once a dataset contains an object with a dnode larger than 512B, which occurs as a result of setting the \fBdnodesize\fR dataset property to a value other than \fBlegacy\fR. The feature will return to being \fBenabled\fR once all filesystems that have ever contained a dnode larger than 512B are destroyed. Large dnodes allow more data to be stored in the bonus buffer, thus potentially improving performance by avoiding the use of spill blocks. .RE .sp .ne 2 .na \fB\fBlivelist\fR\fR .ad .RS 4n .TS l l . GUID com.delphix:livelist READ\-ONLY COMPATIBLE yes DEPENDENCIES none .TE This feature allows clones to be deleted faster than the traditional method when a large number of random/sparse writes have been made to the clone. All blocks allocated and freed after a clone is created are tracked by the the clone's livelist which is referenced during the deletion of the clone. The feature is activated when a clone is created and remains active until all clones have been destroyed. .RE .sp .ne 2 .na \fBlog_spacemap\fR .ad .RS 4n .TS l l . GUID com.delphix:log_spacemap READ\-ONLY COMPATIBLE yes DEPENDENCIES com.delphix:spacemap_v2 .TE This feature improves performance for heavily-fragmented pools, especially when workloads are heavy in random-writes. It does so by logging all the metaslab changes on a single spacemap every TXG instead of scattering multiple writes to all the metaslab spacemaps. This feature becomes \fBactive\fR as soon as it is enabled and will never return to being \fBenabled\fR. .RE .sp .ne 2 .na \fBlz4_compress\fR .ad .RS 4n .TS l l . GUID org.illumos:lz4_compress READ\-ONLY COMPATIBLE no DEPENDENCIES none .TE \fBlz4\fR is a high-performance real-time compression algorithm that features significantly faster compression and decompression as well as a higher compression ratio than the older \fBlzjb\fR compression. Typically, \fBlz4\fR compression is approximately 50% faster on compressible data and 200% faster on incompressible data than \fBlzjb\fR. It is also approximately 80% faster on decompression, while giving approximately 10% better compression ratio. When the \fBlz4_compress\fR feature is set to \fBenabled\fR, the administrator can turn on \fBlz4\fR compression on any dataset on the pool using the zfs(8) command. Please note that doing so will immediately activate the \fBlz4_compress\fR feature on the underlying pool using the zfs(8) command. Also, all newly written metadata will be compressed with \fBlz4\fR algorithm. Since this feature is not read-only compatible, this operation will render the pool unimportable on systems without support for the \fBlz4_compress\fR feature. Booting off of \fBlz4\fR-compressed root pools is supported. This feature becomes \fBactive\fR as soon as it is enabled and will never return to being \fBenabled\fB. .RE .sp .ne 2 .na \fBmulti_vdev_crash_dump\fR .ad .RS 4n .TS l l . GUID com.joyent:multi_vdev_crash_dump READ\-ONLY COMPATIBLE no DEPENDENCIES none .TE This feature allows a dump device to be configured with a pool comprised of multiple vdevs. Those vdevs may be arranged in any mirrored or raidz configuration. When the \fBmulti_vdev_crash_dump\fR feature is set to \fBenabled\fR, the administrator can use the \fBdumpadm\fR(1M) command to configure a dump device on a pool comprised of multiple vdevs. -Under Linux this feature is registered for compatibility but not used. -New pools created under Linux will have the feature \fBenabled\fR but -will never transition to \fB\fBactive\fR. This functionality is not -required in order to support crash dumps under Linux. Existing pools -where this feature is \fB\fBactive\fR can be imported. +Under FreeBSD and Linux this feature is registered for compatibility but not +used. New pools created under FreeBSD and Linux will have the feature +\fBenabled\fR but will never transition to \fB\fBactive\fR. This functionality +is not required in order to support crash dumps under FreeBSD and Linux. +Existing pools where this feature is \fB\fBactive\fR can be imported. .RE .sp .ne 2 .na \fBobsolete_counts\fR .ad .RS 4n .TS l l . GUID com.delphix:obsolete_counts READ\-ONLY COMPATIBLE yes DEPENDENCIES device_removal .TE This feature is an enhancement of device_removal, which will over time reduce the memory used to track removed devices. When indirect blocks are freed or remapped, we note that their part of the indirect mapping is "obsolete", i.e. no longer needed. This feature becomes \fBactive\fR when the \fBzpool remove\fR subcommand is used on a top-level vdev, and will never return to being \fBenabled\fR. .RE .sp .ne 2 .na \fBproject_quota\fR .ad .RS 4n .TS l l . GUID org.zfsonlinux:project_quota READ\-ONLY COMPATIBLE yes DEPENDENCIES extensible_dataset .TE This feature allows administrators to account the spaces and objects usage information against the project identifier (ID). The project ID is new object-based attribute. When upgrading an existing filesystem, object without project ID attribute will be assigned a zero project ID. After this feature is enabled, newly created object will inherit its parent directory's project ID if the parent inherit flag is set (via \fBchattr +/-P\fR or \fBzfs project [-s|-C]\fR). Otherwise, the new object's project ID will be set as zero. An object's project ID can be changed at anytime by the owner (or privileged user) via \fBchattr -p $prjid\fR or \fBzfs project -p $prjid\fR. This feature will become \fBactive\fR as soon as it is enabled and will never return to being \fBdisabled\fR. Each filesystem will be upgraded automatically when remounted or when new file is created under that filesystem. The upgrade can also be triggered on filesystems via `zfs set version=current `. The upgrade process runs in the background and may take a while to complete for the filesystems containing a large number of files. .RE .sp .ne 2 .na \fB\fBredaction_bookmarks\fR\fR .ad .RS 4n .TS l l . GUID com.delphix:redaction_bookmarks READ\-ONLY COMPATIBLE no DEPENDENCIES bookmarks, extensible_dataset .TE This feature enables the use of the redacted zfs send. Redacted \fBzfs send\fR creates redaction bookmarks, which store the list of blocks redacted by the send that created them. For more information about redacted send, see \fBzfs\fR(8). .RE .sp .ne 2 .na \fB\fBredacted_datasets\fR\fR .ad .RS 4n .TS l l . GUID com.delphix:redacted_datasets READ\-ONLY COMPATIBLE no DEPENDENCIES extensible_dataset .TE This feature enables the receiving of redacted zfs send streams. Redacted zfs send streams create redacted datasets when received. These datasets are missing some of their blocks, and so cannot be safely mounted, and their contents cannot be safely read. For more information about redacted receive, see \fBzfs\fR(8). .RE .sp .ne 2 .na \fBresilver_defer\fR .ad .RS 4n .TS l l . GUID com.datto:resilver_defer READ\-ONLY COMPATIBLE yes DEPENDENCIES none .TE This feature allows zfs to postpone new resilvers if an existing one is already in progress. Without this feature, any new resilvers will cause the currently running one to be immediately restarted from the beginning. This feature becomes \fBactive\fR once a resilver has been deferred, and returns to being \fBenabled\fR when the deferred resilver begins. .RE .sp .ne 2 .na \fBsha512\fR .ad .RS 4n .TS l l . GUID org.illumos:sha512 READ\-ONLY COMPATIBLE no DEPENDENCIES extensible_dataset .TE This feature enables the use of the SHA-512/256 truncated hash algorithm (FIPS 180-4) for checksum and dedup. The native 64-bit arithmetic of SHA-512 provides an approximate 50% performance boost over SHA-256 on 64-bit hardware and is thus a good minimum-change replacement candidate for systems where hash performance is important, but these systems cannot for whatever reason utilize the faster \fBskein\fR and \fBedonr\fR algorithms. When the \fBsha512\fR feature is set to \fBenabled\fR, the administrator can turn on the \fBsha512\fR checksum on any dataset using \fBzfs set checksum=sha512\fR. See zfs(8). This feature becomes \fBactive\fR once a \fBchecksum\fR property has been set to \fBsha512\fR, and will return to being \fBenabled\fR once all filesystems that have ever had their checksum set to \fBsha512\fR are destroyed. .RE .sp .ne 2 .na \fBskein\fR .ad .RS 4n .TS l l . GUID org.illumos:skein READ\-ONLY COMPATIBLE no DEPENDENCIES extensible_dataset .TE This feature enables the use of the Skein hash algorithm for checksum and dedup. Skein is a high-performance secure hash algorithm that was a finalist in the NIST SHA-3 competition. It provides a very high security margin and high performance on 64-bit hardware (80% faster than SHA-256). This implementation also utilizes the new salted checksumming functionality in ZFS, which means that the checksum is pre-seeded with a secret 256-bit random key (stored on the pool) before being fed the data block to be checksummed. Thus the produced checksums are unique to a given pool, preventing hash collision attacks on systems with dedup. When the \fBskein\fR feature is set to \fBenabled\fR, the administrator can turn on the \fBskein\fR checksum on any dataset using \fBzfs set checksum=skein\fR. See zfs(8). This feature becomes \fBactive\fR once a \fBchecksum\fR property has been set to \fBskein\fR, and will return to being \fBenabled\fR once all filesystems that have ever had their checksum set to \fBskein\fR are destroyed. .RE .sp .ne 2 .na \fBspacemap_histogram\fR .ad .RS 4n .TS l l . GUID com.delphix:spacemap_histogram READ\-ONLY COMPATIBLE yes DEPENDENCIES none .TE This features allows ZFS to maintain more information about how free space is organized within the pool. If this feature is \fBenabled\fR, ZFS will set this feature to \fBactive\fR when a new space map object is created or an existing space map is upgraded to the new format. Once the feature is \fBactive\fR, it will remain in that state until the pool is destroyed. .RE .sp .ne 2 .na \fBspacemap_v2\fR .ad .RS 4n .TS l l . GUID com.delphix:spacemap_v2 READ\-ONLY COMPATIBLE yes DEPENDENCIES none .TE This feature enables the use of the new space map encoding which consists of two words (instead of one) whenever it is advantageous. The new encoding allows space maps to represent large regions of space more efficiently on-disk while also increasing their maximum addressable offset. This feature becomes \fBactive\fR once it is \fBenabled\fR, and never returns back to being \fBenabled\fR. .RE .sp .ne 2 .na \fBuserobj_accounting\fR .ad .RS 4n .TS l l . GUID org.zfsonlinux:userobj_accounting READ\-ONLY COMPATIBLE yes DEPENDENCIES extensible_dataset .TE This feature allows administrators to account the object usage information by user and group. This feature becomes \fBactive\fR as soon as it is enabled and will never return to being \fBenabled\fR. Each filesystem will be upgraded automatically when remounted, or when new files are created under that filesystem. The upgrade can also be started manually on filesystems by running `zfs set version=current `. The upgrade process runs in the background and may take a while to complete for filesystems containing a large number of files. .RE .sp .ne 2 .na \fBzpool_checkpoint\fR .ad .RS 4n .TS l l . GUID com.delphix:zpool_checkpoint READ\-ONLY COMPATIBLE yes DEPENDENCIES none .TE This feature enables the \fBzpool checkpoint\fR subcommand that can checkpoint the state of the pool at the time it was issued and later rewind back to it or discard it. This feature becomes \fBactive\fR when the \fBzpool checkpoint\fR subcommand is used to checkpoint the pool. The feature will only return back to being \fBenabled\fR when the pool is rewound or the checkpoint has been discarded. .RE .sp .ne 2 .na \fBzstd_compress\fR .ad .RS 4n .TS l l . GUID org.freebsd:zstd_compress READ\-ONLY COMPATIBLE no DEPENDENCIES extensible_dataset .TE \fBzstd\fR is a high-performance compression algorithm that features a combination of high compression ratios and high speed. Compared to \fBgzip\fR, \fBzstd\fR offers slighty better compression at much higher speeds. Compared to \fBlz4\fR, \fBzstd\fR offers much better compression while being only modestly slower. Typically, \fBzstd\fR compression speed ranges from 250 to 500 MB/s per thread and decompression speed is over 1 GB/s per thread. When the \fBzstd\fR feature is set to \fBenabled\fR, the administrator can turn on \fBzstd\fR compression of any dataset by running `zfs set compress=zstd `. This feature becomes \fBactive\fR once a \fBcompress\fR property has been set to \fBzstd\fR, and will return to being \fBenabled\fR once all filesystems that have ever had their compress property set to \fBzstd\fR are destroyed. Booting off of \fBzstd\fR-compressed root pools is not yet supported. .RE .SH "SEE ALSO" zpool(8) diff --git a/module/os/freebsd/zfs/zfs_vfsops.c b/module/os/freebsd/zfs/zfs_vfsops.c index 671b939b593b..55ed2e25d5c3 100644 --- a/module/os/freebsd/zfs/zfs_vfsops.c +++ b/module/os/freebsd/zfs/zfs_vfsops.c @@ -1,2304 +1,2304 @@ /* * 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) 2011 Pawel Jakub Dawidek . * All rights reserved. * Copyright (c) 2012, 2015 by Delphix. All rights reserved. * Copyright (c) 2014 Integros [integros.com] * Copyright 2016 Nexenta Systems, Inc. All rights reserved. */ /* Portions Copyright 2010 Robert Milkowski */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_comutil.h" #ifndef MNTK_VMSETSIZE_BUG #define MNTK_VMSETSIZE_BUG 0 #endif #ifndef MNTK_NOMSYNC #define MNTK_NOMSYNC 8 #endif /* BEGIN CSTYLED */ struct mtx zfs_debug_mtx; MTX_SYSINIT(zfs_debug_mtx, &zfs_debug_mtx, "zfs_debug", MTX_DEF); SYSCTL_NODE(_vfs, OID_AUTO, zfs, CTLFLAG_RW, 0, "ZFS file system"); int zfs_super_owner; SYSCTL_INT(_vfs_zfs, OID_AUTO, super_owner, CTLFLAG_RW, &zfs_super_owner, 0, "File system owner can perform privileged operation on his file systems"); int zfs_debug_level; SYSCTL_INT(_vfs_zfs, OID_AUTO, debug, CTLFLAG_RWTUN, &zfs_debug_level, 0, "Debug level"); SYSCTL_NODE(_vfs_zfs, OID_AUTO, version, CTLFLAG_RD, 0, "ZFS versions"); static int zfs_version_acl = ZFS_ACL_VERSION; SYSCTL_INT(_vfs_zfs_version, OID_AUTO, acl, CTLFLAG_RD, &zfs_version_acl, 0, "ZFS_ACL_VERSION"); static int zfs_version_spa = SPA_VERSION; SYSCTL_INT(_vfs_zfs_version, OID_AUTO, spa, CTLFLAG_RD, &zfs_version_spa, 0, "SPA_VERSION"); static int zfs_version_zpl = ZPL_VERSION; SYSCTL_INT(_vfs_zfs_version, OID_AUTO, zpl, CTLFLAG_RD, &zfs_version_zpl, 0, "ZPL_VERSION"); /* END CSTYLED */ static int zfs_quotactl(vfs_t *vfsp, int cmds, uid_t id, void *arg); static int zfs_mount(vfs_t *vfsp); static int zfs_umount(vfs_t *vfsp, int fflag); static int zfs_root(vfs_t *vfsp, int flags, vnode_t **vpp); static int zfs_statfs(vfs_t *vfsp, struct statfs *statp); static int zfs_vget(vfs_t *vfsp, ino_t ino, int flags, vnode_t **vpp); static int zfs_sync(vfs_t *vfsp, int waitfor); #if __FreeBSD_version >= 1300098 static int zfs_checkexp(vfs_t *vfsp, struct sockaddr *nam, uint64_t *extflagsp, struct ucred **credanonp, int *numsecflavors, int *secflavors); #else static int zfs_checkexp(vfs_t *vfsp, struct sockaddr *nam, int *extflagsp, struct ucred **credanonp, int *numsecflavors, int **secflavors); #endif static int zfs_fhtovp(vfs_t *vfsp, fid_t *fidp, int flags, vnode_t **vpp); static void zfs_freevfs(vfs_t *vfsp); struct vfsops zfs_vfsops = { .vfs_mount = zfs_mount, .vfs_unmount = zfs_umount, #if __FreeBSD_version >= 1300049 .vfs_root = vfs_cache_root, .vfs_cachedroot = zfs_root, #else .vfs_root = zfs_root, #endif .vfs_statfs = zfs_statfs, .vfs_vget = zfs_vget, .vfs_sync = zfs_sync, .vfs_checkexp = zfs_checkexp, .vfs_fhtovp = zfs_fhtovp, .vfs_quotactl = zfs_quotactl, }; VFS_SET(zfs_vfsops, zfs, VFCF_JAIL | VFCF_DELEGADMIN); /* * We need to keep a count of active fs's. * This is necessary to prevent our module * from being unloaded after a umount -f */ static uint32_t zfs_active_fs_count = 0; int zfs_get_temporary_prop(dsl_dataset_t *ds, zfs_prop_t zfs_prop, uint64_t *val, char *setpoint) { int error; zfsvfs_t *zfvp; vfs_t *vfsp; objset_t *os; uint64_t tmp = *val; error = dmu_objset_from_ds(ds, &os); if (error != 0) return (error); error = getzfsvfs_impl(os, &zfvp); if (error != 0) return (error); if (zfvp == NULL) return (ENOENT); vfsp = zfvp->z_vfs; switch (zfs_prop) { case ZFS_PROP_ATIME: if (vfs_optionisset(vfsp, MNTOPT_NOATIME, NULL)) tmp = 0; if (vfs_optionisset(vfsp, MNTOPT_ATIME, NULL)) tmp = 1; break; case ZFS_PROP_DEVICES: if (vfs_optionisset(vfsp, MNTOPT_NODEVICES, NULL)) tmp = 0; if (vfs_optionisset(vfsp, MNTOPT_DEVICES, NULL)) tmp = 1; break; case ZFS_PROP_EXEC: if (vfs_optionisset(vfsp, MNTOPT_NOEXEC, NULL)) tmp = 0; if (vfs_optionisset(vfsp, MNTOPT_EXEC, NULL)) tmp = 1; break; case ZFS_PROP_SETUID: if (vfs_optionisset(vfsp, MNTOPT_NOSETUID, NULL)) tmp = 0; if (vfs_optionisset(vfsp, MNTOPT_SETUID, NULL)) tmp = 1; break; case ZFS_PROP_READONLY: if (vfs_optionisset(vfsp, MNTOPT_RW, NULL)) tmp = 0; if (vfs_optionisset(vfsp, MNTOPT_RO, NULL)) tmp = 1; break; case ZFS_PROP_XATTR: if (zfvp->z_flags & ZSB_XATTR) tmp = zfvp->z_xattr; break; case ZFS_PROP_NBMAND: if (vfs_optionisset(vfsp, MNTOPT_NONBMAND, NULL)) tmp = 0; if (vfs_optionisset(vfsp, MNTOPT_NBMAND, NULL)) tmp = 1; break; default: vfs_unbusy(vfsp); return (ENOENT); } vfs_unbusy(vfsp); if (tmp != *val) { (void) strcpy(setpoint, "temporary"); *val = tmp; } return (0); } static int zfs_getquota(zfsvfs_t *zfsvfs, uid_t id, int isgroup, struct dqblk64 *dqp) { int error = 0; char buf[32]; uint64_t usedobj, quotaobj; uint64_t quota, used = 0; timespec_t now; usedobj = isgroup ? DMU_GROUPUSED_OBJECT : DMU_USERUSED_OBJECT; quotaobj = isgroup ? zfsvfs->z_groupquota_obj : zfsvfs->z_userquota_obj; if (quotaobj == 0 || zfsvfs->z_replay) { error = ENOENT; goto done; } (void) sprintf(buf, "%llx", (longlong_t)id); if ((error = zap_lookup(zfsvfs->z_os, quotaobj, buf, sizeof (quota), 1, "a)) != 0) { dprintf("%s(%d): quotaobj lookup failed\n", __FUNCTION__, __LINE__); goto done; } /* * quota(8) uses bsoftlimit as "quoota", and hardlimit as "limit". * So we set them to be the same. */ dqp->dqb_bsoftlimit = dqp->dqb_bhardlimit = btodb(quota); error = zap_lookup(zfsvfs->z_os, usedobj, buf, sizeof (used), 1, &used); if (error && error != ENOENT) { dprintf("%s(%d): usedobj failed; %d\n", __FUNCTION__, __LINE__, error); goto done; } dqp->dqb_curblocks = btodb(used); dqp->dqb_ihardlimit = dqp->dqb_isoftlimit = 0; vfs_timestamp(&now); /* * Setting this to 0 causes FreeBSD quota(8) to print * the number of days since the epoch, which isn't * particularly useful. */ dqp->dqb_btime = dqp->dqb_itime = now.tv_sec; done: return (error); } static int zfs_quotactl(vfs_t *vfsp, int cmds, uid_t id, void *arg) { zfsvfs_t *zfsvfs = vfsp->vfs_data; struct thread *td; int cmd, type, error = 0; int bitsize; zfs_userquota_prop_t quota_type; struct dqblk64 dqblk = { 0 }; td = curthread; cmd = cmds >> SUBCMDSHIFT; type = cmds & SUBCMDMASK; ZFS_ENTER(zfsvfs); if (id == -1) { switch (type) { case USRQUOTA: id = td->td_ucred->cr_ruid; break; case GRPQUOTA: id = td->td_ucred->cr_rgid; break; default: error = EINVAL; if (cmd == Q_QUOTAON || cmd == Q_QUOTAOFF) vfs_unbusy(vfsp); goto done; } } /* * Map BSD type to: * ZFS_PROP_USERUSED, * ZFS_PROP_USERQUOTA, * ZFS_PROP_GROUPUSED, * ZFS_PROP_GROUPQUOTA */ switch (cmd) { case Q_SETQUOTA: case Q_SETQUOTA32: if (type == USRQUOTA) quota_type = ZFS_PROP_USERQUOTA; else if (type == GRPQUOTA) quota_type = ZFS_PROP_GROUPQUOTA; else error = EINVAL; break; case Q_GETQUOTA: case Q_GETQUOTA32: if (type == USRQUOTA) quota_type = ZFS_PROP_USERUSED; else if (type == GRPQUOTA) quota_type = ZFS_PROP_GROUPUSED; else error = EINVAL; break; } /* * Depending on the cmd, we may need to get * the ruid and domain (see fuidstr_to_sid?), * the fuid (how?), or other information. * Create fuid using zfs_fuid_create(zfsvfs, id, * ZFS_OWNER or ZFS_GROUP, cr, &fuidp)? * I think I can use just the id? * * Look at zfs_id_overquota() to look up a quota. * zap_lookup(something, quotaobj, fuidstring, * sizeof (long long), 1, "a) * * See zfs_set_userquota() to set a quota. */ if ((uint32_t)type >= MAXQUOTAS) { error = EINVAL; goto done; } switch (cmd) { case Q_GETQUOTASIZE: bitsize = 64; error = copyout(&bitsize, arg, sizeof (int)); break; case Q_QUOTAON: // As far as I can tell, you can't turn quotas on or off on zfs error = 0; vfs_unbusy(vfsp); break; case Q_QUOTAOFF: error = ENOTSUP; vfs_unbusy(vfsp); break; case Q_SETQUOTA: error = copyin(arg, &dqblk, sizeof (dqblk)); if (error == 0) error = zfs_set_userquota(zfsvfs, quota_type, "", id, dbtob(dqblk.dqb_bhardlimit)); break; case Q_GETQUOTA: error = zfs_getquota(zfsvfs, id, type == GRPQUOTA, &dqblk); if (error == 0) error = copyout(&dqblk, arg, sizeof (dqblk)); break; default: error = EINVAL; break; } done: ZFS_EXIT(zfsvfs); return (error); } boolean_t zfs_is_readonly(zfsvfs_t *zfsvfs) { return (!!(zfsvfs->z_vfs->vfs_flag & VFS_RDONLY)); } /*ARGSUSED*/ static int zfs_sync(vfs_t *vfsp, int waitfor) { /* * Data integrity is job one. We don't want a compromised kernel * writing to the storage pool, so we never sync during panic. */ if (panicstr) return (0); /* * Ignore the system syncher. ZFS already commits async data * at zfs_txg_timeout intervals. */ if (waitfor == MNT_LAZY) return (0); if (vfsp != NULL) { /* * Sync a specific filesystem. */ zfsvfs_t *zfsvfs = vfsp->vfs_data; dsl_pool_t *dp; int error; error = vfs_stdsync(vfsp, waitfor); if (error != 0) return (error); ZFS_ENTER(zfsvfs); dp = dmu_objset_pool(zfsvfs->z_os); /* * If the system is shutting down, then skip any * filesystems which may exist on a suspended pool. */ if (rebooting && spa_suspended(dp->dp_spa)) { ZFS_EXIT(zfsvfs); return (0); } if (zfsvfs->z_log != NULL) zil_commit(zfsvfs->z_log, 0); ZFS_EXIT(zfsvfs); } else { /* * Sync all ZFS filesystems. This is what happens when you - * run sync(1M). Unlike other filesystems, ZFS honors the + * run sync(8). Unlike other filesystems, ZFS honors the * request by waiting for all pools to commit all dirty data. */ spa_sync_allpools(); } return (0); } static void atime_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; if (newval == TRUE) { zfsvfs->z_atime = TRUE; zfsvfs->z_vfs->vfs_flag &= ~MNT_NOATIME; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_NOATIME); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_ATIME, NULL, 0); } else { zfsvfs->z_atime = FALSE; zfsvfs->z_vfs->vfs_flag |= MNT_NOATIME; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_ATIME); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_NOATIME, NULL, 0); } } static void xattr_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; if (newval == ZFS_XATTR_OFF) { zfsvfs->z_flags &= ~ZSB_XATTR; } else { zfsvfs->z_flags |= ZSB_XATTR; if (newval == ZFS_XATTR_SA) zfsvfs->z_xattr_sa = B_TRUE; else zfsvfs->z_xattr_sa = B_FALSE; } } static void blksz_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; ASSERT3U(newval, <=, spa_maxblocksize(dmu_objset_spa(zfsvfs->z_os))); ASSERT3U(newval, >=, SPA_MINBLOCKSIZE); ASSERT(ISP2(newval)); zfsvfs->z_max_blksz = newval; zfsvfs->z_vfs->mnt_stat.f_iosize = newval; } static void readonly_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; if (newval) { /* XXX locking on vfs_flag? */ zfsvfs->z_vfs->vfs_flag |= VFS_RDONLY; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_RW); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_RO, NULL, 0); } else { /* XXX locking on vfs_flag? */ zfsvfs->z_vfs->vfs_flag &= ~VFS_RDONLY; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_RO); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_RW, NULL, 0); } } static void setuid_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; if (newval == FALSE) { zfsvfs->z_vfs->vfs_flag |= VFS_NOSETUID; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_SETUID); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_NOSETUID, NULL, 0); } else { zfsvfs->z_vfs->vfs_flag &= ~VFS_NOSETUID; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_NOSETUID); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_SETUID, NULL, 0); } } static void exec_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; if (newval == FALSE) { zfsvfs->z_vfs->vfs_flag |= VFS_NOEXEC; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_EXEC); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_NOEXEC, NULL, 0); } else { zfsvfs->z_vfs->vfs_flag &= ~VFS_NOEXEC; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_NOEXEC); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_EXEC, NULL, 0); } } /* * The nbmand mount option can be changed at mount time. * We can't allow it to be toggled on live file systems or incorrect * behavior may be seen from cifs clients * * This property isn't registered via dsl_prop_register(), but this callback * will be called when a file system is first mounted */ static void nbmand_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; if (newval == FALSE) { vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_NBMAND); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_NONBMAND, NULL, 0); } else { vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_NONBMAND); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_NBMAND, NULL, 0); } } static void snapdir_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; zfsvfs->z_show_ctldir = newval; } static void vscan_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; zfsvfs->z_vscan = newval; } static void acl_mode_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; zfsvfs->z_acl_mode = newval; } static void acl_inherit_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; zfsvfs->z_acl_inherit = newval; } static void acl_type_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; zfsvfs->z_acl_type = newval; } static int zfs_register_callbacks(vfs_t *vfsp) { struct dsl_dataset *ds = NULL; objset_t *os = NULL; zfsvfs_t *zfsvfs = NULL; uint64_t nbmand; boolean_t readonly = B_FALSE; boolean_t do_readonly = B_FALSE; boolean_t setuid = B_FALSE; boolean_t do_setuid = B_FALSE; boolean_t exec = B_FALSE; boolean_t do_exec = B_FALSE; boolean_t xattr = B_FALSE; boolean_t atime = B_FALSE; boolean_t do_atime = B_FALSE; boolean_t do_xattr = B_FALSE; int error = 0; ASSERT(vfsp); zfsvfs = vfsp->vfs_data; ASSERT(zfsvfs); os = zfsvfs->z_os; /* * This function can be called for a snapshot when we update snapshot's * mount point, which isn't really supported. */ if (dmu_objset_is_snapshot(os)) return (EOPNOTSUPP); /* * The act of registering our callbacks will destroy any mount * options we may have. In order to enable temporary overrides * of mount options, we stash away the current values and * restore them after we register the callbacks. */ if (vfs_optionisset(vfsp, MNTOPT_RO, NULL) || !spa_writeable(dmu_objset_spa(os))) { readonly = B_TRUE; do_readonly = B_TRUE; } else if (vfs_optionisset(vfsp, MNTOPT_RW, NULL)) { readonly = B_FALSE; do_readonly = B_TRUE; } if (vfs_optionisset(vfsp, MNTOPT_NOSETUID, NULL)) { setuid = B_FALSE; do_setuid = B_TRUE; } else if (vfs_optionisset(vfsp, MNTOPT_SETUID, NULL)) { setuid = B_TRUE; do_setuid = B_TRUE; } if (vfs_optionisset(vfsp, MNTOPT_NOEXEC, NULL)) { exec = B_FALSE; do_exec = B_TRUE; } else if (vfs_optionisset(vfsp, MNTOPT_EXEC, NULL)) { exec = B_TRUE; do_exec = B_TRUE; } if (vfs_optionisset(vfsp, MNTOPT_NOXATTR, NULL)) { zfsvfs->z_xattr = xattr = ZFS_XATTR_OFF; do_xattr = B_TRUE; } else if (vfs_optionisset(vfsp, MNTOPT_XATTR, NULL)) { zfsvfs->z_xattr = xattr = ZFS_XATTR_DIR; do_xattr = B_TRUE; } else if (vfs_optionisset(vfsp, MNTOPT_DIRXATTR, NULL)) { zfsvfs->z_xattr = xattr = ZFS_XATTR_DIR; do_xattr = B_TRUE; } else if (vfs_optionisset(vfsp, MNTOPT_SAXATTR, NULL)) { zfsvfs->z_xattr = xattr = ZFS_XATTR_SA; do_xattr = B_TRUE; } if (vfs_optionisset(vfsp, MNTOPT_NOATIME, NULL)) { atime = B_FALSE; do_atime = B_TRUE; } else if (vfs_optionisset(vfsp, MNTOPT_ATIME, NULL)) { atime = B_TRUE; do_atime = B_TRUE; } /* * We need to enter pool configuration here, so that we can use * dsl_prop_get_int_ds() to handle the special nbmand property below. * dsl_prop_get_integer() can not be used, because it has to acquire * spa_namespace_lock and we can not do that because we already hold * z_teardown_lock. The problem is that spa_write_cachefile() is called * with spa_namespace_lock held and the function calls ZFS vnode * operations to write the cache file and thus z_teardown_lock is * acquired after spa_namespace_lock. */ ds = dmu_objset_ds(os); dsl_pool_config_enter(dmu_objset_pool(os), FTAG); /* * nbmand is a special property. It can only be changed at * mount time. * * This is weird, but it is documented to only be changeable * at mount time. */ if (vfs_optionisset(vfsp, MNTOPT_NONBMAND, NULL)) { nbmand = B_FALSE; } else if (vfs_optionisset(vfsp, MNTOPT_NBMAND, NULL)) { nbmand = B_TRUE; } else if ((error = dsl_prop_get_int_ds(ds, "nbmand", &nbmand) != 0)) { dsl_pool_config_exit(dmu_objset_pool(os), FTAG); return (error); } /* * Register property callbacks. * * It would probably be fine to just check for i/o error from * the first prop_register(), but I guess I like to go * overboard... */ error = dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_ATIME), atime_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_XATTR), xattr_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_RECORDSIZE), blksz_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_READONLY), readonly_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_SETUID), setuid_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_EXEC), exec_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_SNAPDIR), snapdir_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_ACLTYPE), acl_type_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_ACLMODE), acl_mode_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_ACLINHERIT), acl_inherit_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_VSCAN), vscan_changed_cb, zfsvfs); dsl_pool_config_exit(dmu_objset_pool(os), FTAG); if (error) goto unregister; /* * Invoke our callbacks to restore temporary mount options. */ if (do_readonly) readonly_changed_cb(zfsvfs, readonly); if (do_setuid) setuid_changed_cb(zfsvfs, setuid); if (do_exec) exec_changed_cb(zfsvfs, exec); if (do_xattr) xattr_changed_cb(zfsvfs, xattr); if (do_atime) atime_changed_cb(zfsvfs, atime); nbmand_changed_cb(zfsvfs, nbmand); return (0); unregister: dsl_prop_unregister_all(ds, zfsvfs); return (error); } /* * Associate this zfsvfs with the given objset, which must be owned. * This will cache a bunch of on-disk state from the objset in the * zfsvfs. */ static int zfsvfs_init(zfsvfs_t *zfsvfs, objset_t *os) { int error; uint64_t val; zfsvfs->z_max_blksz = SPA_OLD_MAXBLOCKSIZE; zfsvfs->z_show_ctldir = ZFS_SNAPDIR_VISIBLE; zfsvfs->z_os = os; error = zfs_get_zplprop(os, ZFS_PROP_VERSION, &zfsvfs->z_version); if (error != 0) return (error); if (zfsvfs->z_version > zfs_zpl_version_map(spa_version(dmu_objset_spa(os)))) { (void) printf("Can't mount a version %lld file system " "on a version %lld pool\n. Pool must be upgraded to mount " "this file system.", (u_longlong_t)zfsvfs->z_version, (u_longlong_t)spa_version(dmu_objset_spa(os))); return (SET_ERROR(ENOTSUP)); } error = zfs_get_zplprop(os, ZFS_PROP_NORMALIZE, &val); if (error != 0) return (error); zfsvfs->z_norm = (int)val; error = zfs_get_zplprop(os, ZFS_PROP_UTF8ONLY, &val); if (error != 0) return (error); zfsvfs->z_utf8 = (val != 0); error = zfs_get_zplprop(os, ZFS_PROP_CASE, &val); if (error != 0) return (error); zfsvfs->z_case = (uint_t)val; error = zfs_get_zplprop(os, ZFS_PROP_ACLTYPE, &val); if (error != 0) return (error); zfsvfs->z_acl_type = (uint_t)val; /* * Fold case on file systems that are always or sometimes case * insensitive. */ if (zfsvfs->z_case == ZFS_CASE_INSENSITIVE || zfsvfs->z_case == ZFS_CASE_MIXED) zfsvfs->z_norm |= U8_TEXTPREP_TOUPPER; zfsvfs->z_use_fuids = USE_FUIDS(zfsvfs->z_version, zfsvfs->z_os); zfsvfs->z_use_sa = USE_SA(zfsvfs->z_version, zfsvfs->z_os); uint64_t sa_obj = 0; if (zfsvfs->z_use_sa) { /* should either have both of these objects or none */ error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_SA_ATTRS, 8, 1, &sa_obj); if (error != 0) return (error); } error = sa_setup(os, sa_obj, zfs_attr_table, ZPL_END, &zfsvfs->z_attr_table); if (error != 0) return (error); if (zfsvfs->z_version >= ZPL_VERSION_SA) sa_register_update_callback(os, zfs_sa_upgrade); error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_ROOT_OBJ, 8, 1, &zfsvfs->z_root); if (error != 0) return (error); ASSERT(zfsvfs->z_root != 0); error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_UNLINKED_SET, 8, 1, &zfsvfs->z_unlinkedobj); if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_USERQUOTA], 8, 1, &zfsvfs->z_userquota_obj); if (error == ENOENT) zfsvfs->z_userquota_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_GROUPQUOTA], 8, 1, &zfsvfs->z_groupquota_obj); if (error == ENOENT) zfsvfs->z_groupquota_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_PROJECTQUOTA], 8, 1, &zfsvfs->z_projectquota_obj); if (error == ENOENT) zfsvfs->z_projectquota_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_USEROBJQUOTA], 8, 1, &zfsvfs->z_userobjquota_obj); if (error == ENOENT) zfsvfs->z_userobjquota_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_GROUPOBJQUOTA], 8, 1, &zfsvfs->z_groupobjquota_obj); if (error == ENOENT) zfsvfs->z_groupobjquota_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_PROJECTOBJQUOTA], 8, 1, &zfsvfs->z_projectobjquota_obj); if (error == ENOENT) zfsvfs->z_projectobjquota_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_FUID_TABLES, 8, 1, &zfsvfs->z_fuid_obj); if (error == ENOENT) zfsvfs->z_fuid_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_SHARES_DIR, 8, 1, &zfsvfs->z_shares_dir); if (error == ENOENT) zfsvfs->z_shares_dir = 0; else if (error != 0) return (error); /* * Only use the name cache if we are looking for a * name on a file system that does not require normalization * or case folding. We can also look there if we happen to be * on a non-normalizing, mixed sensitivity file system IF we * are looking for the exact name (which is always the case on * FreeBSD). */ zfsvfs->z_use_namecache = !zfsvfs->z_norm || ((zfsvfs->z_case == ZFS_CASE_MIXED) && !(zfsvfs->z_norm & ~U8_TEXTPREP_TOUPPER)); return (0); } taskq_t *zfsvfs_taskq; static void zfsvfs_task_unlinked_drain(void *context, int pending __unused) { zfs_unlinked_drain((zfsvfs_t *)context); } int zfsvfs_create(const char *osname, boolean_t readonly, zfsvfs_t **zfvp) { objset_t *os; zfsvfs_t *zfsvfs; int error; boolean_t ro = (readonly || (strchr(osname, '@') != NULL)); /* * XXX: Fix struct statfs so this isn't necessary! * * The 'osname' is used as the filesystem's special node, which means * it must fit in statfs.f_mntfromname, or else it can't be * enumerated, so libzfs_mnttab_find() returns NULL, which causes * 'zfs unmount' to think it's not mounted when it is. */ if (strlen(osname) >= MNAMELEN) return (SET_ERROR(ENAMETOOLONG)); zfsvfs = kmem_zalloc(sizeof (zfsvfs_t), KM_SLEEP); error = dmu_objset_own(osname, DMU_OST_ZFS, ro, B_TRUE, zfsvfs, &os); if (error != 0) { kmem_free(zfsvfs, sizeof (zfsvfs_t)); return (error); } error = zfsvfs_create_impl(zfvp, zfsvfs, os); return (error); } int zfsvfs_create_impl(zfsvfs_t **zfvp, zfsvfs_t *zfsvfs, objset_t *os) { int error; zfsvfs->z_vfs = NULL; zfsvfs->z_parent = zfsvfs; mutex_init(&zfsvfs->z_znodes_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&zfsvfs->z_lock, NULL, MUTEX_DEFAULT, NULL); list_create(&zfsvfs->z_all_znodes, sizeof (znode_t), offsetof(znode_t, z_link_node)); TASK_INIT(&zfsvfs->z_unlinked_drain_task, 0, zfsvfs_task_unlinked_drain, zfsvfs); #ifdef DIAGNOSTIC rrm_init(&zfsvfs->z_teardown_lock, B_TRUE); #else rrm_init(&zfsvfs->z_teardown_lock, B_FALSE); #endif ZFS_INIT_TEARDOWN_INACTIVE(zfsvfs); rw_init(&zfsvfs->z_fuid_lock, NULL, RW_DEFAULT, NULL); for (int i = 0; i != ZFS_OBJ_MTX_SZ; i++) mutex_init(&zfsvfs->z_hold_mtx[i], NULL, MUTEX_DEFAULT, NULL); error = zfsvfs_init(zfsvfs, os); if (error != 0) { dmu_objset_disown(os, B_TRUE, zfsvfs); *zfvp = NULL; kmem_free(zfsvfs, sizeof (zfsvfs_t)); return (error); } *zfvp = zfsvfs; return (0); } static int zfsvfs_setup(zfsvfs_t *zfsvfs, boolean_t mounting) { int error; /* * Check for a bad on-disk format version now since we * lied about owning the dataset readonly before. */ if (!(zfsvfs->z_vfs->vfs_flag & VFS_RDONLY) && dmu_objset_incompatible_encryption_version(zfsvfs->z_os)) return (SET_ERROR(EROFS)); error = zfs_register_callbacks(zfsvfs->z_vfs); if (error) return (error); zfsvfs->z_log = zil_open(zfsvfs->z_os, zfs_get_data); /* * If we are not mounting (ie: online recv), then we don't * have to worry about replaying the log as we blocked all * operations out since we closed the ZIL. */ if (mounting) { boolean_t readonly; ASSERT3P(zfsvfs->z_kstat.dk_kstats, ==, NULL); dataset_kstats_create(&zfsvfs->z_kstat, zfsvfs->z_os); /* * During replay we remove the read only flag to * allow replays to succeed. */ readonly = zfsvfs->z_vfs->vfs_flag & VFS_RDONLY; if (readonly != 0) { zfsvfs->z_vfs->vfs_flag &= ~VFS_RDONLY; } else { dsl_dir_t *dd; zap_stats_t zs; if (zap_get_stats(zfsvfs->z_os, zfsvfs->z_unlinkedobj, &zs) == 0) { dataset_kstats_update_nunlinks_kstat( &zfsvfs->z_kstat, zs.zs_num_entries); dprintf_ds(zfsvfs->z_os->os_dsl_dataset, "num_entries in unlinked set: %llu", zs.zs_num_entries); } zfs_unlinked_drain(zfsvfs); dd = zfsvfs->z_os->os_dsl_dataset->ds_dir; dd->dd_activity_cancelled = B_FALSE; } /* * Parse and replay the intent log. * * Because of ziltest, this must be done after * zfs_unlinked_drain(). (Further note: ziltest * doesn't use readonly mounts, where * zfs_unlinked_drain() isn't called.) This is because * ziltest causes spa_sync() to think it's committed, * but actually it is not, so the intent log contains * many txg's worth of changes. * * In particular, if object N is in the unlinked set in * the last txg to actually sync, then it could be * actually freed in a later txg and then reallocated * in a yet later txg. This would write a "create * object N" record to the intent log. Normally, this * would be fine because the spa_sync() would have * written out the fact that object N is free, before * we could write the "create object N" intent log * record. * * But when we are in ziltest mode, we advance the "open * txg" without actually spa_sync()-ing the changes to * disk. So we would see that object N is still * allocated and in the unlinked set, and there is an * intent log record saying to allocate it. */ if (spa_writeable(dmu_objset_spa(zfsvfs->z_os))) { if (zil_replay_disable) { zil_destroy(zfsvfs->z_log, B_FALSE); } else { boolean_t use_nc = zfsvfs->z_use_namecache; zfsvfs->z_use_namecache = B_FALSE; zfsvfs->z_replay = B_TRUE; zil_replay(zfsvfs->z_os, zfsvfs, zfs_replay_vector); zfsvfs->z_replay = B_FALSE; zfsvfs->z_use_namecache = use_nc; } } /* restore readonly bit */ if (readonly != 0) zfsvfs->z_vfs->vfs_flag |= VFS_RDONLY; } /* * Set the objset user_ptr to track its zfsvfs. */ mutex_enter(&zfsvfs->z_os->os_user_ptr_lock); dmu_objset_set_user(zfsvfs->z_os, zfsvfs); mutex_exit(&zfsvfs->z_os->os_user_ptr_lock); return (0); } void zfsvfs_free(zfsvfs_t *zfsvfs) { int i; zfs_fuid_destroy(zfsvfs); mutex_destroy(&zfsvfs->z_znodes_lock); mutex_destroy(&zfsvfs->z_lock); ASSERT(zfsvfs->z_nr_znodes == 0); list_destroy(&zfsvfs->z_all_znodes); rrm_destroy(&zfsvfs->z_teardown_lock); ZFS_DESTROY_TEARDOWN_INACTIVE(zfsvfs); rw_destroy(&zfsvfs->z_fuid_lock); for (i = 0; i != ZFS_OBJ_MTX_SZ; i++) mutex_destroy(&zfsvfs->z_hold_mtx[i]); dataset_kstats_destroy(&zfsvfs->z_kstat); kmem_free(zfsvfs, sizeof (zfsvfs_t)); } static void zfs_set_fuid_feature(zfsvfs_t *zfsvfs) { zfsvfs->z_use_fuids = USE_FUIDS(zfsvfs->z_version, zfsvfs->z_os); if (zfsvfs->z_vfs) { if (zfsvfs->z_use_fuids) { vfs_set_feature(zfsvfs->z_vfs, VFSFT_XVATTR); vfs_set_feature(zfsvfs->z_vfs, VFSFT_SYSATTR_VIEWS); vfs_set_feature(zfsvfs->z_vfs, VFSFT_ACEMASKONACCESS); vfs_set_feature(zfsvfs->z_vfs, VFSFT_ACLONCREATE); vfs_set_feature(zfsvfs->z_vfs, VFSFT_ACCESS_FILTER); vfs_set_feature(zfsvfs->z_vfs, VFSFT_REPARSE); } else { vfs_clear_feature(zfsvfs->z_vfs, VFSFT_XVATTR); vfs_clear_feature(zfsvfs->z_vfs, VFSFT_SYSATTR_VIEWS); vfs_clear_feature(zfsvfs->z_vfs, VFSFT_ACEMASKONACCESS); vfs_clear_feature(zfsvfs->z_vfs, VFSFT_ACLONCREATE); vfs_clear_feature(zfsvfs->z_vfs, VFSFT_ACCESS_FILTER); vfs_clear_feature(zfsvfs->z_vfs, VFSFT_REPARSE); } } zfsvfs->z_use_sa = USE_SA(zfsvfs->z_version, zfsvfs->z_os); } static int zfs_domount(vfs_t *vfsp, char *osname) { uint64_t recordsize, fsid_guid; int error = 0; zfsvfs_t *zfsvfs; ASSERT(vfsp); ASSERT(osname); error = zfsvfs_create(osname, vfsp->mnt_flag & MNT_RDONLY, &zfsvfs); if (error) return (error); zfsvfs->z_vfs = vfsp; if ((error = dsl_prop_get_integer(osname, "recordsize", &recordsize, NULL))) goto out; zfsvfs->z_vfs->vfs_bsize = SPA_MINBLOCKSIZE; zfsvfs->z_vfs->mnt_stat.f_iosize = recordsize; vfsp->vfs_data = zfsvfs; vfsp->mnt_flag |= MNT_LOCAL; vfsp->mnt_kern_flag |= MNTK_LOOKUP_SHARED; vfsp->mnt_kern_flag |= MNTK_SHARED_WRITES; vfsp->mnt_kern_flag |= MNTK_EXTENDED_SHARED; /* * This can cause a loss of coherence between ARC and page cache * on ZoF - unclear if the problem is in FreeBSD or ZoF */ vfsp->mnt_kern_flag |= MNTK_NO_IOPF; /* vn_io_fault can be used */ vfsp->mnt_kern_flag |= MNTK_NOMSYNC; vfsp->mnt_kern_flag |= MNTK_VMSETSIZE_BUG; #if defined(_KERNEL) && !defined(KMEM_DEBUG) vfsp->mnt_kern_flag |= MNTK_FPLOOKUP; #endif /* * The fsid is 64 bits, composed of an 8-bit fs type, which * separates our fsid from any other filesystem types, and a * 56-bit objset unique ID. The objset unique ID is unique to * all objsets open on this system, provided by unique_create(). * The 8-bit fs type must be put in the low bits of fsid[1] * because that's where other Solaris filesystems put it. */ fsid_guid = dmu_objset_fsid_guid(zfsvfs->z_os); ASSERT((fsid_guid & ~((1ULL<<56)-1)) == 0); vfsp->vfs_fsid.val[0] = fsid_guid; vfsp->vfs_fsid.val[1] = ((fsid_guid>>32) << 8) | (vfsp->mnt_vfc->vfc_typenum & 0xFF); /* * Set features for file system. */ zfs_set_fuid_feature(zfsvfs); if (zfsvfs->z_case == ZFS_CASE_INSENSITIVE) { vfs_set_feature(vfsp, VFSFT_DIRENTFLAGS); vfs_set_feature(vfsp, VFSFT_CASEINSENSITIVE); vfs_set_feature(vfsp, VFSFT_NOCASESENSITIVE); } else if (zfsvfs->z_case == ZFS_CASE_MIXED) { vfs_set_feature(vfsp, VFSFT_DIRENTFLAGS); vfs_set_feature(vfsp, VFSFT_CASEINSENSITIVE); } vfs_set_feature(vfsp, VFSFT_ZEROCOPY_SUPPORTED); if (dmu_objset_is_snapshot(zfsvfs->z_os)) { uint64_t pval; atime_changed_cb(zfsvfs, B_FALSE); readonly_changed_cb(zfsvfs, B_TRUE); if ((error = dsl_prop_get_integer(osname, "xattr", &pval, NULL))) goto out; xattr_changed_cb(zfsvfs, pval); if ((error = dsl_prop_get_integer(osname, "acltype", &pval, NULL))) goto out; acl_type_changed_cb(zfsvfs, pval); zfsvfs->z_issnap = B_TRUE; zfsvfs->z_os->os_sync = ZFS_SYNC_DISABLED; mutex_enter(&zfsvfs->z_os->os_user_ptr_lock); dmu_objset_set_user(zfsvfs->z_os, zfsvfs); mutex_exit(&zfsvfs->z_os->os_user_ptr_lock); } else { if ((error = zfsvfs_setup(zfsvfs, B_TRUE))) goto out; } vfs_mountedfrom(vfsp, osname); if (!zfsvfs->z_issnap) zfsctl_create(zfsvfs); out: if (error) { dmu_objset_disown(zfsvfs->z_os, B_TRUE, zfsvfs); zfsvfs_free(zfsvfs); } else { atomic_inc_32(&zfs_active_fs_count); } return (error); } static void zfs_unregister_callbacks(zfsvfs_t *zfsvfs) { objset_t *os = zfsvfs->z_os; if (!dmu_objset_is_snapshot(os)) dsl_prop_unregister_all(dmu_objset_ds(os), zfsvfs); } static int getpoolname(const char *osname, char *poolname) { char *p; p = strchr(osname, '/'); if (p == NULL) { if (strlen(osname) >= MAXNAMELEN) return (ENAMETOOLONG); (void) strcpy(poolname, osname); } else { if (p - osname >= MAXNAMELEN) return (ENAMETOOLONG); (void) strncpy(poolname, osname, p - osname); poolname[p - osname] = '\0'; } return (0); } /*ARGSUSED*/ static int zfs_mount(vfs_t *vfsp) { kthread_t *td = curthread; vnode_t *mvp = vfsp->mnt_vnodecovered; cred_t *cr = td->td_ucred; char *osname; int error = 0; int canwrite; if (vfs_getopt(vfsp->mnt_optnew, "from", (void **)&osname, NULL)) return (SET_ERROR(EINVAL)); /* * If full-owner-access is enabled and delegated administration is * turned on, we must set nosuid. */ if (zfs_super_owner && dsl_deleg_access(osname, ZFS_DELEG_PERM_MOUNT, cr) != ECANCELED) { secpolicy_fs_mount_clearopts(cr, vfsp); } /* * Check for mount privilege? * * If we don't have privilege then see if * we have local permission to allow it */ error = secpolicy_fs_mount(cr, mvp, vfsp); if (error) { if (dsl_deleg_access(osname, ZFS_DELEG_PERM_MOUNT, cr) != 0) goto out; if (!(vfsp->vfs_flag & MS_REMOUNT)) { vattr_t vattr; /* * Make sure user is the owner of the mount point * or has sufficient privileges. */ vattr.va_mask = AT_UID; vn_lock(mvp, LK_SHARED | LK_RETRY); if (VOP_GETATTR(mvp, &vattr, cr)) { VOP_UNLOCK1(mvp); goto out; } if (secpolicy_vnode_owner(mvp, cr, vattr.va_uid) != 0 && VOP_ACCESS(mvp, VWRITE, cr, td) != 0) { VOP_UNLOCK1(mvp); goto out; } VOP_UNLOCK1(mvp); } secpolicy_fs_mount_clearopts(cr, vfsp); } /* * Refuse to mount a filesystem if we are in a local zone and the * dataset is not visible. */ if (!INGLOBALZONE(curproc) && (!zone_dataset_visible(osname, &canwrite) || !canwrite)) { error = SET_ERROR(EPERM); goto out; } vfsp->vfs_flag |= MNT_NFS4ACLS; /* * When doing a remount, we simply refresh our temporary properties * according to those options set in the current VFS options. */ if (vfsp->vfs_flag & MS_REMOUNT) { zfsvfs_t *zfsvfs = vfsp->vfs_data; /* * Refresh mount options with z_teardown_lock blocking I/O while * the filesystem is in an inconsistent state. * The lock also serializes this code with filesystem * manipulations between entry to zfs_suspend_fs() and return * from zfs_resume_fs(). */ rrm_enter(&zfsvfs->z_teardown_lock, RW_WRITER, FTAG); zfs_unregister_callbacks(zfsvfs); error = zfs_register_callbacks(vfsp); rrm_exit(&zfsvfs->z_teardown_lock, FTAG); goto out; } /* Initial root mount: try hard to import the requested root pool. */ if ((vfsp->vfs_flag & MNT_ROOTFS) != 0 && (vfsp->vfs_flag & MNT_UPDATE) == 0) { char pname[MAXNAMELEN]; error = getpoolname(osname, pname); if (error == 0) error = spa_import_rootpool(pname, false); if (error) goto out; } DROP_GIANT(); error = zfs_domount(vfsp, osname); PICKUP_GIANT(); out: return (error); } static int zfs_statfs(vfs_t *vfsp, struct statfs *statp) { zfsvfs_t *zfsvfs = vfsp->vfs_data; uint64_t refdbytes, availbytes, usedobjs, availobjs; statp->f_version = STATFS_VERSION; ZFS_ENTER(zfsvfs); dmu_objset_space(zfsvfs->z_os, &refdbytes, &availbytes, &usedobjs, &availobjs); /* * The underlying storage pool actually uses multiple block sizes. * We report the fragsize as the smallest block size we support, * and we report our blocksize as the filesystem's maximum blocksize. */ statp->f_bsize = SPA_MINBLOCKSIZE; statp->f_iosize = zfsvfs->z_vfs->mnt_stat.f_iosize; /* * The following report "total" blocks of various kinds in the * file system, but reported in terms of f_frsize - the * "fragment" size. */ statp->f_blocks = (refdbytes + availbytes) >> SPA_MINBLOCKSHIFT; statp->f_bfree = availbytes / statp->f_bsize; statp->f_bavail = statp->f_bfree; /* no root reservation */ /* * statvfs() should really be called statufs(), because it assumes * static metadata. ZFS doesn't preallocate files, so the best * we can do is report the max that could possibly fit in f_files, * and that minus the number actually used in f_ffree. * For f_ffree, report the smaller of the number of object available * and the number of blocks (each object will take at least a block). */ statp->f_ffree = MIN(availobjs, statp->f_bfree); statp->f_files = statp->f_ffree + usedobjs; /* * We're a zfs filesystem. */ strlcpy(statp->f_fstypename, "zfs", sizeof (statp->f_fstypename)); strlcpy(statp->f_mntfromname, vfsp->mnt_stat.f_mntfromname, sizeof (statp->f_mntfromname)); strlcpy(statp->f_mntonname, vfsp->mnt_stat.f_mntonname, sizeof (statp->f_mntonname)); statp->f_namemax = MAXNAMELEN - 1; ZFS_EXIT(zfsvfs); return (0); } static int zfs_root(vfs_t *vfsp, int flags, vnode_t **vpp) { zfsvfs_t *zfsvfs = vfsp->vfs_data; znode_t *rootzp; int error; ZFS_ENTER(zfsvfs); error = zfs_zget(zfsvfs, zfsvfs->z_root, &rootzp); if (error == 0) *vpp = ZTOV(rootzp); ZFS_EXIT(zfsvfs); if (error == 0) { error = vn_lock(*vpp, flags); if (error != 0) { VN_RELE(*vpp); *vpp = NULL; } } return (error); } /* * Teardown the zfsvfs::z_os. * * Note, if 'unmounting' is FALSE, we return with the 'z_teardown_lock' * and 'z_teardown_inactive_lock' held. */ static int zfsvfs_teardown(zfsvfs_t *zfsvfs, boolean_t unmounting) { znode_t *zp; dsl_dir_t *dd; /* * If someone has not already unmounted this file system, * drain the zrele_taskq to ensure all active references to the * zfsvfs_t have been handled only then can it be safely destroyed. */ if (zfsvfs->z_os) { /* * If we're unmounting we have to wait for the list to * drain completely. * * If we're not unmounting there's no guarantee the list * will drain completely, but zreles run from the taskq * may add the parents of dir-based xattrs to the taskq * so we want to wait for these. * * We can safely read z_nr_znodes without locking because the * VFS has already blocked operations which add to the * z_all_znodes list and thus increment z_nr_znodes. */ int round = 0; while (zfsvfs->z_nr_znodes > 0) { taskq_wait_outstanding(dsl_pool_zrele_taskq( dmu_objset_pool(zfsvfs->z_os)), 0); if (++round > 1 && !unmounting) break; } } rrm_enter(&zfsvfs->z_teardown_lock, RW_WRITER, FTAG); if (!unmounting) { /* * We purge the parent filesystem's vfsp as the parent * filesystem and all of its snapshots have their vnode's * v_vfsp set to the parent's filesystem's vfsp. Note, * 'z_parent' is self referential for non-snapshots. */ #ifdef FREEBSD_NAMECACHE #if __FreeBSD_version >= 1300117 cache_purgevfs(zfsvfs->z_parent->z_vfs); #else cache_purgevfs(zfsvfs->z_parent->z_vfs, true); #endif #endif } /* * Close the zil. NB: Can't close the zil while zfs_inactive * threads are blocked as zil_close can call zfs_inactive. */ if (zfsvfs->z_log) { zil_close(zfsvfs->z_log); zfsvfs->z_log = NULL; } ZFS_WLOCK_TEARDOWN_INACTIVE(zfsvfs); /* * If we are not unmounting (ie: online recv) and someone already * unmounted this file system while we were doing the switcheroo, * or a reopen of z_os failed then just bail out now. */ if (!unmounting && (zfsvfs->z_unmounted || zfsvfs->z_os == NULL)) { ZFS_WUNLOCK_TEARDOWN_INACTIVE(zfsvfs); rrm_exit(&zfsvfs->z_teardown_lock, FTAG); return (SET_ERROR(EIO)); } /* * At this point there are no vops active, and any new vops will * fail with EIO since we have z_teardown_lock for writer (only * relevant for forced unmount). * * Release all holds on dbufs. */ mutex_enter(&zfsvfs->z_znodes_lock); for (zp = list_head(&zfsvfs->z_all_znodes); zp != NULL; zp = list_next(&zfsvfs->z_all_znodes, zp)) if (zp->z_sa_hdl) { ASSERT(ZTOV(zp)->v_count >= 0); zfs_znode_dmu_fini(zp); } mutex_exit(&zfsvfs->z_znodes_lock); /* * If we are unmounting, set the unmounted flag and let new vops * unblock. zfs_inactive will have the unmounted behavior, and all * other vops will fail with EIO. */ if (unmounting) { zfsvfs->z_unmounted = B_TRUE; ZFS_WUNLOCK_TEARDOWN_INACTIVE(zfsvfs); rrm_exit(&zfsvfs->z_teardown_lock, FTAG); } /* * z_os will be NULL if there was an error in attempting to reopen * zfsvfs, so just return as the properties had already been * unregistered and cached data had been evicted before. */ if (zfsvfs->z_os == NULL) return (0); /* * Unregister properties. */ zfs_unregister_callbacks(zfsvfs); /* * Evict cached data */ if (!zfs_is_readonly(zfsvfs)) txg_wait_synced(dmu_objset_pool(zfsvfs->z_os), 0); dmu_objset_evict_dbufs(zfsvfs->z_os); dd = zfsvfs->z_os->os_dsl_dataset->ds_dir; dsl_dir_cancel_waiters(dd); return (0); } /*ARGSUSED*/ static int zfs_umount(vfs_t *vfsp, int fflag) { kthread_t *td = curthread; zfsvfs_t *zfsvfs = vfsp->vfs_data; objset_t *os; cred_t *cr = td->td_ucred; int ret; ret = secpolicy_fs_unmount(cr, vfsp); if (ret) { if (dsl_deleg_access((char *)vfsp->vfs_resource, ZFS_DELEG_PERM_MOUNT, cr)) return (ret); } /* * Unmount any snapshots mounted under .zfs before unmounting the * dataset itself. */ if (zfsvfs->z_ctldir != NULL) { if ((ret = zfsctl_umount_snapshots(vfsp, fflag, cr)) != 0) return (ret); } if (fflag & MS_FORCE) { /* * Mark file system as unmounted before calling * vflush(FORCECLOSE). This way we ensure no future vnops * will be called and risk operating on DOOMED vnodes. */ rrm_enter(&zfsvfs->z_teardown_lock, RW_WRITER, FTAG); zfsvfs->z_unmounted = B_TRUE; rrm_exit(&zfsvfs->z_teardown_lock, FTAG); } /* * Flush all the files. */ ret = vflush(vfsp, 0, (fflag & MS_FORCE) ? FORCECLOSE : 0, td); if (ret != 0) return (ret); while (taskqueue_cancel(zfsvfs_taskq->tq_queue, &zfsvfs->z_unlinked_drain_task, NULL) != 0) taskqueue_drain(zfsvfs_taskq->tq_queue, &zfsvfs->z_unlinked_drain_task); VERIFY(zfsvfs_teardown(zfsvfs, B_TRUE) == 0); os = zfsvfs->z_os; /* * z_os will be NULL if there was an error in * attempting to reopen zfsvfs. */ if (os != NULL) { /* * Unset the objset user_ptr. */ mutex_enter(&os->os_user_ptr_lock); dmu_objset_set_user(os, NULL); mutex_exit(&os->os_user_ptr_lock); /* * Finally release the objset */ dmu_objset_disown(os, B_TRUE, zfsvfs); } /* * We can now safely destroy the '.zfs' directory node. */ if (zfsvfs->z_ctldir != NULL) zfsctl_destroy(zfsvfs); zfs_freevfs(vfsp); return (0); } static int zfs_vget(vfs_t *vfsp, ino_t ino, int flags, vnode_t **vpp) { zfsvfs_t *zfsvfs = vfsp->vfs_data; znode_t *zp; int err; /* * zfs_zget() can't operate on virtual entries like .zfs/ or * .zfs/snapshot/ directories, that's why we return EOPNOTSUPP. * This will make NFS to switch to LOOKUP instead of using VGET. */ if (ino == ZFSCTL_INO_ROOT || ino == ZFSCTL_INO_SNAPDIR || (zfsvfs->z_shares_dir != 0 && ino == zfsvfs->z_shares_dir)) return (EOPNOTSUPP); ZFS_ENTER(zfsvfs); err = zfs_zget(zfsvfs, ino, &zp); if (err == 0 && zp->z_unlinked) { vrele(ZTOV(zp)); err = EINVAL; } if (err == 0) *vpp = ZTOV(zp); ZFS_EXIT(zfsvfs); if (err == 0) { err = vn_lock(*vpp, flags); if (err != 0) vrele(*vpp); } if (err != 0) *vpp = NULL; return (err); } static int #if __FreeBSD_version >= 1300098 zfs_checkexp(vfs_t *vfsp, struct sockaddr *nam, uint64_t *extflagsp, struct ucred **credanonp, int *numsecflavors, int *secflavors) #else zfs_checkexp(vfs_t *vfsp, struct sockaddr *nam, int *extflagsp, struct ucred **credanonp, int *numsecflavors, int **secflavors) #endif { zfsvfs_t *zfsvfs = vfsp->vfs_data; /* * If this is regular file system vfsp is the same as * zfsvfs->z_parent->z_vfs, but if it is snapshot, * zfsvfs->z_parent->z_vfs represents parent file system * which we have to use here, because only this file system * has mnt_export configured. */ return (vfs_stdcheckexp(zfsvfs->z_parent->z_vfs, nam, extflagsp, credanonp, numsecflavors, secflavors)); } CTASSERT(SHORT_FID_LEN <= sizeof (struct fid)); CTASSERT(LONG_FID_LEN <= sizeof (struct fid)); static int zfs_fhtovp(vfs_t *vfsp, fid_t *fidp, int flags, vnode_t **vpp) { struct componentname cn; zfsvfs_t *zfsvfs = vfsp->vfs_data; znode_t *zp; vnode_t *dvp; uint64_t object = 0; uint64_t fid_gen = 0; uint64_t gen_mask; uint64_t zp_gen; int i, err; *vpp = NULL; ZFS_ENTER(zfsvfs); /* * On FreeBSD we can get snapshot's mount point or its parent file * system mount point depending if snapshot is already mounted or not. */ if (zfsvfs->z_parent == zfsvfs && fidp->fid_len == LONG_FID_LEN) { zfid_long_t *zlfid = (zfid_long_t *)fidp; uint64_t objsetid = 0; uint64_t setgen = 0; for (i = 0; i < sizeof (zlfid->zf_setid); i++) objsetid |= ((uint64_t)zlfid->zf_setid[i]) << (8 * i); for (i = 0; i < sizeof (zlfid->zf_setgen); i++) setgen |= ((uint64_t)zlfid->zf_setgen[i]) << (8 * i); ZFS_EXIT(zfsvfs); err = zfsctl_lookup_objset(vfsp, objsetid, &zfsvfs); if (err) return (SET_ERROR(EINVAL)); ZFS_ENTER(zfsvfs); } if (fidp->fid_len == SHORT_FID_LEN || fidp->fid_len == LONG_FID_LEN) { zfid_short_t *zfid = (zfid_short_t *)fidp; for (i = 0; i < sizeof (zfid->zf_object); i++) object |= ((uint64_t)zfid->zf_object[i]) << (8 * i); for (i = 0; i < sizeof (zfid->zf_gen); i++) fid_gen |= ((uint64_t)zfid->zf_gen[i]) << (8 * i); } else { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } /* * A zero fid_gen means we are in .zfs or the .zfs/snapshot * directory tree. If the object == zfsvfs->z_shares_dir, then * we are in the .zfs/shares directory tree. */ if ((fid_gen == 0 && (object == ZFSCTL_INO_ROOT || object == ZFSCTL_INO_SNAPDIR)) || (zfsvfs->z_shares_dir != 0 && object == zfsvfs->z_shares_dir)) { ZFS_EXIT(zfsvfs); VERIFY0(zfsctl_root(zfsvfs, LK_SHARED, &dvp)); if (object == ZFSCTL_INO_SNAPDIR) { cn.cn_nameptr = "snapshot"; cn.cn_namelen = strlen(cn.cn_nameptr); cn.cn_nameiop = LOOKUP; cn.cn_flags = ISLASTCN | LOCKLEAF; cn.cn_lkflags = flags; VERIFY0(VOP_LOOKUP(dvp, vpp, &cn)); vput(dvp); } else if (object == zfsvfs->z_shares_dir) { /* * XXX This branch must not be taken, * if it is, then the lookup below will * explode. */ cn.cn_nameptr = "shares"; cn.cn_namelen = strlen(cn.cn_nameptr); cn.cn_nameiop = LOOKUP; cn.cn_flags = ISLASTCN; cn.cn_lkflags = flags; VERIFY0(VOP_LOOKUP(dvp, vpp, &cn)); vput(dvp); } else { *vpp = dvp; } return (err); } gen_mask = -1ULL >> (64 - 8 * i); dprintf("getting %llu [%u mask %llx]\n", object, fid_gen, gen_mask); if ((err = zfs_zget(zfsvfs, object, &zp))) { ZFS_EXIT(zfsvfs); return (err); } (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zfsvfs), &zp_gen, sizeof (uint64_t)); zp_gen = zp_gen & gen_mask; if (zp_gen == 0) zp_gen = 1; if (zp->z_unlinked || zp_gen != fid_gen) { dprintf("znode gen (%u) != fid gen (%u)\n", zp_gen, fid_gen); vrele(ZTOV(zp)); ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } *vpp = ZTOV(zp); ZFS_EXIT(zfsvfs); err = vn_lock(*vpp, flags); if (err == 0) vnode_create_vobject(*vpp, zp->z_size, curthread); else *vpp = NULL; return (err); } /* * Block out VOPs and close zfsvfs_t::z_os * * Note, if successful, then we return with the 'z_teardown_lock' and * 'z_teardown_inactive_lock' write held. We leave ownership of the underlying * dataset and objset intact so that they can be atomically handed off during * a subsequent rollback or recv operation and the resume thereafter. */ int zfs_suspend_fs(zfsvfs_t *zfsvfs) { int error; if ((error = zfsvfs_teardown(zfsvfs, B_FALSE)) != 0) return (error); return (0); } /* * Rebuild SA and release VOPs. Note that ownership of the underlying dataset * is an invariant across any of the operations that can be performed while the * filesystem was suspended. Whether it succeeded or failed, the preconditions * are the same: the relevant objset and associated dataset are owned by * zfsvfs, held, and long held on entry. */ int zfs_resume_fs(zfsvfs_t *zfsvfs, dsl_dataset_t *ds) { int err; znode_t *zp; ASSERT(RRM_WRITE_HELD(&zfsvfs->z_teardown_lock)); ASSERT(ZFS_TEARDOWN_INACTIVE_WLOCKED(zfsvfs)); /* * We already own this, so just update the objset_t, as the one we * had before may have been evicted. */ objset_t *os; VERIFY3P(ds->ds_owner, ==, zfsvfs); VERIFY(dsl_dataset_long_held(ds)); dsl_pool_t *dp = spa_get_dsl(dsl_dataset_get_spa(ds)); dsl_pool_config_enter(dp, FTAG); VERIFY0(dmu_objset_from_ds(ds, &os)); dsl_pool_config_exit(dp, FTAG); err = zfsvfs_init(zfsvfs, os); if (err != 0) goto bail; ds->ds_dir->dd_activity_cancelled = B_FALSE; VERIFY(zfsvfs_setup(zfsvfs, B_FALSE) == 0); zfs_set_fuid_feature(zfsvfs); /* * Attempt to re-establish all the active znodes with * their dbufs. If a zfs_rezget() fails, then we'll let * any potential callers discover that via ZFS_ENTER_VERIFY_VP * when they try to use their znode. */ mutex_enter(&zfsvfs->z_znodes_lock); for (zp = list_head(&zfsvfs->z_all_znodes); zp; zp = list_next(&zfsvfs->z_all_znodes, zp)) { (void) zfs_rezget(zp); } mutex_exit(&zfsvfs->z_znodes_lock); bail: /* release the VOPs */ ZFS_WUNLOCK_TEARDOWN_INACTIVE(zfsvfs); rrm_exit(&zfsvfs->z_teardown_lock, FTAG); if (err) { /* * Since we couldn't setup the sa framework, try to force * unmount this file system. */ if (vn_vfswlock(zfsvfs->z_vfs->vfs_vnodecovered) == 0) { vfs_ref(zfsvfs->z_vfs); (void) dounmount(zfsvfs->z_vfs, MS_FORCE, curthread); } } return (err); } static void zfs_freevfs(vfs_t *vfsp) { zfsvfs_t *zfsvfs = vfsp->vfs_data; zfsvfs_free(zfsvfs); atomic_dec_32(&zfs_active_fs_count); } #ifdef __i386__ static int desiredvnodes_backup; #include #include #include #include #include #endif static void zfs_vnodes_adjust(void) { #ifdef __i386__ int newdesiredvnodes; desiredvnodes_backup = desiredvnodes; /* * We calculate newdesiredvnodes the same way it is done in * vntblinit(). If it is equal to desiredvnodes, it means that * it wasn't tuned by the administrator and we can tune it down. */ newdesiredvnodes = min(maxproc + vm_cnt.v_page_count / 4, 2 * vm_kmem_size / (5 * (sizeof (struct vm_object) + sizeof (struct vnode)))); if (newdesiredvnodes == desiredvnodes) desiredvnodes = (3 * newdesiredvnodes) / 4; #endif } static void zfs_vnodes_adjust_back(void) { #ifdef __i386__ desiredvnodes = desiredvnodes_backup; #endif } void zfs_init(void) { printf("ZFS filesystem version: " ZPL_VERSION_STRING "\n"); /* * Initialize .zfs directory structures */ zfsctl_init(); /* * Initialize znode cache, vnode ops, etc... */ zfs_znode_init(); /* * Reduce number of vnodes. Originally number of vnodes is calculated * with UFS inode in mind. We reduce it here, because it's too big for * ZFS/i386. */ zfs_vnodes_adjust(); dmu_objset_register_type(DMU_OST_ZFS, zpl_get_file_info); zfsvfs_taskq = taskq_create("zfsvfs", 1, minclsyspri, 0, 0, 0); } void zfs_fini(void) { taskq_destroy(zfsvfs_taskq); zfsctl_fini(); zfs_znode_fini(); zfs_vnodes_adjust_back(); } int zfs_busy(void) { return (zfs_active_fs_count != 0); } /* * Release VOPs and unmount a suspended filesystem. */ int zfs_end_fs(zfsvfs_t *zfsvfs, dsl_dataset_t *ds) { ASSERT(RRM_WRITE_HELD(&zfsvfs->z_teardown_lock)); ASSERT(ZFS_TEARDOWN_INACTIVE_WLOCKED(zfsvfs)); /* * We already own this, so just hold and rele it to update the * objset_t, as the one we had before may have been evicted. */ objset_t *os; VERIFY3P(ds->ds_owner, ==, zfsvfs); VERIFY(dsl_dataset_long_held(ds)); dsl_pool_t *dp = spa_get_dsl(dsl_dataset_get_spa(ds)); dsl_pool_config_enter(dp, FTAG); VERIFY0(dmu_objset_from_ds(ds, &os)); dsl_pool_config_exit(dp, FTAG); zfsvfs->z_os = os; /* release the VOPs */ ZFS_WUNLOCK_TEARDOWN_INACTIVE(zfsvfs); rrm_exit(&zfsvfs->z_teardown_lock, FTAG); /* * Try to force unmount this file system. */ (void) zfs_umount(zfsvfs->z_vfs, 0); zfsvfs->z_unmounted = B_TRUE; return (0); } int zfs_set_version(zfsvfs_t *zfsvfs, uint64_t newvers) { int error; objset_t *os = zfsvfs->z_os; dmu_tx_t *tx; if (newvers < ZPL_VERSION_INITIAL || newvers > ZPL_VERSION) return (SET_ERROR(EINVAL)); if (newvers < zfsvfs->z_version) return (SET_ERROR(EINVAL)); if (zfs_spa_version_map(newvers) > spa_version(dmu_objset_spa(zfsvfs->z_os))) return (SET_ERROR(ENOTSUP)); tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, B_FALSE, ZPL_VERSION_STR); if (newvers >= ZPL_VERSION_SA && !zfsvfs->z_use_sa) { dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, B_TRUE, ZFS_SA_ATTRS); dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL); } error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); return (error); } error = zap_update(os, MASTER_NODE_OBJ, ZPL_VERSION_STR, 8, 1, &newvers, tx); if (error) { dmu_tx_commit(tx); return (error); } if (newvers >= ZPL_VERSION_SA && !zfsvfs->z_use_sa) { uint64_t sa_obj; ASSERT3U(spa_version(dmu_objset_spa(zfsvfs->z_os)), >=, SPA_VERSION_SA); sa_obj = zap_create(os, DMU_OT_SA_MASTER_NODE, DMU_OT_NONE, 0, tx); error = zap_add(os, MASTER_NODE_OBJ, ZFS_SA_ATTRS, 8, 1, &sa_obj, tx); ASSERT0(error); VERIFY(0 == sa_set_sa_object(os, sa_obj)); sa_register_update_callback(os, zfs_sa_upgrade); } spa_history_log_internal_ds(dmu_objset_ds(os), "upgrade", tx, "from %ju to %ju", (uintmax_t)zfsvfs->z_version, (uintmax_t)newvers); dmu_tx_commit(tx); zfsvfs->z_version = newvers; os->os_version = newvers; zfs_set_fuid_feature(zfsvfs); return (0); } /* * Read a property stored within the master node. */ int zfs_get_zplprop(objset_t *os, zfs_prop_t prop, uint64_t *value) { uint64_t *cached_copy = NULL; /* * Figure out where in the objset_t the cached copy would live, if it * is available for the requested property. */ if (os != NULL) { switch (prop) { case ZFS_PROP_VERSION: cached_copy = &os->os_version; break; case ZFS_PROP_NORMALIZE: cached_copy = &os->os_normalization; break; case ZFS_PROP_UTF8ONLY: cached_copy = &os->os_utf8only; break; case ZFS_PROP_CASE: cached_copy = &os->os_casesensitivity; break; default: break; } } if (cached_copy != NULL && *cached_copy != OBJSET_PROP_UNINITIALIZED) { *value = *cached_copy; return (0); } /* * If the property wasn't cached, look up the file system's value for * the property. For the version property, we look up a slightly * different string. */ const char *pname; int error = ENOENT; if (prop == ZFS_PROP_VERSION) { pname = ZPL_VERSION_STR; } else { pname = zfs_prop_to_name(prop); } if (os != NULL) { ASSERT3U(os->os_phys->os_type, ==, DMU_OST_ZFS); error = zap_lookup(os, MASTER_NODE_OBJ, pname, 8, 1, value); } if (error == ENOENT) { /* No value set, use the default value */ switch (prop) { case ZFS_PROP_VERSION: *value = ZPL_VERSION; break; case ZFS_PROP_NORMALIZE: case ZFS_PROP_UTF8ONLY: *value = 0; break; case ZFS_PROP_CASE: *value = ZFS_CASE_SENSITIVE; break; case ZFS_PROP_ACLTYPE: *value = ZFS_ACLTYPE_NFSV4; break; default: return (error); } error = 0; } /* * If one of the methods for getting the property value above worked, * copy it into the objset_t's cache. */ if (error == 0 && cached_copy != NULL) { *cached_copy = *value; } return (error); } /* * Return true if the corresponding vfs's unmounted flag is set. * Otherwise return false. * If this function returns true we know VFS unmount has been initiated. */ boolean_t zfs_get_vfs_flag_unmounted(objset_t *os) { zfsvfs_t *zfvp; boolean_t unmounted = B_FALSE; ASSERT(dmu_objset_type(os) == DMU_OST_ZFS); mutex_enter(&os->os_user_ptr_lock); zfvp = dmu_objset_get_user(os); if (zfvp != NULL && zfvp->z_vfs != NULL && (zfvp->z_vfs->mnt_kern_flag & MNTK_UNMOUNT)) unmounted = B_TRUE; mutex_exit(&os->os_user_ptr_lock); return (unmounted); } #ifdef _KERNEL void zfsvfs_update_fromname(const char *oldname, const char *newname) { char tmpbuf[MAXPATHLEN]; struct mount *mp; char *fromname; size_t oldlen; oldlen = strlen(oldname); mtx_lock(&mountlist_mtx); TAILQ_FOREACH(mp, &mountlist, mnt_list) { fromname = mp->mnt_stat.f_mntfromname; if (strcmp(fromname, oldname) == 0) { (void) strlcpy(fromname, newname, sizeof (mp->mnt_stat.f_mntfromname)); continue; } if (strncmp(fromname, oldname, oldlen) == 0 && (fromname[oldlen] == '/' || fromname[oldlen] == '@')) { (void) snprintf(tmpbuf, sizeof (tmpbuf), "%s%s", newname, fromname + oldlen); (void) strlcpy(fromname, tmpbuf, sizeof (mp->mnt_stat.f_mntfromname)); continue; } } mtx_unlock(&mountlist_mtx); } #endif diff --git a/module/os/linux/zfs/zfs_vfsops.c b/module/os/linux/zfs/zfs_vfsops.c index b218237d07ff..6c0d8134c486 100644 --- a/module/os/linux/zfs/zfs_vfsops.c +++ b/module/os/linux/zfs/zfs_vfsops.c @@ -1,2176 +1,2176 @@ /* * 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, 2018 by Delphix. All rights reserved. */ /* Portions Copyright 2010 Robert Milkowski */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_comutil.h" enum { TOKEN_RO, TOKEN_RW, TOKEN_SETUID, TOKEN_NOSETUID, TOKEN_EXEC, TOKEN_NOEXEC, TOKEN_DEVICES, TOKEN_NODEVICES, TOKEN_DIRXATTR, TOKEN_SAXATTR, TOKEN_XATTR, TOKEN_NOXATTR, TOKEN_ATIME, TOKEN_NOATIME, TOKEN_RELATIME, TOKEN_NORELATIME, TOKEN_NBMAND, TOKEN_NONBMAND, TOKEN_MNTPOINT, TOKEN_LAST, }; static const match_table_t zpl_tokens = { { TOKEN_RO, MNTOPT_RO }, { TOKEN_RW, MNTOPT_RW }, { TOKEN_SETUID, MNTOPT_SETUID }, { TOKEN_NOSETUID, MNTOPT_NOSETUID }, { TOKEN_EXEC, MNTOPT_EXEC }, { TOKEN_NOEXEC, MNTOPT_NOEXEC }, { TOKEN_DEVICES, MNTOPT_DEVICES }, { TOKEN_NODEVICES, MNTOPT_NODEVICES }, { TOKEN_DIRXATTR, MNTOPT_DIRXATTR }, { TOKEN_SAXATTR, MNTOPT_SAXATTR }, { TOKEN_XATTR, MNTOPT_XATTR }, { TOKEN_NOXATTR, MNTOPT_NOXATTR }, { TOKEN_ATIME, MNTOPT_ATIME }, { TOKEN_NOATIME, MNTOPT_NOATIME }, { TOKEN_RELATIME, MNTOPT_RELATIME }, { TOKEN_NORELATIME, MNTOPT_NORELATIME }, { TOKEN_NBMAND, MNTOPT_NBMAND }, { TOKEN_NONBMAND, MNTOPT_NONBMAND }, { TOKEN_MNTPOINT, MNTOPT_MNTPOINT "=%s" }, { TOKEN_LAST, NULL }, }; static void zfsvfs_vfs_free(vfs_t *vfsp) { if (vfsp != NULL) { if (vfsp->vfs_mntpoint != NULL) kmem_strfree(vfsp->vfs_mntpoint); kmem_free(vfsp, sizeof (vfs_t)); } } static int zfsvfs_parse_option(char *option, int token, substring_t *args, vfs_t *vfsp) { switch (token) { case TOKEN_RO: vfsp->vfs_readonly = B_TRUE; vfsp->vfs_do_readonly = B_TRUE; break; case TOKEN_RW: vfsp->vfs_readonly = B_FALSE; vfsp->vfs_do_readonly = B_TRUE; break; case TOKEN_SETUID: vfsp->vfs_setuid = B_TRUE; vfsp->vfs_do_setuid = B_TRUE; break; case TOKEN_NOSETUID: vfsp->vfs_setuid = B_FALSE; vfsp->vfs_do_setuid = B_TRUE; break; case TOKEN_EXEC: vfsp->vfs_exec = B_TRUE; vfsp->vfs_do_exec = B_TRUE; break; case TOKEN_NOEXEC: vfsp->vfs_exec = B_FALSE; vfsp->vfs_do_exec = B_TRUE; break; case TOKEN_DEVICES: vfsp->vfs_devices = B_TRUE; vfsp->vfs_do_devices = B_TRUE; break; case TOKEN_NODEVICES: vfsp->vfs_devices = B_FALSE; vfsp->vfs_do_devices = B_TRUE; break; case TOKEN_DIRXATTR: vfsp->vfs_xattr = ZFS_XATTR_DIR; vfsp->vfs_do_xattr = B_TRUE; break; case TOKEN_SAXATTR: vfsp->vfs_xattr = ZFS_XATTR_SA; vfsp->vfs_do_xattr = B_TRUE; break; case TOKEN_XATTR: vfsp->vfs_xattr = ZFS_XATTR_DIR; vfsp->vfs_do_xattr = B_TRUE; break; case TOKEN_NOXATTR: vfsp->vfs_xattr = ZFS_XATTR_OFF; vfsp->vfs_do_xattr = B_TRUE; break; case TOKEN_ATIME: vfsp->vfs_atime = B_TRUE; vfsp->vfs_do_atime = B_TRUE; break; case TOKEN_NOATIME: vfsp->vfs_atime = B_FALSE; vfsp->vfs_do_atime = B_TRUE; break; case TOKEN_RELATIME: vfsp->vfs_relatime = B_TRUE; vfsp->vfs_do_relatime = B_TRUE; break; case TOKEN_NORELATIME: vfsp->vfs_relatime = B_FALSE; vfsp->vfs_do_relatime = B_TRUE; break; case TOKEN_NBMAND: vfsp->vfs_nbmand = B_TRUE; vfsp->vfs_do_nbmand = B_TRUE; break; case TOKEN_NONBMAND: vfsp->vfs_nbmand = B_FALSE; vfsp->vfs_do_nbmand = B_TRUE; break; case TOKEN_MNTPOINT: vfsp->vfs_mntpoint = match_strdup(&args[0]); if (vfsp->vfs_mntpoint == NULL) return (SET_ERROR(ENOMEM)); break; default: break; } return (0); } /* * Parse the raw mntopts and return a vfs_t describing the options. */ static int zfsvfs_parse_options(char *mntopts, vfs_t **vfsp) { vfs_t *tmp_vfsp; int error; tmp_vfsp = kmem_zalloc(sizeof (vfs_t), KM_SLEEP); if (mntopts != NULL) { substring_t args[MAX_OPT_ARGS]; char *tmp_mntopts, *p, *t; int token; tmp_mntopts = t = kmem_strdup(mntopts); if (tmp_mntopts == NULL) return (SET_ERROR(ENOMEM)); while ((p = strsep(&t, ",")) != NULL) { if (!*p) continue; args[0].to = args[0].from = NULL; token = match_token(p, zpl_tokens, args); error = zfsvfs_parse_option(p, token, args, tmp_vfsp); if (error) { kmem_strfree(tmp_mntopts); zfsvfs_vfs_free(tmp_vfsp); return (error); } } kmem_strfree(tmp_mntopts); } *vfsp = tmp_vfsp; return (0); } boolean_t zfs_is_readonly(zfsvfs_t *zfsvfs) { return (!!(zfsvfs->z_sb->s_flags & SB_RDONLY)); } /*ARGSUSED*/ int zfs_sync(struct super_block *sb, int wait, cred_t *cr) { zfsvfs_t *zfsvfs = sb->s_fs_info; /* * Semantically, the only requirement is that the sync be initiated. * The DMU syncs out txgs frequently, so there's nothing to do. */ if (!wait) return (0); if (zfsvfs != NULL) { /* * Sync a specific filesystem. */ dsl_pool_t *dp; ZFS_ENTER(zfsvfs); dp = dmu_objset_pool(zfsvfs->z_os); /* * If the system is shutting down, then skip any * filesystems which may exist on a suspended pool. */ if (spa_suspended(dp->dp_spa)) { ZFS_EXIT(zfsvfs); return (0); } if (zfsvfs->z_log != NULL) zil_commit(zfsvfs->z_log, 0); ZFS_EXIT(zfsvfs); } else { /* * Sync all ZFS filesystems. This is what happens when you - * run sync(1M). Unlike other filesystems, ZFS honors the + * run sync(1). Unlike other filesystems, ZFS honors the * request by waiting for all pools to commit all dirty data. */ spa_sync_allpools(); } return (0); } static void atime_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; struct super_block *sb = zfsvfs->z_sb; if (sb == NULL) return; /* * Update SB_NOATIME bit in VFS super block. Since atime update is * determined by atime_needs_update(), atime_needs_update() needs to * return false if atime is turned off, and not unconditionally return * false if atime is turned on. */ if (newval) sb->s_flags &= ~SB_NOATIME; else sb->s_flags |= SB_NOATIME; } static void relatime_changed_cb(void *arg, uint64_t newval) { ((zfsvfs_t *)arg)->z_relatime = newval; } static void xattr_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; if (newval == ZFS_XATTR_OFF) { zfsvfs->z_flags &= ~ZSB_XATTR; } else { zfsvfs->z_flags |= ZSB_XATTR; if (newval == ZFS_XATTR_SA) zfsvfs->z_xattr_sa = B_TRUE; else zfsvfs->z_xattr_sa = B_FALSE; } } static void acltype_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; switch (newval) { case ZFS_ACLTYPE_NFSV4: case ZFS_ACLTYPE_OFF: zfsvfs->z_acl_type = ZFS_ACLTYPE_OFF; zfsvfs->z_sb->s_flags &= ~SB_POSIXACL; break; case ZFS_ACLTYPE_POSIX: #ifdef CONFIG_FS_POSIX_ACL zfsvfs->z_acl_type = ZFS_ACLTYPE_POSIX; zfsvfs->z_sb->s_flags |= SB_POSIXACL; #else zfsvfs->z_acl_type = ZFS_ACLTYPE_OFF; zfsvfs->z_sb->s_flags &= ~SB_POSIXACL; #endif /* CONFIG_FS_POSIX_ACL */ break; default: break; } } static void blksz_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; ASSERT3U(newval, <=, spa_maxblocksize(dmu_objset_spa(zfsvfs->z_os))); ASSERT3U(newval, >=, SPA_MINBLOCKSIZE); ASSERT(ISP2(newval)); zfsvfs->z_max_blksz = newval; } static void readonly_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; struct super_block *sb = zfsvfs->z_sb; if (sb == NULL) return; if (newval) sb->s_flags |= SB_RDONLY; else sb->s_flags &= ~SB_RDONLY; } static void devices_changed_cb(void *arg, uint64_t newval) { } static void setuid_changed_cb(void *arg, uint64_t newval) { } static void exec_changed_cb(void *arg, uint64_t newval) { } static void nbmand_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; struct super_block *sb = zfsvfs->z_sb; if (sb == NULL) return; if (newval == TRUE) sb->s_flags |= SB_MANDLOCK; else sb->s_flags &= ~SB_MANDLOCK; } static void snapdir_changed_cb(void *arg, uint64_t newval) { ((zfsvfs_t *)arg)->z_show_ctldir = newval; } static void vscan_changed_cb(void *arg, uint64_t newval) { ((zfsvfs_t *)arg)->z_vscan = newval; } static void acl_mode_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; zfsvfs->z_acl_mode = newval; } static void acl_inherit_changed_cb(void *arg, uint64_t newval) { ((zfsvfs_t *)arg)->z_acl_inherit = newval; } static int zfs_register_callbacks(vfs_t *vfsp) { struct dsl_dataset *ds = NULL; objset_t *os = NULL; zfsvfs_t *zfsvfs = NULL; int error = 0; ASSERT(vfsp); zfsvfs = vfsp->vfs_data; ASSERT(zfsvfs); os = zfsvfs->z_os; /* * The act of registering our callbacks will destroy any mount * options we may have. In order to enable temporary overrides * of mount options, we stash away the current values and * restore them after we register the callbacks. */ if (zfs_is_readonly(zfsvfs) || !spa_writeable(dmu_objset_spa(os))) { vfsp->vfs_do_readonly = B_TRUE; vfsp->vfs_readonly = B_TRUE; } /* * Register property callbacks. * * It would probably be fine to just check for i/o error from * the first prop_register(), but I guess I like to go * overboard... */ ds = dmu_objset_ds(os); dsl_pool_config_enter(dmu_objset_pool(os), FTAG); error = dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_ATIME), atime_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_RELATIME), relatime_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_XATTR), xattr_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_RECORDSIZE), blksz_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_READONLY), readonly_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_DEVICES), devices_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_SETUID), setuid_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_EXEC), exec_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_SNAPDIR), snapdir_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_ACLTYPE), acltype_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_ACLMODE), acl_mode_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_ACLINHERIT), acl_inherit_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_VSCAN), vscan_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_NBMAND), nbmand_changed_cb, zfsvfs); dsl_pool_config_exit(dmu_objset_pool(os), FTAG); if (error) goto unregister; /* * Invoke our callbacks to restore temporary mount options. */ if (vfsp->vfs_do_readonly) readonly_changed_cb(zfsvfs, vfsp->vfs_readonly); if (vfsp->vfs_do_setuid) setuid_changed_cb(zfsvfs, vfsp->vfs_setuid); if (vfsp->vfs_do_exec) exec_changed_cb(zfsvfs, vfsp->vfs_exec); if (vfsp->vfs_do_devices) devices_changed_cb(zfsvfs, vfsp->vfs_devices); if (vfsp->vfs_do_xattr) xattr_changed_cb(zfsvfs, vfsp->vfs_xattr); if (vfsp->vfs_do_atime) atime_changed_cb(zfsvfs, vfsp->vfs_atime); if (vfsp->vfs_do_relatime) relatime_changed_cb(zfsvfs, vfsp->vfs_relatime); if (vfsp->vfs_do_nbmand) nbmand_changed_cb(zfsvfs, vfsp->vfs_nbmand); return (0); unregister: dsl_prop_unregister_all(ds, zfsvfs); return (error); } /* * Takes a dataset, a property, a value and that value's setpoint as * found in the ZAP. Checks if the property has been changed in the vfs. * If so, val and setpoint will be overwritten with updated content. * Otherwise, they are left unchanged. */ int zfs_get_temporary_prop(dsl_dataset_t *ds, zfs_prop_t zfs_prop, uint64_t *val, char *setpoint) { int error; zfsvfs_t *zfvp; vfs_t *vfsp; objset_t *os; uint64_t tmp = *val; error = dmu_objset_from_ds(ds, &os); if (error != 0) return (error); if (dmu_objset_type(os) != DMU_OST_ZFS) return (EINVAL); mutex_enter(&os->os_user_ptr_lock); zfvp = dmu_objset_get_user(os); mutex_exit(&os->os_user_ptr_lock); if (zfvp == NULL) return (ESRCH); vfsp = zfvp->z_vfs; switch (zfs_prop) { case ZFS_PROP_ATIME: if (vfsp->vfs_do_atime) tmp = vfsp->vfs_atime; break; case ZFS_PROP_RELATIME: if (vfsp->vfs_do_relatime) tmp = vfsp->vfs_relatime; break; case ZFS_PROP_DEVICES: if (vfsp->vfs_do_devices) tmp = vfsp->vfs_devices; break; case ZFS_PROP_EXEC: if (vfsp->vfs_do_exec) tmp = vfsp->vfs_exec; break; case ZFS_PROP_SETUID: if (vfsp->vfs_do_setuid) tmp = vfsp->vfs_setuid; break; case ZFS_PROP_READONLY: if (vfsp->vfs_do_readonly) tmp = vfsp->vfs_readonly; break; case ZFS_PROP_XATTR: if (vfsp->vfs_do_xattr) tmp = vfsp->vfs_xattr; break; case ZFS_PROP_NBMAND: if (vfsp->vfs_do_nbmand) tmp = vfsp->vfs_nbmand; break; default: return (ENOENT); } if (tmp != *val) { (void) strcpy(setpoint, "temporary"); *val = tmp; } return (0); } /* * Associate this zfsvfs with the given objset, which must be owned. * This will cache a bunch of on-disk state from the objset in the * zfsvfs. */ static int zfsvfs_init(zfsvfs_t *zfsvfs, objset_t *os) { int error; uint64_t val; zfsvfs->z_max_blksz = SPA_OLD_MAXBLOCKSIZE; zfsvfs->z_show_ctldir = ZFS_SNAPDIR_VISIBLE; zfsvfs->z_os = os; error = zfs_get_zplprop(os, ZFS_PROP_VERSION, &zfsvfs->z_version); if (error != 0) return (error); if (zfsvfs->z_version > zfs_zpl_version_map(spa_version(dmu_objset_spa(os)))) { (void) printk("Can't mount a version %lld file system " "on a version %lld pool\n. Pool must be upgraded to mount " "this file system.\n", (u_longlong_t)zfsvfs->z_version, (u_longlong_t)spa_version(dmu_objset_spa(os))); return (SET_ERROR(ENOTSUP)); } error = zfs_get_zplprop(os, ZFS_PROP_NORMALIZE, &val); if (error != 0) return (error); zfsvfs->z_norm = (int)val; error = zfs_get_zplprop(os, ZFS_PROP_UTF8ONLY, &val); if (error != 0) return (error); zfsvfs->z_utf8 = (val != 0); error = zfs_get_zplprop(os, ZFS_PROP_CASE, &val); if (error != 0) return (error); zfsvfs->z_case = (uint_t)val; if ((error = zfs_get_zplprop(os, ZFS_PROP_ACLTYPE, &val)) != 0) return (error); zfsvfs->z_acl_type = (uint_t)val; /* * Fold case on file systems that are always or sometimes case * insensitive. */ if (zfsvfs->z_case == ZFS_CASE_INSENSITIVE || zfsvfs->z_case == ZFS_CASE_MIXED) zfsvfs->z_norm |= U8_TEXTPREP_TOUPPER; zfsvfs->z_use_fuids = USE_FUIDS(zfsvfs->z_version, zfsvfs->z_os); zfsvfs->z_use_sa = USE_SA(zfsvfs->z_version, zfsvfs->z_os); uint64_t sa_obj = 0; if (zfsvfs->z_use_sa) { /* should either have both of these objects or none */ error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_SA_ATTRS, 8, 1, &sa_obj); if (error != 0) return (error); error = zfs_get_zplprop(os, ZFS_PROP_XATTR, &val); if ((error == 0) && (val == ZFS_XATTR_SA)) zfsvfs->z_xattr_sa = B_TRUE; } error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_ROOT_OBJ, 8, 1, &zfsvfs->z_root); if (error != 0) return (error); ASSERT(zfsvfs->z_root != 0); error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_UNLINKED_SET, 8, 1, &zfsvfs->z_unlinkedobj); if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_USERQUOTA], 8, 1, &zfsvfs->z_userquota_obj); if (error == ENOENT) zfsvfs->z_userquota_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_GROUPQUOTA], 8, 1, &zfsvfs->z_groupquota_obj); if (error == ENOENT) zfsvfs->z_groupquota_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_PROJECTQUOTA], 8, 1, &zfsvfs->z_projectquota_obj); if (error == ENOENT) zfsvfs->z_projectquota_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_USEROBJQUOTA], 8, 1, &zfsvfs->z_userobjquota_obj); if (error == ENOENT) zfsvfs->z_userobjquota_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_GROUPOBJQUOTA], 8, 1, &zfsvfs->z_groupobjquota_obj); if (error == ENOENT) zfsvfs->z_groupobjquota_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_PROJECTOBJQUOTA], 8, 1, &zfsvfs->z_projectobjquota_obj); if (error == ENOENT) zfsvfs->z_projectobjquota_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_FUID_TABLES, 8, 1, &zfsvfs->z_fuid_obj); if (error == ENOENT) zfsvfs->z_fuid_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_SHARES_DIR, 8, 1, &zfsvfs->z_shares_dir); if (error == ENOENT) zfsvfs->z_shares_dir = 0; else if (error != 0) return (error); error = sa_setup(os, sa_obj, zfs_attr_table, ZPL_END, &zfsvfs->z_attr_table); if (error != 0) return (error); if (zfsvfs->z_version >= ZPL_VERSION_SA) sa_register_update_callback(os, zfs_sa_upgrade); return (0); } int zfsvfs_create(const char *osname, boolean_t readonly, zfsvfs_t **zfvp) { objset_t *os; zfsvfs_t *zfsvfs; int error; boolean_t ro = (readonly || (strchr(osname, '@') != NULL)); zfsvfs = kmem_zalloc(sizeof (zfsvfs_t), KM_SLEEP); error = dmu_objset_own(osname, DMU_OST_ZFS, ro, B_TRUE, zfsvfs, &os); if (error != 0) { kmem_free(zfsvfs, sizeof (zfsvfs_t)); return (error); } error = zfsvfs_create_impl(zfvp, zfsvfs, os); if (error != 0) { dmu_objset_disown(os, B_TRUE, zfsvfs); } return (error); } /* * Note: zfsvfs is assumed to be malloc'd, and will be freed by this function * on a failure. Do not pass in a statically allocated zfsvfs. */ int zfsvfs_create_impl(zfsvfs_t **zfvp, zfsvfs_t *zfsvfs, objset_t *os) { int error; zfsvfs->z_vfs = NULL; zfsvfs->z_sb = NULL; zfsvfs->z_parent = zfsvfs; mutex_init(&zfsvfs->z_znodes_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&zfsvfs->z_lock, NULL, MUTEX_DEFAULT, NULL); list_create(&zfsvfs->z_all_znodes, sizeof (znode_t), offsetof(znode_t, z_link_node)); rrm_init(&zfsvfs->z_teardown_lock, B_FALSE); rw_init(&zfsvfs->z_teardown_inactive_lock, NULL, RW_DEFAULT, NULL); rw_init(&zfsvfs->z_fuid_lock, NULL, RW_DEFAULT, NULL); int size = MIN(1 << (highbit64(zfs_object_mutex_size) - 1), ZFS_OBJ_MTX_MAX); zfsvfs->z_hold_size = size; zfsvfs->z_hold_trees = vmem_zalloc(sizeof (avl_tree_t) * size, KM_SLEEP); zfsvfs->z_hold_locks = vmem_zalloc(sizeof (kmutex_t) * size, KM_SLEEP); for (int i = 0; i != size; i++) { avl_create(&zfsvfs->z_hold_trees[i], zfs_znode_hold_compare, sizeof (znode_hold_t), offsetof(znode_hold_t, zh_node)); mutex_init(&zfsvfs->z_hold_locks[i], NULL, MUTEX_DEFAULT, NULL); } error = zfsvfs_init(zfsvfs, os); if (error != 0) { *zfvp = NULL; zfsvfs_free(zfsvfs); return (error); } zfsvfs->z_drain_task = TASKQID_INVALID; zfsvfs->z_draining = B_FALSE; zfsvfs->z_drain_cancel = B_TRUE; *zfvp = zfsvfs; return (0); } static int zfsvfs_setup(zfsvfs_t *zfsvfs, boolean_t mounting) { int error; boolean_t readonly = zfs_is_readonly(zfsvfs); error = zfs_register_callbacks(zfsvfs->z_vfs); if (error) return (error); zfsvfs->z_log = zil_open(zfsvfs->z_os, zfs_get_data); /* * If we are not mounting (ie: online recv), then we don't * have to worry about replaying the log as we blocked all * operations out since we closed the ZIL. */ if (mounting) { ASSERT3P(zfsvfs->z_kstat.dk_kstats, ==, NULL); dataset_kstats_create(&zfsvfs->z_kstat, zfsvfs->z_os); /* * During replay we remove the read only flag to * allow replays to succeed. */ if (readonly != 0) { readonly_changed_cb(zfsvfs, B_FALSE); } else { zap_stats_t zs; if (zap_get_stats(zfsvfs->z_os, zfsvfs->z_unlinkedobj, &zs) == 0) { dataset_kstats_update_nunlinks_kstat( &zfsvfs->z_kstat, zs.zs_num_entries); dprintf_ds(zfsvfs->z_os->os_dsl_dataset, "num_entries in unlinked set: %llu", zs.zs_num_entries); } zfs_unlinked_drain(zfsvfs); dsl_dir_t *dd = zfsvfs->z_os->os_dsl_dataset->ds_dir; dd->dd_activity_cancelled = B_FALSE; } /* * Parse and replay the intent log. * * Because of ziltest, this must be done after * zfs_unlinked_drain(). (Further note: ziltest * doesn't use readonly mounts, where * zfs_unlinked_drain() isn't called.) This is because * ziltest causes spa_sync() to think it's committed, * but actually it is not, so the intent log contains * many txg's worth of changes. * * In particular, if object N is in the unlinked set in * the last txg to actually sync, then it could be * actually freed in a later txg and then reallocated * in a yet later txg. This would write a "create * object N" record to the intent log. Normally, this * would be fine because the spa_sync() would have * written out the fact that object N is free, before * we could write the "create object N" intent log * record. * * But when we are in ziltest mode, we advance the "open * txg" without actually spa_sync()-ing the changes to * disk. So we would see that object N is still * allocated and in the unlinked set, and there is an * intent log record saying to allocate it. */ if (spa_writeable(dmu_objset_spa(zfsvfs->z_os))) { if (zil_replay_disable) { zil_destroy(zfsvfs->z_log, B_FALSE); } else { zfsvfs->z_replay = B_TRUE; zil_replay(zfsvfs->z_os, zfsvfs, zfs_replay_vector); zfsvfs->z_replay = B_FALSE; } } /* restore readonly bit */ if (readonly != 0) readonly_changed_cb(zfsvfs, B_TRUE); } /* * Set the objset user_ptr to track its zfsvfs. */ mutex_enter(&zfsvfs->z_os->os_user_ptr_lock); dmu_objset_set_user(zfsvfs->z_os, zfsvfs); mutex_exit(&zfsvfs->z_os->os_user_ptr_lock); return (0); } void zfsvfs_free(zfsvfs_t *zfsvfs) { int i, size = zfsvfs->z_hold_size; zfs_fuid_destroy(zfsvfs); mutex_destroy(&zfsvfs->z_znodes_lock); mutex_destroy(&zfsvfs->z_lock); list_destroy(&zfsvfs->z_all_znodes); rrm_destroy(&zfsvfs->z_teardown_lock); rw_destroy(&zfsvfs->z_teardown_inactive_lock); rw_destroy(&zfsvfs->z_fuid_lock); for (i = 0; i != size; i++) { avl_destroy(&zfsvfs->z_hold_trees[i]); mutex_destroy(&zfsvfs->z_hold_locks[i]); } vmem_free(zfsvfs->z_hold_trees, sizeof (avl_tree_t) * size); vmem_free(zfsvfs->z_hold_locks, sizeof (kmutex_t) * size); zfsvfs_vfs_free(zfsvfs->z_vfs); dataset_kstats_destroy(&zfsvfs->z_kstat); kmem_free(zfsvfs, sizeof (zfsvfs_t)); } static void zfs_set_fuid_feature(zfsvfs_t *zfsvfs) { zfsvfs->z_use_fuids = USE_FUIDS(zfsvfs->z_version, zfsvfs->z_os); zfsvfs->z_use_sa = USE_SA(zfsvfs->z_version, zfsvfs->z_os); } static void zfs_unregister_callbacks(zfsvfs_t *zfsvfs) { objset_t *os = zfsvfs->z_os; if (!dmu_objset_is_snapshot(os)) dsl_prop_unregister_all(dmu_objset_ds(os), zfsvfs); } #ifdef HAVE_MLSLABEL /* * Check that the hex label string is appropriate for the dataset being * mounted into the global_zone proper. * * Return an error if the hex label string is not default or * admin_low/admin_high. For admin_low labels, the corresponding * dataset must be readonly. */ int zfs_check_global_label(const char *dsname, const char *hexsl) { if (strcasecmp(hexsl, ZFS_MLSLABEL_DEFAULT) == 0) return (0); if (strcasecmp(hexsl, ADMIN_HIGH) == 0) return (0); if (strcasecmp(hexsl, ADMIN_LOW) == 0) { /* must be readonly */ uint64_t rdonly; if (dsl_prop_get_integer(dsname, zfs_prop_to_name(ZFS_PROP_READONLY), &rdonly, NULL)) return (SET_ERROR(EACCES)); return (rdonly ? 0 : SET_ERROR(EACCES)); } return (SET_ERROR(EACCES)); } #endif /* HAVE_MLSLABEL */ static int zfs_statfs_project(zfsvfs_t *zfsvfs, znode_t *zp, struct kstatfs *statp, uint32_t bshift) { char buf[20 + DMU_OBJACCT_PREFIX_LEN]; uint64_t offset = DMU_OBJACCT_PREFIX_LEN; uint64_t quota; uint64_t used; int err; strlcpy(buf, DMU_OBJACCT_PREFIX, DMU_OBJACCT_PREFIX_LEN + 1); err = zfs_id_to_fuidstr(zfsvfs, NULL, zp->z_projid, buf + offset, sizeof (buf) - offset, B_FALSE); if (err) return (err); if (zfsvfs->z_projectquota_obj == 0) goto objs; err = zap_lookup(zfsvfs->z_os, zfsvfs->z_projectquota_obj, buf + offset, 8, 1, "a); if (err == ENOENT) goto objs; else if (err) return (err); err = zap_lookup(zfsvfs->z_os, DMU_PROJECTUSED_OBJECT, buf + offset, 8, 1, &used); if (unlikely(err == ENOENT)) { uint32_t blksize; u_longlong_t nblocks; /* * Quota accounting is async, so it is possible race case. * There is at least one object with the given project ID. */ sa_object_size(zp->z_sa_hdl, &blksize, &nblocks); if (unlikely(zp->z_blksz == 0)) blksize = zfsvfs->z_max_blksz; used = blksize * nblocks; } else if (err) { return (err); } statp->f_blocks = quota >> bshift; statp->f_bfree = (quota > used) ? ((quota - used) >> bshift) : 0; statp->f_bavail = statp->f_bfree; objs: if (zfsvfs->z_projectobjquota_obj == 0) return (0); err = zap_lookup(zfsvfs->z_os, zfsvfs->z_projectobjquota_obj, buf + offset, 8, 1, "a); if (err == ENOENT) return (0); else if (err) return (err); err = zap_lookup(zfsvfs->z_os, DMU_PROJECTUSED_OBJECT, buf, 8, 1, &used); if (unlikely(err == ENOENT)) { /* * Quota accounting is async, so it is possible race case. * There is at least one object with the given project ID. */ used = 1; } else if (err) { return (err); } statp->f_files = quota; statp->f_ffree = (quota > used) ? (quota - used) : 0; return (0); } int zfs_statvfs(struct inode *ip, struct kstatfs *statp) { zfsvfs_t *zfsvfs = ITOZSB(ip); uint64_t refdbytes, availbytes, usedobjs, availobjs; int err = 0; ZFS_ENTER(zfsvfs); dmu_objset_space(zfsvfs->z_os, &refdbytes, &availbytes, &usedobjs, &availobjs); uint64_t fsid = dmu_objset_fsid_guid(zfsvfs->z_os); /* * The underlying storage pool actually uses multiple block * size. Under Solaris frsize (fragment size) is reported as * the smallest block size we support, and bsize (block size) * as the filesystem's maximum block size. Unfortunately, * under Linux the fragment size and block size are often used * interchangeably. Thus we are forced to report both of them * as the filesystem's maximum block size. */ statp->f_frsize = zfsvfs->z_max_blksz; statp->f_bsize = zfsvfs->z_max_blksz; uint32_t bshift = fls(statp->f_bsize) - 1; /* * The following report "total" blocks of various kinds in * the file system, but reported in terms of f_bsize - the * "preferred" size. */ /* Round up so we never have a filesystem using 0 blocks. */ refdbytes = P2ROUNDUP(refdbytes, statp->f_bsize); statp->f_blocks = (refdbytes + availbytes) >> bshift; statp->f_bfree = availbytes >> bshift; statp->f_bavail = statp->f_bfree; /* no root reservation */ /* * statvfs() should really be called statufs(), because it assumes * static metadata. ZFS doesn't preallocate files, so the best * we can do is report the max that could possibly fit in f_files, * and that minus the number actually used in f_ffree. * For f_ffree, report the smaller of the number of objects available * and the number of blocks (each object will take at least a block). */ statp->f_ffree = MIN(availobjs, availbytes >> DNODE_SHIFT); statp->f_files = statp->f_ffree + usedobjs; statp->f_fsid.val[0] = (uint32_t)fsid; statp->f_fsid.val[1] = (uint32_t)(fsid >> 32); statp->f_type = ZFS_SUPER_MAGIC; statp->f_namelen = MAXNAMELEN - 1; /* * We have all of 40 characters to stuff a string here. * Is there anything useful we could/should provide? */ bzero(statp->f_spare, sizeof (statp->f_spare)); if (dmu_objset_projectquota_enabled(zfsvfs->z_os) && dmu_objset_projectquota_present(zfsvfs->z_os)) { znode_t *zp = ITOZ(ip); if (zp->z_pflags & ZFS_PROJINHERIT && zp->z_projid && zpl_is_valid_projid(zp->z_projid)) err = zfs_statfs_project(zfsvfs, zp, statp, bshift); } ZFS_EXIT(zfsvfs); return (err); } static int zfs_root(zfsvfs_t *zfsvfs, struct inode **ipp) { znode_t *rootzp; int error; ZFS_ENTER(zfsvfs); error = zfs_zget(zfsvfs, zfsvfs->z_root, &rootzp); if (error == 0) *ipp = ZTOI(rootzp); ZFS_EXIT(zfsvfs); return (error); } /* * Linux kernels older than 3.1 do not support a per-filesystem shrinker. * To accommodate this we must improvise and manually walk the list of znodes * attempting to prune dentries in order to be able to drop the inodes. * * To avoid scanning the same znodes multiple times they are always rotated * to the end of the z_all_znodes list. New znodes are inserted at the * end of the list so we're always scanning the oldest znodes first. */ static int zfs_prune_aliases(zfsvfs_t *zfsvfs, unsigned long nr_to_scan) { znode_t **zp_array, *zp; int max_array = MIN(nr_to_scan, PAGE_SIZE * 8 / sizeof (znode_t *)); int objects = 0; int i = 0, j = 0; zp_array = kmem_zalloc(max_array * sizeof (znode_t *), KM_SLEEP); mutex_enter(&zfsvfs->z_znodes_lock); while ((zp = list_head(&zfsvfs->z_all_znodes)) != NULL) { if ((i++ > nr_to_scan) || (j >= max_array)) break; ASSERT(list_link_active(&zp->z_link_node)); list_remove(&zfsvfs->z_all_znodes, zp); list_insert_tail(&zfsvfs->z_all_znodes, zp); /* Skip active znodes and .zfs entries */ if (MUTEX_HELD(&zp->z_lock) || zp->z_is_ctldir) continue; if (igrab(ZTOI(zp)) == NULL) continue; zp_array[j] = zp; j++; } mutex_exit(&zfsvfs->z_znodes_lock); for (i = 0; i < j; i++) { zp = zp_array[i]; ASSERT3P(zp, !=, NULL); d_prune_aliases(ZTOI(zp)); if (atomic_read(&ZTOI(zp)->i_count) == 1) objects++; zrele(zp); } kmem_free(zp_array, max_array * sizeof (znode_t *)); return (objects); } /* * The ARC has requested that the filesystem drop entries from the dentry * and inode caches. This can occur when the ARC needs to free meta data * blocks but can't because they are all pinned by entries in these caches. */ int zfs_prune(struct super_block *sb, unsigned long nr_to_scan, int *objects) { zfsvfs_t *zfsvfs = sb->s_fs_info; int error = 0; struct shrinker *shrinker = &sb->s_shrink; struct shrink_control sc = { .nr_to_scan = nr_to_scan, .gfp_mask = GFP_KERNEL, }; ZFS_ENTER(zfsvfs); #if defined(HAVE_SPLIT_SHRINKER_CALLBACK) && \ defined(SHRINK_CONTROL_HAS_NID) && \ defined(SHRINKER_NUMA_AWARE) if (sb->s_shrink.flags & SHRINKER_NUMA_AWARE) { *objects = 0; for_each_online_node(sc.nid) { *objects += (*shrinker->scan_objects)(shrinker, &sc); } } else { *objects = (*shrinker->scan_objects)(shrinker, &sc); } #elif defined(HAVE_SPLIT_SHRINKER_CALLBACK) *objects = (*shrinker->scan_objects)(shrinker, &sc); #elif defined(HAVE_SINGLE_SHRINKER_CALLBACK) *objects = (*shrinker->shrink)(shrinker, &sc); #elif defined(HAVE_D_PRUNE_ALIASES) #define D_PRUNE_ALIASES_IS_DEFAULT *objects = zfs_prune_aliases(zfsvfs, nr_to_scan); #else #error "No available dentry and inode cache pruning mechanism." #endif #if defined(HAVE_D_PRUNE_ALIASES) && !defined(D_PRUNE_ALIASES_IS_DEFAULT) #undef D_PRUNE_ALIASES_IS_DEFAULT /* * Fall back to zfs_prune_aliases if the kernel's per-superblock * shrinker couldn't free anything, possibly due to the inodes being * allocated in a different memcg. */ if (*objects == 0) *objects = zfs_prune_aliases(zfsvfs, nr_to_scan); #endif ZFS_EXIT(zfsvfs); dprintf_ds(zfsvfs->z_os->os_dsl_dataset, "pruning, nr_to_scan=%lu objects=%d error=%d\n", nr_to_scan, *objects, error); return (error); } /* * Teardown the zfsvfs_t. * * Note, if 'unmounting' is FALSE, we return with the 'z_teardown_lock' * and 'z_teardown_inactive_lock' held. */ static int zfsvfs_teardown(zfsvfs_t *zfsvfs, boolean_t unmounting) { znode_t *zp; zfs_unlinked_drain_stop_wait(zfsvfs); /* * If someone has not already unmounted this file system, * drain the zrele_taskq to ensure all active references to the * zfsvfs_t have been handled only then can it be safely destroyed. */ if (zfsvfs->z_os) { /* * If we're unmounting we have to wait for the list to * drain completely. * * If we're not unmounting there's no guarantee the list * will drain completely, but iputs run from the taskq * may add the parents of dir-based xattrs to the taskq * so we want to wait for these. * * We can safely read z_nr_znodes without locking because the * VFS has already blocked operations which add to the * z_all_znodes list and thus increment z_nr_znodes. */ int round = 0; while (zfsvfs->z_nr_znodes > 0) { taskq_wait_outstanding(dsl_pool_zrele_taskq( dmu_objset_pool(zfsvfs->z_os)), 0); if (++round > 1 && !unmounting) break; } } rrm_enter(&zfsvfs->z_teardown_lock, RW_WRITER, FTAG); if (!unmounting) { /* * We purge the parent filesystem's super block as the * parent filesystem and all of its snapshots have their * inode's super block set to the parent's filesystem's * super block. Note, 'z_parent' is self referential * for non-snapshots. */ shrink_dcache_sb(zfsvfs->z_parent->z_sb); } /* * Close the zil. NB: Can't close the zil while zfs_inactive * threads are blocked as zil_close can call zfs_inactive. */ if (zfsvfs->z_log) { zil_close(zfsvfs->z_log); zfsvfs->z_log = NULL; } rw_enter(&zfsvfs->z_teardown_inactive_lock, RW_WRITER); /* * If we are not unmounting (ie: online recv) and someone already * unmounted this file system while we were doing the switcheroo, * or a reopen of z_os failed then just bail out now. */ if (!unmounting && (zfsvfs->z_unmounted || zfsvfs->z_os == NULL)) { rw_exit(&zfsvfs->z_teardown_inactive_lock); rrm_exit(&zfsvfs->z_teardown_lock, FTAG); return (SET_ERROR(EIO)); } /* * At this point there are no VFS ops active, and any new VFS ops * will fail with EIO since we have z_teardown_lock for writer (only * relevant for forced unmount). * * Release all holds on dbufs. We also grab an extra reference to all * the remaining inodes so that the kernel does not attempt to free * any inodes of a suspended fs. This can cause deadlocks since the * zfs_resume_fs() process may involve starting threads, which might * attempt to free unreferenced inodes to free up memory for the new * thread. */ if (!unmounting) { mutex_enter(&zfsvfs->z_znodes_lock); for (zp = list_head(&zfsvfs->z_all_znodes); zp != NULL; zp = list_next(&zfsvfs->z_all_znodes, zp)) { if (zp->z_sa_hdl) zfs_znode_dmu_fini(zp); if (igrab(ZTOI(zp)) != NULL) zp->z_suspended = B_TRUE; } mutex_exit(&zfsvfs->z_znodes_lock); } /* * If we are unmounting, set the unmounted flag and let new VFS ops * unblock. zfs_inactive will have the unmounted behavior, and all * other VFS ops will fail with EIO. */ if (unmounting) { zfsvfs->z_unmounted = B_TRUE; rw_exit(&zfsvfs->z_teardown_inactive_lock); rrm_exit(&zfsvfs->z_teardown_lock, FTAG); } /* * z_os will be NULL if there was an error in attempting to reopen * zfsvfs, so just return as the properties had already been * * unregistered and cached data had been evicted before. */ if (zfsvfs->z_os == NULL) return (0); /* * Unregister properties. */ zfs_unregister_callbacks(zfsvfs); /* * Evict cached data. We must write out any dirty data before * disowning the dataset. */ objset_t *os = zfsvfs->z_os; boolean_t os_dirty = B_FALSE; for (int t = 0; t < TXG_SIZE; t++) { if (dmu_objset_is_dirty(os, t)) { os_dirty = B_TRUE; break; } } if (!zfs_is_readonly(zfsvfs) && os_dirty) { txg_wait_synced(dmu_objset_pool(zfsvfs->z_os), 0); } dmu_objset_evict_dbufs(zfsvfs->z_os); dsl_dir_t *dd = os->os_dsl_dataset->ds_dir; dsl_dir_cancel_waiters(dd); return (0); } #if defined(HAVE_SUPER_SETUP_BDI_NAME) atomic_long_t zfs_bdi_seq = ATOMIC_LONG_INIT(0); #endif int zfs_domount(struct super_block *sb, zfs_mnt_t *zm, int silent) { const char *osname = zm->mnt_osname; struct inode *root_inode; uint64_t recordsize; int error = 0; zfsvfs_t *zfsvfs = NULL; vfs_t *vfs = NULL; ASSERT(zm); ASSERT(osname); error = zfsvfs_parse_options(zm->mnt_data, &vfs); if (error) return (error); error = zfsvfs_create(osname, vfs->vfs_readonly, &zfsvfs); if (error) { zfsvfs_vfs_free(vfs); goto out; } if ((error = dsl_prop_get_integer(osname, "recordsize", &recordsize, NULL))) { zfsvfs_vfs_free(vfs); goto out; } vfs->vfs_data = zfsvfs; zfsvfs->z_vfs = vfs; zfsvfs->z_sb = sb; sb->s_fs_info = zfsvfs; sb->s_magic = ZFS_SUPER_MAGIC; sb->s_maxbytes = MAX_LFS_FILESIZE; sb->s_time_gran = 1; sb->s_blocksize = recordsize; sb->s_blocksize_bits = ilog2(recordsize); error = -zpl_bdi_setup(sb, "zfs"); if (error) goto out; sb->s_bdi->ra_pages = 0; /* Set callback operations for the file system. */ sb->s_op = &zpl_super_operations; sb->s_xattr = zpl_xattr_handlers; sb->s_export_op = &zpl_export_operations; sb->s_d_op = &zpl_dentry_operations; /* Set features for file system. */ zfs_set_fuid_feature(zfsvfs); if (dmu_objset_is_snapshot(zfsvfs->z_os)) { uint64_t pval; atime_changed_cb(zfsvfs, B_FALSE); readonly_changed_cb(zfsvfs, B_TRUE); if ((error = dsl_prop_get_integer(osname, "xattr", &pval, NULL))) goto out; xattr_changed_cb(zfsvfs, pval); if ((error = dsl_prop_get_integer(osname, "acltype", &pval, NULL))) goto out; acltype_changed_cb(zfsvfs, pval); zfsvfs->z_issnap = B_TRUE; zfsvfs->z_os->os_sync = ZFS_SYNC_DISABLED; zfsvfs->z_snap_defer_time = jiffies; mutex_enter(&zfsvfs->z_os->os_user_ptr_lock); dmu_objset_set_user(zfsvfs->z_os, zfsvfs); mutex_exit(&zfsvfs->z_os->os_user_ptr_lock); } else { if ((error = zfsvfs_setup(zfsvfs, B_TRUE))) goto out; } /* Allocate a root inode for the filesystem. */ error = zfs_root(zfsvfs, &root_inode); if (error) { (void) zfs_umount(sb); goto out; } /* Allocate a root dentry for the filesystem */ sb->s_root = d_make_root(root_inode); if (sb->s_root == NULL) { (void) zfs_umount(sb); error = SET_ERROR(ENOMEM); goto out; } if (!zfsvfs->z_issnap) zfsctl_create(zfsvfs); zfsvfs->z_arc_prune = arc_add_prune_callback(zpl_prune_sb, sb); out: if (error) { if (zfsvfs != NULL) { dmu_objset_disown(zfsvfs->z_os, B_TRUE, zfsvfs); zfsvfs_free(zfsvfs); } /* * make sure we don't have dangling sb->s_fs_info which * zfs_preumount will use. */ sb->s_fs_info = NULL; } return (error); } /* * Called when an unmount is requested and certain sanity checks have * already passed. At this point no dentries or inodes have been reclaimed * from their respective caches. We drop the extra reference on the .zfs * control directory to allow everything to be reclaimed. All snapshots * must already have been unmounted to reach this point. */ void zfs_preumount(struct super_block *sb) { zfsvfs_t *zfsvfs = sb->s_fs_info; /* zfsvfs is NULL when zfs_domount fails during mount */ if (zfsvfs) { zfs_unlinked_drain_stop_wait(zfsvfs); zfsctl_destroy(sb->s_fs_info); /* * Wait for zrele_async before entering evict_inodes in * generic_shutdown_super. The reason we must finish before * evict_inodes is when lazytime is on, or when zfs_purgedir * calls zfs_zget, zrele would bump i_count from 0 to 1. This * would race with the i_count check in evict_inodes. This means * it could destroy the inode while we are still using it. * * We wait for two passes. xattr directories in the first pass * may add xattr entries in zfs_purgedir, so in the second pass * we wait for them. We don't use taskq_wait here because it is * a pool wide taskq. Other mounted filesystems can constantly * do zrele_async and there's no guarantee when taskq will be * empty. */ taskq_wait_outstanding(dsl_pool_zrele_taskq( dmu_objset_pool(zfsvfs->z_os)), 0); taskq_wait_outstanding(dsl_pool_zrele_taskq( dmu_objset_pool(zfsvfs->z_os)), 0); } } /* * Called once all other unmount released tear down has occurred. * It is our responsibility to release any remaining infrastructure. */ /*ARGSUSED*/ int zfs_umount(struct super_block *sb) { zfsvfs_t *zfsvfs = sb->s_fs_info; objset_t *os; if (zfsvfs->z_arc_prune != NULL) arc_remove_prune_callback(zfsvfs->z_arc_prune); VERIFY(zfsvfs_teardown(zfsvfs, B_TRUE) == 0); os = zfsvfs->z_os; zpl_bdi_destroy(sb); /* * z_os will be NULL if there was an error in * attempting to reopen zfsvfs. */ if (os != NULL) { /* * Unset the objset user_ptr. */ mutex_enter(&os->os_user_ptr_lock); dmu_objset_set_user(os, NULL); mutex_exit(&os->os_user_ptr_lock); /* * Finally release the objset */ dmu_objset_disown(os, B_TRUE, zfsvfs); } zfsvfs_free(zfsvfs); return (0); } int zfs_remount(struct super_block *sb, int *flags, zfs_mnt_t *zm) { zfsvfs_t *zfsvfs = sb->s_fs_info; vfs_t *vfsp; boolean_t issnap = dmu_objset_is_snapshot(zfsvfs->z_os); int error; if ((issnap || !spa_writeable(dmu_objset_spa(zfsvfs->z_os))) && !(*flags & SB_RDONLY)) { *flags |= SB_RDONLY; return (EROFS); } error = zfsvfs_parse_options(zm->mnt_data, &vfsp); if (error) return (error); if (!zfs_is_readonly(zfsvfs) && (*flags & SB_RDONLY)) txg_wait_synced(dmu_objset_pool(zfsvfs->z_os), 0); zfs_unregister_callbacks(zfsvfs); zfsvfs_vfs_free(zfsvfs->z_vfs); vfsp->vfs_data = zfsvfs; zfsvfs->z_vfs = vfsp; if (!issnap) (void) zfs_register_callbacks(vfsp); return (error); } int zfs_vget(struct super_block *sb, struct inode **ipp, fid_t *fidp) { zfsvfs_t *zfsvfs = sb->s_fs_info; znode_t *zp; uint64_t object = 0; uint64_t fid_gen = 0; uint64_t gen_mask; uint64_t zp_gen; int i, err; *ipp = NULL; if (fidp->fid_len == SHORT_FID_LEN || fidp->fid_len == LONG_FID_LEN) { zfid_short_t *zfid = (zfid_short_t *)fidp; for (i = 0; i < sizeof (zfid->zf_object); i++) object |= ((uint64_t)zfid->zf_object[i]) << (8 * i); for (i = 0; i < sizeof (zfid->zf_gen); i++) fid_gen |= ((uint64_t)zfid->zf_gen[i]) << (8 * i); } else { return (SET_ERROR(EINVAL)); } /* LONG_FID_LEN means snapdirs */ if (fidp->fid_len == LONG_FID_LEN) { zfid_long_t *zlfid = (zfid_long_t *)fidp; uint64_t objsetid = 0; uint64_t setgen = 0; for (i = 0; i < sizeof (zlfid->zf_setid); i++) objsetid |= ((uint64_t)zlfid->zf_setid[i]) << (8 * i); for (i = 0; i < sizeof (zlfid->zf_setgen); i++) setgen |= ((uint64_t)zlfid->zf_setgen[i]) << (8 * i); if (objsetid != ZFSCTL_INO_SNAPDIRS - object) { dprintf("snapdir fid: objsetid (%llu) != " "ZFSCTL_INO_SNAPDIRS (%llu) - object (%llu)\n", objsetid, ZFSCTL_INO_SNAPDIRS, object); return (SET_ERROR(EINVAL)); } if (fid_gen > 1 || setgen != 0) { dprintf("snapdir fid: fid_gen (%llu) and setgen " "(%llu)\n", fid_gen, setgen); return (SET_ERROR(EINVAL)); } return (zfsctl_snapdir_vget(sb, objsetid, fid_gen, ipp)); } ZFS_ENTER(zfsvfs); /* A zero fid_gen means we are in the .zfs control directories */ if (fid_gen == 0 && (object == ZFSCTL_INO_ROOT || object == ZFSCTL_INO_SNAPDIR)) { *ipp = zfsvfs->z_ctldir; ASSERT(*ipp != NULL); if (object == ZFSCTL_INO_SNAPDIR) { VERIFY(zfsctl_root_lookup(*ipp, "snapshot", ipp, 0, kcred, NULL, NULL) == 0); } else { igrab(*ipp); } ZFS_EXIT(zfsvfs); return (0); } gen_mask = -1ULL >> (64 - 8 * i); dprintf("getting %llu [%llu mask %llx]\n", object, fid_gen, gen_mask); if ((err = zfs_zget(zfsvfs, object, &zp))) { ZFS_EXIT(zfsvfs); return (err); } /* Don't export xattr stuff */ if (zp->z_pflags & ZFS_XATTR) { zrele(zp); ZFS_EXIT(zfsvfs); return (SET_ERROR(ENOENT)); } (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zfsvfs), &zp_gen, sizeof (uint64_t)); zp_gen = zp_gen & gen_mask; if (zp_gen == 0) zp_gen = 1; if ((fid_gen == 0) && (zfsvfs->z_root == object)) fid_gen = zp_gen; if (zp->z_unlinked || zp_gen != fid_gen) { dprintf("znode gen (%llu) != fid gen (%llu)\n", zp_gen, fid_gen); zrele(zp); ZFS_EXIT(zfsvfs); return (SET_ERROR(ENOENT)); } *ipp = ZTOI(zp); if (*ipp) zfs_inode_update(ITOZ(*ipp)); ZFS_EXIT(zfsvfs); return (0); } /* * Block out VFS ops and close zfsvfs_t * * Note, if successful, then we return with the 'z_teardown_lock' and * 'z_teardown_inactive_lock' write held. We leave ownership of the underlying * dataset and objset intact so that they can be atomically handed off during * a subsequent rollback or recv operation and the resume thereafter. */ int zfs_suspend_fs(zfsvfs_t *zfsvfs) { int error; if ((error = zfsvfs_teardown(zfsvfs, B_FALSE)) != 0) return (error); return (0); } /* * Rebuild SA and release VOPs. Note that ownership of the underlying dataset * is an invariant across any of the operations that can be performed while the * filesystem was suspended. Whether it succeeded or failed, the preconditions * are the same: the relevant objset and associated dataset are owned by * zfsvfs, held, and long held on entry. */ int zfs_resume_fs(zfsvfs_t *zfsvfs, dsl_dataset_t *ds) { int err, err2; znode_t *zp; ASSERT(RRM_WRITE_HELD(&zfsvfs->z_teardown_lock)); ASSERT(RW_WRITE_HELD(&zfsvfs->z_teardown_inactive_lock)); /* * We already own this, so just update the objset_t, as the one we * had before may have been evicted. */ objset_t *os; VERIFY3P(ds->ds_owner, ==, zfsvfs); VERIFY(dsl_dataset_long_held(ds)); dsl_pool_t *dp = spa_get_dsl(dsl_dataset_get_spa(ds)); dsl_pool_config_enter(dp, FTAG); VERIFY0(dmu_objset_from_ds(ds, &os)); dsl_pool_config_exit(dp, FTAG); err = zfsvfs_init(zfsvfs, os); if (err != 0) goto bail; ds->ds_dir->dd_activity_cancelled = B_FALSE; VERIFY(zfsvfs_setup(zfsvfs, B_FALSE) == 0); zfs_set_fuid_feature(zfsvfs); zfsvfs->z_rollback_time = jiffies; /* * Attempt to re-establish all the active inodes with their * dbufs. If a zfs_rezget() fails, then we unhash the inode * and mark it stale. This prevents a collision if a new * inode/object is created which must use the same inode * number. The stale inode will be be released when the * VFS prunes the dentry holding the remaining references * on the stale inode. */ mutex_enter(&zfsvfs->z_znodes_lock); for (zp = list_head(&zfsvfs->z_all_znodes); zp; zp = list_next(&zfsvfs->z_all_znodes, zp)) { err2 = zfs_rezget(zp); if (err2) { remove_inode_hash(ZTOI(zp)); zp->z_is_stale = B_TRUE; } /* see comment in zfs_suspend_fs() */ if (zp->z_suspended) { zfs_zrele_async(zp); zp->z_suspended = B_FALSE; } } mutex_exit(&zfsvfs->z_znodes_lock); if (!zfs_is_readonly(zfsvfs) && !zfsvfs->z_unmounted) { /* * zfs_suspend_fs() could have interrupted freeing * of dnodes. We need to restart this freeing so * that we don't "leak" the space. */ zfs_unlinked_drain(zfsvfs); } /* * Most of the time zfs_suspend_fs is used for changing the contents * of the underlying dataset. ZFS rollback and receive operations * might create files for which negative dentries are present in * the cache. Since walking the dcache would require a lot of GPL-only * code duplication, it's much easier on these rather rare occasions * just to flush the whole dcache for the given dataset/filesystem. */ shrink_dcache_sb(zfsvfs->z_sb); bail: if (err != 0) zfsvfs->z_unmounted = B_TRUE; /* release the VFS ops */ rw_exit(&zfsvfs->z_teardown_inactive_lock); rrm_exit(&zfsvfs->z_teardown_lock, FTAG); if (err != 0) { /* * Since we couldn't setup the sa framework, try to force * unmount this file system. */ if (zfsvfs->z_os) (void) zfs_umount(zfsvfs->z_sb); } return (err); } /* * Release VOPs and unmount a suspended filesystem. */ int zfs_end_fs(zfsvfs_t *zfsvfs, dsl_dataset_t *ds) { ASSERT(RRM_WRITE_HELD(&zfsvfs->z_teardown_lock)); ASSERT(RW_WRITE_HELD(&zfsvfs->z_teardown_inactive_lock)); /* * We already own this, so just hold and rele it to update the * objset_t, as the one we had before may have been evicted. */ objset_t *os; VERIFY3P(ds->ds_owner, ==, zfsvfs); VERIFY(dsl_dataset_long_held(ds)); dsl_pool_t *dp = spa_get_dsl(dsl_dataset_get_spa(ds)); dsl_pool_config_enter(dp, FTAG); VERIFY0(dmu_objset_from_ds(ds, &os)); dsl_pool_config_exit(dp, FTAG); zfsvfs->z_os = os; /* release the VOPs */ rw_exit(&zfsvfs->z_teardown_inactive_lock); rrm_exit(&zfsvfs->z_teardown_lock, FTAG); /* * Try to force unmount this file system. */ (void) zfs_umount(zfsvfs->z_sb); zfsvfs->z_unmounted = B_TRUE; return (0); } /* * Automounted snapshots rely on periodic revalidation * to defer snapshots from being automatically unmounted. */ inline void zfs_exit_fs(zfsvfs_t *zfsvfs) { if (!zfsvfs->z_issnap) return; if (time_after(jiffies, zfsvfs->z_snap_defer_time + MAX(zfs_expire_snapshot * HZ / 2, HZ))) { zfsvfs->z_snap_defer_time = jiffies; zfsctl_snapshot_unmount_delay(zfsvfs->z_os->os_spa, dmu_objset_id(zfsvfs->z_os), zfs_expire_snapshot); } } int zfs_set_version(zfsvfs_t *zfsvfs, uint64_t newvers) { int error; objset_t *os = zfsvfs->z_os; dmu_tx_t *tx; if (newvers < ZPL_VERSION_INITIAL || newvers > ZPL_VERSION) return (SET_ERROR(EINVAL)); if (newvers < zfsvfs->z_version) return (SET_ERROR(EINVAL)); if (zfs_spa_version_map(newvers) > spa_version(dmu_objset_spa(zfsvfs->z_os))) return (SET_ERROR(ENOTSUP)); tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, B_FALSE, ZPL_VERSION_STR); if (newvers >= ZPL_VERSION_SA && !zfsvfs->z_use_sa) { dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, B_TRUE, ZFS_SA_ATTRS); dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL); } error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); return (error); } error = zap_update(os, MASTER_NODE_OBJ, ZPL_VERSION_STR, 8, 1, &newvers, tx); if (error) { dmu_tx_commit(tx); return (error); } if (newvers >= ZPL_VERSION_SA && !zfsvfs->z_use_sa) { uint64_t sa_obj; ASSERT3U(spa_version(dmu_objset_spa(zfsvfs->z_os)), >=, SPA_VERSION_SA); sa_obj = zap_create(os, DMU_OT_SA_MASTER_NODE, DMU_OT_NONE, 0, tx); error = zap_add(os, MASTER_NODE_OBJ, ZFS_SA_ATTRS, 8, 1, &sa_obj, tx); ASSERT0(error); VERIFY(0 == sa_set_sa_object(os, sa_obj)); sa_register_update_callback(os, zfs_sa_upgrade); } spa_history_log_internal_ds(dmu_objset_ds(os), "upgrade", tx, "from %llu to %llu", zfsvfs->z_version, newvers); dmu_tx_commit(tx); zfsvfs->z_version = newvers; os->os_version = newvers; zfs_set_fuid_feature(zfsvfs); return (0); } /* * Read a property stored within the master node. */ int zfs_get_zplprop(objset_t *os, zfs_prop_t prop, uint64_t *value) { uint64_t *cached_copy = NULL; /* * Figure out where in the objset_t the cached copy would live, if it * is available for the requested property. */ if (os != NULL) { switch (prop) { case ZFS_PROP_VERSION: cached_copy = &os->os_version; break; case ZFS_PROP_NORMALIZE: cached_copy = &os->os_normalization; break; case ZFS_PROP_UTF8ONLY: cached_copy = &os->os_utf8only; break; case ZFS_PROP_CASE: cached_copy = &os->os_casesensitivity; break; default: break; } } if (cached_copy != NULL && *cached_copy != OBJSET_PROP_UNINITIALIZED) { *value = *cached_copy; return (0); } /* * If the property wasn't cached, look up the file system's value for * the property. For the version property, we look up a slightly * different string. */ const char *pname; int error = ENOENT; if (prop == ZFS_PROP_VERSION) pname = ZPL_VERSION_STR; else pname = zfs_prop_to_name(prop); if (os != NULL) { ASSERT3U(os->os_phys->os_type, ==, DMU_OST_ZFS); error = zap_lookup(os, MASTER_NODE_OBJ, pname, 8, 1, value); } if (error == ENOENT) { /* No value set, use the default value */ switch (prop) { case ZFS_PROP_VERSION: *value = ZPL_VERSION; break; case ZFS_PROP_NORMALIZE: case ZFS_PROP_UTF8ONLY: *value = 0; break; case ZFS_PROP_CASE: *value = ZFS_CASE_SENSITIVE; break; case ZFS_PROP_ACLTYPE: *value = ZFS_ACLTYPE_OFF; break; default: return (error); } error = 0; } /* * If one of the methods for getting the property value above worked, * copy it into the objset_t's cache. */ if (error == 0 && cached_copy != NULL) { *cached_copy = *value; } return (error); } /* * Return true if the corresponding vfs's unmounted flag is set. * Otherwise return false. * If this function returns true we know VFS unmount has been initiated. */ boolean_t zfs_get_vfs_flag_unmounted(objset_t *os) { zfsvfs_t *zfvp; boolean_t unmounted = B_FALSE; ASSERT(dmu_objset_type(os) == DMU_OST_ZFS); mutex_enter(&os->os_user_ptr_lock); zfvp = dmu_objset_get_user(os); if (zfvp != NULL && zfvp->z_unmounted) unmounted = B_TRUE; mutex_exit(&os->os_user_ptr_lock); return (unmounted); } /*ARGSUSED*/ void zfsvfs_update_fromname(const char *oldname, const char *newname) { /* * We don't need to do anything here, the devname is always current by * virtue of zfsvfs->z_sb->s_op->show_devname. */ } void zfs_init(void) { zfsctl_init(); zfs_znode_init(); dmu_objset_register_type(DMU_OST_ZFS, zpl_get_file_info); register_filesystem(&zpl_fs_type); } void zfs_fini(void) { /* * we don't use outstanding because zpl_posix_acl_free might add more. */ taskq_wait(system_delay_taskq); taskq_wait(system_taskq); unregister_filesystem(&zpl_fs_type); zfs_znode_fini(); zfsctl_fini(); } #if defined(_KERNEL) EXPORT_SYMBOL(zfs_suspend_fs); EXPORT_SYMBOL(zfs_resume_fs); EXPORT_SYMBOL(zfs_set_version); EXPORT_SYMBOL(zfsvfs_create); EXPORT_SYMBOL(zfsvfs_free); EXPORT_SYMBOL(zfs_is_readonly); EXPORT_SYMBOL(zfs_domount); EXPORT_SYMBOL(zfs_preumount); EXPORT_SYMBOL(zfs_umount); EXPORT_SYMBOL(zfs_remount); EXPORT_SYMBOL(zfs_statvfs); EXPORT_SYMBOL(zfs_vget); EXPORT_SYMBOL(zfs_prune); #endif diff --git a/module/zcommon/zfs_prop.c b/module/zcommon/zfs_prop.c index 0352b13aa240..89ddc59b226c 100644 --- a/module/zcommon/zfs_prop.c +++ b/module/zcommon/zfs_prop.c @@ -1,1052 +1,1052 @@ /* * 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) 2011, 2018 by Delphix. All rights reserved. * Copyright (c) 2013 by Saso Kiselkov. All rights reserved. * Copyright 2016, Joyent, Inc. * Copyright (c) 2019, Klara Inc. * Copyright (c) 2019, Allan Jude */ /* Portions Copyright 2010 Robert Milkowski */ #include #include #include #include #include #include #include #include "zfs_prop.h" #include "zfs_deleg.h" #include "zfs_fletcher.h" #if !defined(_KERNEL) #include #include #include #endif static zprop_desc_t zfs_prop_table[ZFS_NUM_PROPS]; /* Note this is indexed by zfs_userquota_prop_t, keep the order the same */ const char *zfs_userquota_prop_prefixes[] = { "userused@", "userquota@", "groupused@", "groupquota@", "userobjused@", "userobjquota@", "groupobjused@", "groupobjquota@", "projectused@", "projectquota@", "projectobjused@", "projectobjquota@" }; zprop_desc_t * zfs_prop_get_table(void) { return (zfs_prop_table); } void zfs_prop_init(void) { static zprop_index_t checksum_table[] = { { "on", ZIO_CHECKSUM_ON }, { "off", ZIO_CHECKSUM_OFF }, { "fletcher2", ZIO_CHECKSUM_FLETCHER_2 }, { "fletcher4", ZIO_CHECKSUM_FLETCHER_4 }, { "sha256", ZIO_CHECKSUM_SHA256 }, { "noparity", ZIO_CHECKSUM_NOPARITY }, { "sha512", ZIO_CHECKSUM_SHA512 }, { "skein", ZIO_CHECKSUM_SKEIN }, #if !defined(__FreeBSD__) { "edonr", ZIO_CHECKSUM_EDONR }, #endif { NULL } }; static zprop_index_t dedup_table[] = { { "on", ZIO_CHECKSUM_ON }, { "off", ZIO_CHECKSUM_OFF }, { "verify", ZIO_CHECKSUM_ON | ZIO_CHECKSUM_VERIFY }, { "sha256", ZIO_CHECKSUM_SHA256 }, { "sha256,verify", ZIO_CHECKSUM_SHA256 | ZIO_CHECKSUM_VERIFY }, { "sha512", ZIO_CHECKSUM_SHA512 }, { "sha512,verify", ZIO_CHECKSUM_SHA512 | ZIO_CHECKSUM_VERIFY }, { "skein", ZIO_CHECKSUM_SKEIN }, { "skein,verify", ZIO_CHECKSUM_SKEIN | ZIO_CHECKSUM_VERIFY }, #if !defined(__FreeBSD__) { "edonr,verify", ZIO_CHECKSUM_EDONR | ZIO_CHECKSUM_VERIFY }, #endif { NULL } }; static zprop_index_t compress_table[] = { { "on", ZIO_COMPRESS_ON }, { "off", ZIO_COMPRESS_OFF }, { "lzjb", ZIO_COMPRESS_LZJB }, { "gzip", ZIO_COMPRESS_GZIP_6 }, /* gzip default */ { "gzip-1", ZIO_COMPRESS_GZIP_1 }, { "gzip-2", ZIO_COMPRESS_GZIP_2 }, { "gzip-3", ZIO_COMPRESS_GZIP_3 }, { "gzip-4", ZIO_COMPRESS_GZIP_4 }, { "gzip-5", ZIO_COMPRESS_GZIP_5 }, { "gzip-6", ZIO_COMPRESS_GZIP_6 }, { "gzip-7", ZIO_COMPRESS_GZIP_7 }, { "gzip-8", ZIO_COMPRESS_GZIP_8 }, { "gzip-9", ZIO_COMPRESS_GZIP_9 }, { "zle", ZIO_COMPRESS_ZLE }, { "lz4", ZIO_COMPRESS_LZ4 }, { "zstd", ZIO_COMPRESS_ZSTD }, { "zstd-fast", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_DEFAULT) }, /* * ZSTD 1-19 are synthetic. We store the compression level in a * separate hidden property to avoid wasting a large amount of * space in the ZIO_COMPRESS enum. * * The compression level is also stored within the header of the * compressed block since we may need it for later recompression * to avoid checksum errors (L2ARC). * * Note that the level here is defined as bit shifted mask on * top of the method. */ { "zstd-1", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_1) }, { "zstd-2", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_2) }, { "zstd-3", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_3) }, { "zstd-4", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_4) }, { "zstd-5", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_5) }, { "zstd-6", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_6) }, { "zstd-7", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_7) }, { "zstd-8", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_8) }, { "zstd-9", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_9) }, { "zstd-10", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_10) }, { "zstd-11", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_11) }, { "zstd-12", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_12) }, { "zstd-13", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_13) }, { "zstd-14", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_14) }, { "zstd-15", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_15) }, { "zstd-16", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_16) }, { "zstd-17", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_17) }, { "zstd-18", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_18) }, { "zstd-19", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_19) }, /* * The ZSTD-Fast levels are also synthetic. */ { "zstd-fast-1", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_1) }, { "zstd-fast-2", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_2) }, { "zstd-fast-3", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_3) }, { "zstd-fast-4", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_4) }, { "zstd-fast-5", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_5) }, { "zstd-fast-6", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_6) }, { "zstd-fast-7", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_7) }, { "zstd-fast-8", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_8) }, { "zstd-fast-9", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_9) }, { "zstd-fast-10", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_10) }, { "zstd-fast-20", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_20) }, { "zstd-fast-30", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_30) }, { "zstd-fast-40", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_40) }, { "zstd-fast-50", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_50) }, { "zstd-fast-60", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_60) }, { "zstd-fast-70", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_70) }, { "zstd-fast-80", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_80) }, { "zstd-fast-90", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_90) }, { "zstd-fast-100", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_100) }, { "zstd-fast-500", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_500) }, { "zstd-fast-1000", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_1000) }, { NULL } }; static zprop_index_t crypto_table[] = { { "on", ZIO_CRYPT_ON }, { "off", ZIO_CRYPT_OFF }, { "aes-128-ccm", ZIO_CRYPT_AES_128_CCM }, { "aes-192-ccm", ZIO_CRYPT_AES_192_CCM }, { "aes-256-ccm", ZIO_CRYPT_AES_256_CCM }, { "aes-128-gcm", ZIO_CRYPT_AES_128_GCM }, { "aes-192-gcm", ZIO_CRYPT_AES_192_GCM }, { "aes-256-gcm", ZIO_CRYPT_AES_256_GCM }, { NULL } }; static zprop_index_t keyformat_table[] = { { "none", ZFS_KEYFORMAT_NONE }, { "raw", ZFS_KEYFORMAT_RAW }, { "hex", ZFS_KEYFORMAT_HEX }, { "passphrase", ZFS_KEYFORMAT_PASSPHRASE }, { NULL } }; static zprop_index_t snapdir_table[] = { { "hidden", ZFS_SNAPDIR_HIDDEN }, { "visible", ZFS_SNAPDIR_VISIBLE }, { NULL } }; static zprop_index_t snapdev_table[] = { { "hidden", ZFS_SNAPDEV_HIDDEN }, { "visible", ZFS_SNAPDEV_VISIBLE }, { NULL } }; static zprop_index_t acl_mode_table[] = { { "discard", ZFS_ACL_DISCARD }, { "groupmask", ZFS_ACL_GROUPMASK }, { "passthrough", ZFS_ACL_PASSTHROUGH }, { "restricted", ZFS_ACL_RESTRICTED }, { NULL } }; static zprop_index_t acltype_table[] = { { "off", ZFS_ACLTYPE_OFF }, { "posix", ZFS_ACLTYPE_POSIX }, { "nfsv4", ZFS_ACLTYPE_NFSV4 }, { "disabled", ZFS_ACLTYPE_OFF }, /* bkwrd compatibility */ { "noacl", ZFS_ACLTYPE_OFF }, /* bkwrd compatibility */ { "posixacl", ZFS_ACLTYPE_POSIX }, /* bkwrd compatibility */ { NULL } }; static zprop_index_t acl_inherit_table[] = { { "discard", ZFS_ACL_DISCARD }, { "noallow", ZFS_ACL_NOALLOW }, { "restricted", ZFS_ACL_RESTRICTED }, { "passthrough", ZFS_ACL_PASSTHROUGH }, { "secure", ZFS_ACL_RESTRICTED }, /* bkwrd compatibility */ { "passthrough-x", ZFS_ACL_PASSTHROUGH_X }, { NULL } }; static zprop_index_t case_table[] = { { "sensitive", ZFS_CASE_SENSITIVE }, { "insensitive", ZFS_CASE_INSENSITIVE }, { "mixed", ZFS_CASE_MIXED }, { NULL } }; static zprop_index_t copies_table[] = { { "1", 1 }, { "2", 2 }, { "3", 3 }, { NULL } }; /* * Use the unique flags we have to send to u8_strcmp() and/or * u8_textprep() to represent the various normalization property * values. */ static zprop_index_t normalize_table[] = { { "none", 0 }, { "formD", U8_TEXTPREP_NFD }, { "formKC", U8_TEXTPREP_NFKC }, { "formC", U8_TEXTPREP_NFC }, { "formKD", U8_TEXTPREP_NFKD }, { NULL } }; static zprop_index_t version_table[] = { { "1", 1 }, { "2", 2 }, { "3", 3 }, { "4", 4 }, { "5", 5 }, { "current", ZPL_VERSION }, { NULL } }; static zprop_index_t boolean_table[] = { { "off", 0 }, { "on", 1 }, { NULL } }; static zprop_index_t keystatus_table[] = { { "none", ZFS_KEYSTATUS_NONE}, { "unavailable", ZFS_KEYSTATUS_UNAVAILABLE}, { "available", ZFS_KEYSTATUS_AVAILABLE}, { NULL } }; static zprop_index_t logbias_table[] = { { "latency", ZFS_LOGBIAS_LATENCY }, { "throughput", ZFS_LOGBIAS_THROUGHPUT }, { NULL } }; static zprop_index_t canmount_table[] = { { "off", ZFS_CANMOUNT_OFF }, { "on", ZFS_CANMOUNT_ON }, { "noauto", ZFS_CANMOUNT_NOAUTO }, { NULL } }; static zprop_index_t cache_table[] = { { "none", ZFS_CACHE_NONE }, { "metadata", ZFS_CACHE_METADATA }, { "all", ZFS_CACHE_ALL }, { NULL } }; static zprop_index_t sync_table[] = { { "standard", ZFS_SYNC_STANDARD }, { "always", ZFS_SYNC_ALWAYS }, { "disabled", ZFS_SYNC_DISABLED }, { NULL } }; static zprop_index_t xattr_table[] = { { "off", ZFS_XATTR_OFF }, { "on", ZFS_XATTR_DIR }, { "sa", ZFS_XATTR_SA }, { "dir", ZFS_XATTR_DIR }, { NULL } }; static zprop_index_t dnsize_table[] = { { "legacy", ZFS_DNSIZE_LEGACY }, { "auto", ZFS_DNSIZE_AUTO }, { "1k", ZFS_DNSIZE_1K }, { "2k", ZFS_DNSIZE_2K }, { "4k", ZFS_DNSIZE_4K }, { "8k", ZFS_DNSIZE_8K }, { "16k", ZFS_DNSIZE_16K }, { NULL } }; static zprop_index_t redundant_metadata_table[] = { { "all", ZFS_REDUNDANT_METADATA_ALL }, { "most", ZFS_REDUNDANT_METADATA_MOST }, { NULL } }; static zprop_index_t volmode_table[] = { { "default", ZFS_VOLMODE_DEFAULT }, { "full", ZFS_VOLMODE_GEOM }, { "geom", ZFS_VOLMODE_GEOM }, { "dev", ZFS_VOLMODE_DEV }, { "none", ZFS_VOLMODE_NONE }, { NULL } }; /* inherit index properties */ zprop_register_index(ZFS_PROP_REDUNDANT_METADATA, "redundant_metadata", ZFS_REDUNDANT_METADATA_ALL, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "all | most", "REDUND_MD", redundant_metadata_table); zprop_register_index(ZFS_PROP_SYNC, "sync", ZFS_SYNC_STANDARD, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "standard | always | disabled", "SYNC", sync_table); zprop_register_index(ZFS_PROP_CHECKSUM, "checksum", ZIO_CHECKSUM_DEFAULT, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, #if !defined(__FreeBSD__) "on | off | fletcher2 | fletcher4 | sha256 | sha512 | skein" " | edonr", #else "on | off | fletcher2 | fletcher4 | sha256 | sha512 | skein", #endif "CHECKSUM", checksum_table); zprop_register_index(ZFS_PROP_DEDUP, "dedup", ZIO_CHECKSUM_OFF, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "on | off | verify | sha256[,verify] | sha512[,verify] | " #if !defined(__FreeBSD__) "skein[,verify] | edonr,verify", #else "skein[,verify]", #endif "DEDUP", dedup_table); zprop_register_index(ZFS_PROP_COMPRESSION, "compression", ZIO_COMPRESS_DEFAULT, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "on | off | lzjb | gzip | gzip-[1-9] | zle | lz4 | " "zstd | zstd-[1-19] | " "zstd-fast-[1-10,20,30,40,50,60,70,80,90,100,500,1000]", "COMPRESS", compress_table); zprop_register_index(ZFS_PROP_SNAPDIR, "snapdir", ZFS_SNAPDIR_HIDDEN, PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "hidden | visible", "SNAPDIR", snapdir_table); zprop_register_index(ZFS_PROP_SNAPDEV, "snapdev", ZFS_SNAPDEV_HIDDEN, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "hidden | visible", "SNAPDEV", snapdev_table); zprop_register_index(ZFS_PROP_ACLMODE, "aclmode", ZFS_ACL_DISCARD, PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "discard | groupmask | passthrough | restricted", "ACLMODE", acl_mode_table); zprop_register_index(ZFS_PROP_ACLTYPE, "acltype", #ifdef __linux__ /* Linux doesn't natively support ZFS's NFSv4-style ACLs. */ ZFS_ACLTYPE_OFF, #else ZFS_ACLTYPE_NFSV4, #endif PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT, "off | nfsv4 | posix", "ACLTYPE", acltype_table); zprop_register_index(ZFS_PROP_ACLINHERIT, "aclinherit", ZFS_ACL_RESTRICTED, PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "discard | noallow | restricted | passthrough | passthrough-x", "ACLINHERIT", acl_inherit_table); zprop_register_index(ZFS_PROP_COPIES, "copies", 1, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "1 | 2 | 3", "COPIES", copies_table); zprop_register_index(ZFS_PROP_PRIMARYCACHE, "primarycache", ZFS_CACHE_ALL, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT | ZFS_TYPE_VOLUME, "all | none | metadata", "PRIMARYCACHE", cache_table); zprop_register_index(ZFS_PROP_SECONDARYCACHE, "secondarycache", ZFS_CACHE_ALL, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT | ZFS_TYPE_VOLUME, "all | none | metadata", "SECONDARYCACHE", cache_table); zprop_register_index(ZFS_PROP_LOGBIAS, "logbias", ZFS_LOGBIAS_LATENCY, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "latency | throughput", "LOGBIAS", logbias_table); zprop_register_index(ZFS_PROP_XATTR, "xattr", ZFS_XATTR_DIR, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT, "on | off | dir | sa", "XATTR", xattr_table); zprop_register_index(ZFS_PROP_DNODESIZE, "dnodesize", ZFS_DNSIZE_LEGACY, PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "legacy | auto | 1k | 2k | 4k | 8k | 16k", "DNSIZE", dnsize_table); zprop_register_index(ZFS_PROP_VOLMODE, "volmode", ZFS_VOLMODE_DEFAULT, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "default | full | geom | dev | none", "VOLMODE", volmode_table); /* inherit index (boolean) properties */ zprop_register_index(ZFS_PROP_ATIME, "atime", 1, PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "on | off", "ATIME", boolean_table); zprop_register_index(ZFS_PROP_RELATIME, "relatime", 0, PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "on | off", "RELATIME", boolean_table); zprop_register_index(ZFS_PROP_DEVICES, "devices", 1, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT, "on | off", "DEVICES", boolean_table); zprop_register_index(ZFS_PROP_EXEC, "exec", 1, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT, "on | off", "EXEC", boolean_table); zprop_register_index(ZFS_PROP_SETUID, "setuid", 1, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT, "on | off", "SETUID", boolean_table); zprop_register_index(ZFS_PROP_READONLY, "readonly", 0, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "on | off", "RDONLY", boolean_table); #ifdef __FreeBSD__ zprop_register_index(ZFS_PROP_ZONED, "jailed", 0, PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "on | off", "JAILED", boolean_table); #else zprop_register_index(ZFS_PROP_ZONED, "zoned", 0, PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "on | off", "ZONED", boolean_table); #endif zprop_register_index(ZFS_PROP_VSCAN, "vscan", 0, PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "on | off", "VSCAN", boolean_table); zprop_register_index(ZFS_PROP_NBMAND, "nbmand", 0, PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT, "on | off", "NBMAND", boolean_table); zprop_register_index(ZFS_PROP_OVERLAY, "overlay", 1, PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "on | off", "OVERLAY", boolean_table); /* default index properties */ zprop_register_index(ZFS_PROP_VERSION, "version", 0, PROP_DEFAULT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT, "1 | 2 | 3 | 4 | 5 | current", "VERSION", version_table); zprop_register_index(ZFS_PROP_CANMOUNT, "canmount", ZFS_CANMOUNT_ON, PROP_DEFAULT, ZFS_TYPE_FILESYSTEM, "on | off | noauto", "CANMOUNT", canmount_table); /* readonly index properties */ zprop_register_index(ZFS_PROP_MOUNTED, "mounted", 0, PROP_READONLY, ZFS_TYPE_FILESYSTEM, "yes | no", "MOUNTED", boolean_table); zprop_register_index(ZFS_PROP_DEFER_DESTROY, "defer_destroy", 0, PROP_READONLY, ZFS_TYPE_SNAPSHOT, "yes | no", "DEFER_DESTROY", boolean_table); zprop_register_index(ZFS_PROP_KEYSTATUS, "keystatus", ZFS_KEYSTATUS_NONE, PROP_READONLY, ZFS_TYPE_DATASET, "none | unavailable | available", "KEYSTATUS", keystatus_table); /* set once index properties */ zprop_register_index(ZFS_PROP_NORMALIZE, "normalization", 0, PROP_ONETIME, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT, "none | formC | formD | formKC | formKD", "NORMALIZATION", normalize_table); zprop_register_index(ZFS_PROP_CASE, "casesensitivity", ZFS_CASE_SENSITIVE, PROP_ONETIME, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT, "sensitive | insensitive | mixed", "CASE", case_table); zprop_register_index(ZFS_PROP_KEYFORMAT, "keyformat", ZFS_KEYFORMAT_NONE, PROP_ONETIME_DEFAULT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "none | raw | hex | passphrase", "KEYFORMAT", keyformat_table); zprop_register_index(ZFS_PROP_ENCRYPTION, "encryption", ZIO_CRYPT_DEFAULT, PROP_ONETIME, ZFS_TYPE_DATASET, "on | off | aes-128-ccm | aes-192-ccm | aes-256-ccm | " "aes-128-gcm | aes-192-gcm | aes-256-gcm", "ENCRYPTION", crypto_table); /* set once index (boolean) properties */ zprop_register_index(ZFS_PROP_UTF8ONLY, "utf8only", 0, PROP_ONETIME, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT, "on | off", "UTF8ONLY", boolean_table); /* string properties */ zprop_register_string(ZFS_PROP_ORIGIN, "origin", NULL, PROP_READONLY, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "", "ORIGIN"); zprop_register_string(ZFS_PROP_CLONES, "clones", NULL, PROP_READONLY, ZFS_TYPE_SNAPSHOT, "[,...]", "CLONES"); zprop_register_string(ZFS_PROP_MOUNTPOINT, "mountpoint", "/", PROP_INHERIT, ZFS_TYPE_FILESYSTEM, " | legacy | none", "MOUNTPOINT"); zprop_register_string(ZFS_PROP_SHARENFS, "sharenfs", "off", - PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "on | off | share(1M) options", + PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "on | off | NFS share options", "SHARENFS"); zprop_register_string(ZFS_PROP_TYPE, "type", NULL, PROP_READONLY, ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK, "filesystem | volume | snapshot | bookmark", "TYPE"); zprop_register_string(ZFS_PROP_SHARESMB, "sharesmb", "off", PROP_INHERIT, ZFS_TYPE_FILESYSTEM, - "on | off | sharemgr(1M) options", "SHARESMB"); + "on | off | SMB share options", "SHARESMB"); zprop_register_string(ZFS_PROP_MLSLABEL, "mlslabel", ZFS_MLSLABEL_DEFAULT, PROP_INHERIT, ZFS_TYPE_DATASET, "", "MLSLABEL"); zprop_register_string(ZFS_PROP_SELINUX_CONTEXT, "context", "none", PROP_DEFAULT, ZFS_TYPE_DATASET, "", "CONTEXT"); zprop_register_string(ZFS_PROP_SELINUX_FSCONTEXT, "fscontext", "none", PROP_DEFAULT, ZFS_TYPE_DATASET, "", "FSCONTEXT"); zprop_register_string(ZFS_PROP_SELINUX_DEFCONTEXT, "defcontext", "none", PROP_DEFAULT, ZFS_TYPE_DATASET, "", "DEFCONTEXT"); zprop_register_string(ZFS_PROP_SELINUX_ROOTCONTEXT, "rootcontext", "none", PROP_DEFAULT, ZFS_TYPE_DATASET, "", "ROOTCONTEXT"); zprop_register_string(ZFS_PROP_RECEIVE_RESUME_TOKEN, "receive_resume_token", NULL, PROP_READONLY, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "", "RESUMETOK"); zprop_register_string(ZFS_PROP_ENCRYPTION_ROOT, "encryptionroot", NULL, PROP_READONLY, ZFS_TYPE_DATASET, "", "ENCROOT"); zprop_register_string(ZFS_PROP_KEYLOCATION, "keylocation", "none", PROP_DEFAULT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "prompt | ", "KEYLOCATION"); zprop_register_string(ZFS_PROP_REDACT_SNAPS, "redact_snaps", NULL, PROP_READONLY, ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK, "[,...]", "RSNAPS"); /* readonly number properties */ zprop_register_number(ZFS_PROP_USED, "used", 0, PROP_READONLY, ZFS_TYPE_DATASET, "", "USED"); zprop_register_number(ZFS_PROP_AVAILABLE, "available", 0, PROP_READONLY, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "", "AVAIL"); zprop_register_number(ZFS_PROP_REFERENCED, "referenced", 0, PROP_READONLY, ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK, "", "REFER"); zprop_register_number(ZFS_PROP_COMPRESSRATIO, "compressratio", 0, PROP_READONLY, ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK, "<1.00x or higher if compressed>", "RATIO"); zprop_register_number(ZFS_PROP_REFRATIO, "refcompressratio", 0, PROP_READONLY, ZFS_TYPE_DATASET, "<1.00x or higher if compressed>", "REFRATIO"); zprop_register_number(ZFS_PROP_VOLBLOCKSIZE, "volblocksize", ZVOL_DEFAULT_BLOCKSIZE, PROP_ONETIME, ZFS_TYPE_VOLUME, "512 to 128k, power of 2", "VOLBLOCK"); zprop_register_number(ZFS_PROP_USEDSNAP, "usedbysnapshots", 0, PROP_READONLY, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "", "USEDSNAP"); zprop_register_number(ZFS_PROP_USEDDS, "usedbydataset", 0, PROP_READONLY, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "", "USEDDS"); zprop_register_number(ZFS_PROP_USEDCHILD, "usedbychildren", 0, PROP_READONLY, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "", "USEDCHILD"); zprop_register_number(ZFS_PROP_USEDREFRESERV, "usedbyrefreservation", 0, PROP_READONLY, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "", "USEDREFRESERV"); zprop_register_number(ZFS_PROP_USERREFS, "userrefs", 0, PROP_READONLY, ZFS_TYPE_SNAPSHOT, "", "USERREFS"); zprop_register_number(ZFS_PROP_WRITTEN, "written", 0, PROP_READONLY, ZFS_TYPE_DATASET, "", "WRITTEN"); zprop_register_number(ZFS_PROP_LOGICALUSED, "logicalused", 0, PROP_READONLY, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "", "LUSED"); zprop_register_number(ZFS_PROP_LOGICALREFERENCED, "logicalreferenced", 0, PROP_READONLY, ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK, "", "LREFER"); zprop_register_number(ZFS_PROP_FILESYSTEM_COUNT, "filesystem_count", UINT64_MAX, PROP_READONLY, ZFS_TYPE_FILESYSTEM, "", "FSCOUNT"); zprop_register_number(ZFS_PROP_SNAPSHOT_COUNT, "snapshot_count", UINT64_MAX, PROP_READONLY, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "", "SSCOUNT"); zprop_register_number(ZFS_PROP_GUID, "guid", 0, PROP_READONLY, ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK, "", "GUID"); zprop_register_number(ZFS_PROP_CREATETXG, "createtxg", 0, PROP_READONLY, ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK, "", "CREATETXG"); zprop_register_number(ZFS_PROP_PBKDF2_ITERS, "pbkdf2iters", 0, PROP_ONETIME_DEFAULT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "", "PBKDF2ITERS"); zprop_register_number(ZFS_PROP_OBJSETID, "objsetid", 0, PROP_READONLY, ZFS_TYPE_DATASET, "", "OBJSETID"); /* default number properties */ zprop_register_number(ZFS_PROP_QUOTA, "quota", 0, PROP_DEFAULT, ZFS_TYPE_FILESYSTEM, " | none", "QUOTA"); zprop_register_number(ZFS_PROP_RESERVATION, "reservation", 0, PROP_DEFAULT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, " | none", "RESERV"); zprop_register_number(ZFS_PROP_VOLSIZE, "volsize", 0, PROP_DEFAULT, ZFS_TYPE_SNAPSHOT | ZFS_TYPE_VOLUME, "", "VOLSIZE"); zprop_register_number(ZFS_PROP_REFQUOTA, "refquota", 0, PROP_DEFAULT, ZFS_TYPE_FILESYSTEM, " | none", "REFQUOTA"); zprop_register_number(ZFS_PROP_REFRESERVATION, "refreservation", 0, PROP_DEFAULT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, " | none", "REFRESERV"); zprop_register_number(ZFS_PROP_FILESYSTEM_LIMIT, "filesystem_limit", UINT64_MAX, PROP_DEFAULT, ZFS_TYPE_FILESYSTEM, " | none", "FSLIMIT"); zprop_register_number(ZFS_PROP_SNAPSHOT_LIMIT, "snapshot_limit", UINT64_MAX, PROP_DEFAULT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, " | none", "SSLIMIT"); /* inherit number properties */ zprop_register_number(ZFS_PROP_RECORDSIZE, "recordsize", SPA_OLD_MAXBLOCKSIZE, PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "512 to 1M, power of 2", "RECSIZE"); zprop_register_number(ZFS_PROP_SPECIAL_SMALL_BLOCKS, "special_small_blocks", 0, PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "zero or 512 to 1M, power of 2", "SPECIAL_SMALL_BLOCKS"); /* hidden properties */ zprop_register_hidden(ZFS_PROP_NUMCLONES, "numclones", PROP_TYPE_NUMBER, PROP_READONLY, ZFS_TYPE_SNAPSHOT, "NUMCLONES"); zprop_register_hidden(ZFS_PROP_NAME, "name", PROP_TYPE_STRING, PROP_READONLY, ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK, "NAME"); zprop_register_hidden(ZFS_PROP_ISCSIOPTIONS, "iscsioptions", PROP_TYPE_STRING, PROP_INHERIT, ZFS_TYPE_VOLUME, "ISCSIOPTIONS"); zprop_register_hidden(ZFS_PROP_STMF_SHAREINFO, "stmf_sbd_lu", PROP_TYPE_STRING, PROP_INHERIT, ZFS_TYPE_VOLUME, "STMF_SBD_LU"); zprop_register_hidden(ZFS_PROP_USERACCOUNTING, "useraccounting", PROP_TYPE_NUMBER, PROP_READONLY, ZFS_TYPE_DATASET, "USERACCOUNTING"); zprop_register_hidden(ZFS_PROP_UNIQUE, "unique", PROP_TYPE_NUMBER, PROP_READONLY, ZFS_TYPE_DATASET, "UNIQUE"); zprop_register_hidden(ZFS_PROP_INCONSISTENT, "inconsistent", PROP_TYPE_NUMBER, PROP_READONLY, ZFS_TYPE_DATASET, "INCONSISTENT"); zprop_register_hidden(ZFS_PROP_IVSET_GUID, "ivsetguid", PROP_TYPE_NUMBER, PROP_READONLY, ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK, "IVSETGUID"); zprop_register_hidden(ZFS_PROP_PREV_SNAP, "prevsnap", PROP_TYPE_STRING, PROP_READONLY, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "PREVSNAP"); zprop_register_hidden(ZFS_PROP_PBKDF2_SALT, "pbkdf2salt", PROP_TYPE_NUMBER, PROP_ONETIME_DEFAULT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME, "PBKDF2SALT"); zprop_register_hidden(ZFS_PROP_KEY_GUID, "keyguid", PROP_TYPE_NUMBER, PROP_READONLY, ZFS_TYPE_DATASET, "KEYGUID"); zprop_register_hidden(ZFS_PROP_REDACTED, "redacted", PROP_TYPE_NUMBER, PROP_READONLY, ZFS_TYPE_DATASET, "REDACTED"); /* * Properties that are obsolete and not used. These are retained so * that we don't have to change the values of the zfs_prop_t enum, or * have NULL pointers in the zfs_prop_table[]. */ zprop_register_hidden(ZFS_PROP_REMAPTXG, "remaptxg", PROP_TYPE_NUMBER, PROP_READONLY, ZFS_TYPE_DATASET, "REMAPTXG"); /* oddball properties */ zprop_register_impl(ZFS_PROP_CREATION, "creation", PROP_TYPE_NUMBER, 0, NULL, PROP_READONLY, ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK, "", "CREATION", B_FALSE, B_TRUE, NULL); } boolean_t zfs_prop_delegatable(zfs_prop_t prop) { zprop_desc_t *pd = &zfs_prop_table[prop]; /* The mlslabel property is never delegatable. */ if (prop == ZFS_PROP_MLSLABEL) return (B_FALSE); return (pd->pd_attr != PROP_READONLY); } /* * Given a zfs dataset property name, returns the corresponding property ID. */ zfs_prop_t zfs_name_to_prop(const char *propname) { return (zprop_name_to_prop(propname, ZFS_TYPE_DATASET)); } /* * For user property names, we allow all lowercase alphanumeric characters, plus * a few useful punctuation characters. */ static int valid_char(char c) { return ((c >= 'a' && c <= 'z') || (c >= '0' && c <= '9') || c == '-' || c == '_' || c == '.' || c == ':'); } /* * Returns true if this is a valid user-defined property (one with a ':'). */ boolean_t zfs_prop_user(const char *name) { int i; char c; boolean_t foundsep = B_FALSE; for (i = 0; i < strlen(name); i++) { c = name[i]; if (!valid_char(c)) return (B_FALSE); if (c == ':') foundsep = B_TRUE; } if (!foundsep) return (B_FALSE); return (B_TRUE); } /* * Returns true if this is a valid userspace-type property (one with a '@'). * Note that after the @, any character is valid (eg, another @, for SID * user@domain). */ boolean_t zfs_prop_userquota(const char *name) { zfs_userquota_prop_t prop; for (prop = 0; prop < ZFS_NUM_USERQUOTA_PROPS; prop++) { if (strncmp(name, zfs_userquota_prop_prefixes[prop], strlen(zfs_userquota_prop_prefixes[prop])) == 0) { return (B_TRUE); } } return (B_FALSE); } /* * Returns true if this is a valid written@ property. * Note that after the @, any character is valid (eg, another @, for * written@pool/fs@origin). */ boolean_t zfs_prop_written(const char *name) { static const char *prop_prefix = "written@"; static const char *book_prefix = "written#"; return (strncmp(name, prop_prefix, strlen(prop_prefix)) == 0 || strncmp(name, book_prefix, strlen(book_prefix)) == 0); } /* * Tables of index types, plus functions to convert between the user view * (strings) and internal representation (uint64_t). */ int zfs_prop_string_to_index(zfs_prop_t prop, const char *string, uint64_t *index) { return (zprop_string_to_index(prop, string, index, ZFS_TYPE_DATASET)); } int zfs_prop_index_to_string(zfs_prop_t prop, uint64_t index, const char **string) { return (zprop_index_to_string(prop, index, string, ZFS_TYPE_DATASET)); } uint64_t zfs_prop_random_value(zfs_prop_t prop, uint64_t seed) { return (zprop_random_value(prop, seed, ZFS_TYPE_DATASET)); } /* * Returns TRUE if the property applies to any of the given dataset types. */ boolean_t zfs_prop_valid_for_type(int prop, zfs_type_t types, boolean_t headcheck) { return (zprop_valid_for_type(prop, types, headcheck)); } zprop_type_t zfs_prop_get_type(zfs_prop_t prop) { return (zfs_prop_table[prop].pd_proptype); } /* * Returns TRUE if the property is readonly. */ boolean_t zfs_prop_readonly(zfs_prop_t prop) { return (zfs_prop_table[prop].pd_attr == PROP_READONLY || zfs_prop_table[prop].pd_attr == PROP_ONETIME || zfs_prop_table[prop].pd_attr == PROP_ONETIME_DEFAULT); } /* * Returns TRUE if the property is visible (not hidden). */ boolean_t zfs_prop_visible(zfs_prop_t prop) { return (zfs_prop_table[prop].pd_visible && zfs_prop_table[prop].pd_zfs_mod_supported); } /* * Returns TRUE if the property is only allowed to be set once. */ boolean_t zfs_prop_setonce(zfs_prop_t prop) { return (zfs_prop_table[prop].pd_attr == PROP_ONETIME || zfs_prop_table[prop].pd_attr == PROP_ONETIME_DEFAULT); } const char * zfs_prop_default_string(zfs_prop_t prop) { return (zfs_prop_table[prop].pd_strdefault); } uint64_t zfs_prop_default_numeric(zfs_prop_t prop) { return (zfs_prop_table[prop].pd_numdefault); } /* * Given a dataset property ID, returns the corresponding name. * Assuming the zfs dataset property ID is valid. */ const char * zfs_prop_to_name(zfs_prop_t prop) { return (zfs_prop_table[prop].pd_name); } /* * Returns TRUE if the property is inheritable. */ boolean_t zfs_prop_inheritable(zfs_prop_t prop) { return (zfs_prop_table[prop].pd_attr == PROP_INHERIT || zfs_prop_table[prop].pd_attr == PROP_ONETIME); } /* * Returns TRUE if property is one of the encryption properties that requires * a loaded encryption key to modify. */ boolean_t zfs_prop_encryption_key_param(zfs_prop_t prop) { /* * keylocation does not count as an encryption property. It can be * changed at will without needing the master keys. */ return (prop == ZFS_PROP_PBKDF2_SALT || prop == ZFS_PROP_PBKDF2_ITERS || prop == ZFS_PROP_KEYFORMAT); } /* * Helper function used by both kernelspace and userspace to check the * keylocation property. If encrypted is set, the keylocation must be valid * for an encrypted dataset. */ boolean_t zfs_prop_valid_keylocation(const char *str, boolean_t encrypted) { if (strcmp("none", str) == 0) return (!encrypted); else if (strcmp("prompt", str) == 0) return (B_TRUE); else if (strlen(str) > 8 && strncmp("file:///", str, 8) == 0) return (B_TRUE); return (B_FALSE); } #ifndef _KERNEL #include /* * Returns a string describing the set of acceptable values for the given * zfs property, or NULL if it cannot be set. */ const char * zfs_prop_values(zfs_prop_t prop) { return (zfs_prop_table[prop].pd_values); } /* * Returns TRUE if this property is a string type. Note that index types * (compression, checksum) are treated as strings in userland, even though they * are stored numerically on disk. */ int zfs_prop_is_string(zfs_prop_t prop) { return (zfs_prop_table[prop].pd_proptype == PROP_TYPE_STRING || zfs_prop_table[prop].pd_proptype == PROP_TYPE_INDEX); } /* * Returns the column header for the given property. Used only in * 'zfs list -o', but centralized here with the other property information. */ const char * zfs_prop_column_name(zfs_prop_t prop) { return (zfs_prop_table[prop].pd_colname); } /* * Returns whether the given property should be displayed right-justified for * 'zfs list'. */ boolean_t zfs_prop_align_right(zfs_prop_t prop) { return (zfs_prop_table[prop].pd_rightalign); } #endif #if defined(_KERNEL) #include #if defined(HAVE_KERNEL_FPU_INTERNAL) union fpregs_state **zfs_kfpu_fpregs; EXPORT_SYMBOL(zfs_kfpu_fpregs); #endif /* HAVE_KERNEL_FPU_INTERNAL */ static int __init zcommon_init(void) { int error = kfpu_init(); if (error) return (error); fletcher_4_init(); return (0); } static void __exit zcommon_fini(void) { fletcher_4_fini(); kfpu_fini(); } module_init(zcommon_init); module_exit(zcommon_fini); #endif ZFS_MODULE_DESCRIPTION("Generic ZFS support"); ZFS_MODULE_AUTHOR(ZFS_META_AUTHOR); ZFS_MODULE_LICENSE(ZFS_META_LICENSE); ZFS_MODULE_VERSION(ZFS_META_VERSION "-" ZFS_META_RELEASE); /* zfs dataset property functions */ EXPORT_SYMBOL(zfs_userquota_prop_prefixes); EXPORT_SYMBOL(zfs_prop_init); EXPORT_SYMBOL(zfs_prop_get_type); EXPORT_SYMBOL(zfs_prop_get_table); EXPORT_SYMBOL(zfs_prop_delegatable); EXPORT_SYMBOL(zfs_prop_visible); /* Dataset property functions shared between libzfs and kernel. */ EXPORT_SYMBOL(zfs_prop_default_string); EXPORT_SYMBOL(zfs_prop_default_numeric); EXPORT_SYMBOL(zfs_prop_readonly); EXPORT_SYMBOL(zfs_prop_inheritable); EXPORT_SYMBOL(zfs_prop_encryption_key_param); EXPORT_SYMBOL(zfs_prop_valid_keylocation); EXPORT_SYMBOL(zfs_prop_setonce); EXPORT_SYMBOL(zfs_prop_to_name); EXPORT_SYMBOL(zfs_name_to_prop); EXPORT_SYMBOL(zfs_prop_user); EXPORT_SYMBOL(zfs_prop_userquota); EXPORT_SYMBOL(zfs_prop_index_to_string); EXPORT_SYMBOL(zfs_prop_string_to_index); EXPORT_SYMBOL(zfs_prop_valid_for_type); EXPORT_SYMBOL(zfs_prop_written); diff --git a/module/zfs/dsl_scan.c b/module/zfs/dsl_scan.c index 4704781bfa45..f6a5ceca6a0e 100644 --- a/module/zfs/dsl_scan.c +++ b/module/zfs/dsl_scan.c @@ -1,4429 +1,4429 @@ /* * 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) 2008, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2018 by Delphix. All rights reserved. * Copyright 2016 Gary Mills * Copyright (c) 2017, 2019, Datto Inc. All rights reserved. * Copyright (c) 2015, Nexenta Systems, Inc. All rights reserved. * Copyright 2019 Joyent, Inc. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef _KERNEL #include #endif /* * Grand theory statement on scan queue sorting * * Scanning is implemented by recursively traversing all indirection levels * in an object and reading all blocks referenced from said objects. This * results in us approximately traversing the object from lowest logical * offset to the highest. For best performance, we would want the logical * blocks to be physically contiguous. However, this is frequently not the * case with pools given the allocation patterns of copy-on-write filesystems. * So instead, we put the I/Os into a reordering queue and issue them in a * way that will most benefit physical disks (LBA-order). * * Queue management: * * Ideally, we would want to scan all metadata and queue up all block I/O * prior to starting to issue it, because that allows us to do an optimal * sorting job. This can however consume large amounts of memory. Therefore * we continuously monitor the size of the queues and constrain them to 5% * (zfs_scan_mem_lim_fact) of physmem. If the queues grow larger than this * limit, we clear out a few of the largest extents at the head of the queues * to make room for more scanning. Hopefully, these extents will be fairly * large and contiguous, allowing us to approach sequential I/O throughput * even without a fully sorted tree. * * Metadata scanning takes place in dsl_scan_visit(), which is called from * dsl_scan_sync() every spa_sync(). If we have either fully scanned all * metadata on the pool, or we need to make room in memory because our * queues are too large, dsl_scan_visit() is postponed and * scan_io_queues_run() is called from dsl_scan_sync() instead. This implies * that metadata scanning and queued I/O issuing are mutually exclusive. This * allows us to provide maximum sequential I/O throughput for the majority of * I/O's issued since sequential I/O performance is significantly negatively * impacted if it is interleaved with random I/O. * * Implementation Notes * * One side effect of the queued scanning algorithm is that the scanning code * needs to be notified whenever a block is freed. This is needed to allow * the scanning code to remove these I/Os from the issuing queue. Additionally, * we do not attempt to queue gang blocks to be issued sequentially since this * is very hard to do and would have an extremely limited performance benefit. * Instead, we simply issue gang I/Os as soon as we find them using the legacy * algorithm. * * Backwards compatibility * * This new algorithm is backwards compatible with the legacy on-disk data * structures (and therefore does not require a new feature flag). * Periodically during scanning (see zfs_scan_checkpoint_intval), the scan * will stop scanning metadata (in logical order) and wait for all outstanding * sorted I/O to complete. Once this is done, we write out a checkpoint * bookmark, indicating that we have scanned everything logically before it. * If the pool is imported on a machine without the new sorting algorithm, * the scan simply resumes from the last checkpoint using the legacy algorithm. */ typedef int (scan_cb_t)(dsl_pool_t *, const blkptr_t *, const zbookmark_phys_t *); static scan_cb_t dsl_scan_scrub_cb; static int scan_ds_queue_compare(const void *a, const void *b); static int scan_prefetch_queue_compare(const void *a, const void *b); static void scan_ds_queue_clear(dsl_scan_t *scn); static void scan_ds_prefetch_queue_clear(dsl_scan_t *scn); static boolean_t scan_ds_queue_contains(dsl_scan_t *scn, uint64_t dsobj, uint64_t *txg); static void scan_ds_queue_insert(dsl_scan_t *scn, uint64_t dsobj, uint64_t txg); static void scan_ds_queue_remove(dsl_scan_t *scn, uint64_t dsobj); static void scan_ds_queue_sync(dsl_scan_t *scn, dmu_tx_t *tx); static uint64_t dsl_scan_count_leaves(vdev_t *vd); extern int zfs_vdev_async_write_active_min_dirty_percent; /* * By default zfs will check to ensure it is not over the hard memory * limit before each txg. If finer-grained control of this is needed * this value can be set to 1 to enable checking before scanning each * block. */ int zfs_scan_strict_mem_lim = B_FALSE; /* * Maximum number of parallelly executed bytes per leaf vdev. We attempt * to strike a balance here between keeping the vdev queues full of I/Os * at all times and not overflowing the queues to cause long latency, * which would cause long txg sync times. No matter what, we will not * overload the drives with I/O, since that is protected by * zfs_vdev_scrub_max_active. */ unsigned long zfs_scan_vdev_limit = 4 << 20; int zfs_scan_issue_strategy = 0; int zfs_scan_legacy = B_FALSE; /* don't queue & sort zios, go direct */ unsigned long zfs_scan_max_ext_gap = 2 << 20; /* in bytes */ /* * fill_weight is non-tunable at runtime, so we copy it at module init from * zfs_scan_fill_weight. Runtime adjustments to zfs_scan_fill_weight would * break queue sorting. */ int zfs_scan_fill_weight = 3; static uint64_t fill_weight; /* See dsl_scan_should_clear() for details on the memory limit tunables */ uint64_t zfs_scan_mem_lim_min = 16 << 20; /* bytes */ uint64_t zfs_scan_mem_lim_soft_max = 128 << 20; /* bytes */ int zfs_scan_mem_lim_fact = 20; /* fraction of physmem */ int zfs_scan_mem_lim_soft_fact = 20; /* fraction of mem lim above */ int zfs_scrub_min_time_ms = 1000; /* min millisecs to scrub per txg */ int zfs_obsolete_min_time_ms = 500; /* min millisecs to obsolete per txg */ int zfs_free_min_time_ms = 1000; /* min millisecs to free per txg */ int zfs_resilver_min_time_ms = 3000; /* min millisecs to resilver per txg */ int zfs_scan_checkpoint_intval = 7200; /* in seconds */ int zfs_scan_suspend_progress = 0; /* set to prevent scans from progressing */ int zfs_no_scrub_io = B_FALSE; /* set to disable scrub i/o */ int zfs_no_scrub_prefetch = B_FALSE; /* set to disable scrub prefetch */ enum ddt_class zfs_scrub_ddt_class_max = DDT_CLASS_DUPLICATE; /* max number of blocks to free in a single TXG */ unsigned long zfs_async_block_max_blocks = ULONG_MAX; /* max number of dedup blocks to free in a single TXG */ unsigned long zfs_max_async_dedup_frees = 100000; int zfs_resilver_disable_defer = 0; /* set to disable resilver deferring */ /* * We wait a few txgs after importing a pool to begin scanning so that * the import / mounting code isn't held up by scrub / resilver IO. * Unfortunately, it is a bit difficult to determine exactly how long * this will take since userspace will trigger fs mounts asynchronously * and the kernel will create zvol minors asynchronously. As a result, * the value provided here is a bit arbitrary, but represents a * reasonable estimate of how many txgs it will take to finish fully * importing a pool */ #define SCAN_IMPORT_WAIT_TXGS 5 #define DSL_SCAN_IS_SCRUB_RESILVER(scn) \ ((scn)->scn_phys.scn_func == POOL_SCAN_SCRUB || \ (scn)->scn_phys.scn_func == POOL_SCAN_RESILVER) /* * Enable/disable the processing of the free_bpobj object. */ int zfs_free_bpobj_enabled = 1; /* the order has to match pool_scan_type */ static scan_cb_t *scan_funcs[POOL_SCAN_FUNCS] = { NULL, dsl_scan_scrub_cb, /* POOL_SCAN_SCRUB */ dsl_scan_scrub_cb, /* POOL_SCAN_RESILVER */ }; /* In core node for the scn->scn_queue. Represents a dataset to be scanned */ typedef struct { uint64_t sds_dsobj; uint64_t sds_txg; avl_node_t sds_node; } scan_ds_t; /* * This controls what conditions are placed on dsl_scan_sync_state(): * SYNC_OPTIONAL) write out scn_phys iff scn_bytes_pending == 0 * SYNC_MANDATORY) write out scn_phys always. scn_bytes_pending must be 0. * SYNC_CACHED) if scn_bytes_pending == 0, write out scn_phys. Otherwise * write out the scn_phys_cached version. * See dsl_scan_sync_state for details. */ typedef enum { SYNC_OPTIONAL, SYNC_MANDATORY, SYNC_CACHED } state_sync_type_t; /* * This struct represents the minimum information needed to reconstruct a * zio for sequential scanning. This is useful because many of these will * accumulate in the sequential IO queues before being issued, so saving * memory matters here. */ typedef struct scan_io { /* fields from blkptr_t */ uint64_t sio_blk_prop; uint64_t sio_phys_birth; uint64_t sio_birth; zio_cksum_t sio_cksum; uint32_t sio_nr_dvas; /* fields from zio_t */ uint32_t sio_flags; zbookmark_phys_t sio_zb; /* members for queue sorting */ union { avl_node_t sio_addr_node; /* link into issuing queue */ list_node_t sio_list_node; /* link for issuing to disk */ } sio_nodes; /* * There may be up to SPA_DVAS_PER_BP DVAs here from the bp, * depending on how many were in the original bp. Only the * first DVA is really used for sorting and issuing purposes. * The other DVAs (if provided) simply exist so that the zio * layer can find additional copies to repair from in the * event of an error. This array must go at the end of the * struct to allow this for the variable number of elements. */ dva_t sio_dva[0]; } scan_io_t; #define SIO_SET_OFFSET(sio, x) DVA_SET_OFFSET(&(sio)->sio_dva[0], x) #define SIO_SET_ASIZE(sio, x) DVA_SET_ASIZE(&(sio)->sio_dva[0], x) #define SIO_GET_OFFSET(sio) DVA_GET_OFFSET(&(sio)->sio_dva[0]) #define SIO_GET_ASIZE(sio) DVA_GET_ASIZE(&(sio)->sio_dva[0]) #define SIO_GET_END_OFFSET(sio) \ (SIO_GET_OFFSET(sio) + SIO_GET_ASIZE(sio)) #define SIO_GET_MUSED(sio) \ (sizeof (scan_io_t) + ((sio)->sio_nr_dvas * sizeof (dva_t))) struct dsl_scan_io_queue { dsl_scan_t *q_scn; /* associated dsl_scan_t */ vdev_t *q_vd; /* top-level vdev that this queue represents */ /* trees used for sorting I/Os and extents of I/Os */ range_tree_t *q_exts_by_addr; zfs_btree_t q_exts_by_size; avl_tree_t q_sios_by_addr; uint64_t q_sio_memused; /* members for zio rate limiting */ uint64_t q_maxinflight_bytes; uint64_t q_inflight_bytes; kcondvar_t q_zio_cv; /* used under vd->vdev_scan_io_queue_lock */ /* per txg statistics */ uint64_t q_total_seg_size_this_txg; uint64_t q_segs_this_txg; uint64_t q_total_zio_size_this_txg; uint64_t q_zios_this_txg; }; /* private data for dsl_scan_prefetch_cb() */ typedef struct scan_prefetch_ctx { zfs_refcount_t spc_refcnt; /* refcount for memory management */ dsl_scan_t *spc_scn; /* dsl_scan_t for the pool */ boolean_t spc_root; /* is this prefetch for an objset? */ uint8_t spc_indblkshift; /* dn_indblkshift of current dnode */ uint16_t spc_datablkszsec; /* dn_idatablkszsec of current dnode */ } scan_prefetch_ctx_t; /* private data for dsl_scan_prefetch() */ typedef struct scan_prefetch_issue_ctx { avl_node_t spic_avl_node; /* link into scn->scn_prefetch_queue */ scan_prefetch_ctx_t *spic_spc; /* spc for the callback */ blkptr_t spic_bp; /* bp to prefetch */ zbookmark_phys_t spic_zb; /* bookmark to prefetch */ } scan_prefetch_issue_ctx_t; static void scan_exec_io(dsl_pool_t *dp, const blkptr_t *bp, int zio_flags, const zbookmark_phys_t *zb, dsl_scan_io_queue_t *queue); static void scan_io_queue_insert_impl(dsl_scan_io_queue_t *queue, scan_io_t *sio); static dsl_scan_io_queue_t *scan_io_queue_create(vdev_t *vd); static void scan_io_queues_destroy(dsl_scan_t *scn); static kmem_cache_t *sio_cache[SPA_DVAS_PER_BP]; /* sio->sio_nr_dvas must be set so we know which cache to free from */ static void sio_free(scan_io_t *sio) { ASSERT3U(sio->sio_nr_dvas, >, 0); ASSERT3U(sio->sio_nr_dvas, <=, SPA_DVAS_PER_BP); kmem_cache_free(sio_cache[sio->sio_nr_dvas - 1], sio); } /* It is up to the caller to set sio->sio_nr_dvas for freeing */ static scan_io_t * sio_alloc(unsigned short nr_dvas) { ASSERT3U(nr_dvas, >, 0); ASSERT3U(nr_dvas, <=, SPA_DVAS_PER_BP); return (kmem_cache_alloc(sio_cache[nr_dvas - 1], KM_SLEEP)); } void scan_init(void) { /* * This is used in ext_size_compare() to weight segments * based on how sparse they are. This cannot be changed * mid-scan and the tree comparison functions don't currently * have a mechanism for passing additional context to the * compare functions. Thus we store this value globally and * we only allow it to be set at module initialization time */ fill_weight = zfs_scan_fill_weight; for (int i = 0; i < SPA_DVAS_PER_BP; i++) { char name[36]; (void) snprintf(name, sizeof (name), "sio_cache_%d", i); sio_cache[i] = kmem_cache_create(name, (sizeof (scan_io_t) + ((i + 1) * sizeof (dva_t))), 0, NULL, NULL, NULL, NULL, NULL, 0); } } void scan_fini(void) { for (int i = 0; i < SPA_DVAS_PER_BP; i++) { kmem_cache_destroy(sio_cache[i]); } } static inline boolean_t dsl_scan_is_running(const dsl_scan_t *scn) { return (scn->scn_phys.scn_state == DSS_SCANNING); } boolean_t dsl_scan_resilvering(dsl_pool_t *dp) { return (dsl_scan_is_running(dp->dp_scan) && dp->dp_scan->scn_phys.scn_func == POOL_SCAN_RESILVER); } static inline void sio2bp(const scan_io_t *sio, blkptr_t *bp) { bzero(bp, sizeof (*bp)); bp->blk_prop = sio->sio_blk_prop; bp->blk_phys_birth = sio->sio_phys_birth; bp->blk_birth = sio->sio_birth; bp->blk_fill = 1; /* we always only work with data pointers */ bp->blk_cksum = sio->sio_cksum; ASSERT3U(sio->sio_nr_dvas, >, 0); ASSERT3U(sio->sio_nr_dvas, <=, SPA_DVAS_PER_BP); bcopy(sio->sio_dva, bp->blk_dva, sio->sio_nr_dvas * sizeof (dva_t)); } static inline void bp2sio(const blkptr_t *bp, scan_io_t *sio, int dva_i) { sio->sio_blk_prop = bp->blk_prop; sio->sio_phys_birth = bp->blk_phys_birth; sio->sio_birth = bp->blk_birth; sio->sio_cksum = bp->blk_cksum; sio->sio_nr_dvas = BP_GET_NDVAS(bp); /* * Copy the DVAs to the sio. We need all copies of the block so * that the self healing code can use the alternate copies if the * first is corrupted. We want the DVA at index dva_i to be first * in the sio since this is the primary one that we want to issue. */ for (int i = 0, j = dva_i; i < sio->sio_nr_dvas; i++, j++) { sio->sio_dva[i] = bp->blk_dva[j % sio->sio_nr_dvas]; } } int dsl_scan_init(dsl_pool_t *dp, uint64_t txg) { int err; dsl_scan_t *scn; spa_t *spa = dp->dp_spa; uint64_t f; scn = dp->dp_scan = kmem_zalloc(sizeof (dsl_scan_t), KM_SLEEP); scn->scn_dp = dp; /* * It's possible that we're resuming a scan after a reboot so * make sure that the scan_async_destroying flag is initialized * appropriately. */ ASSERT(!scn->scn_async_destroying); scn->scn_async_destroying = spa_feature_is_active(dp->dp_spa, SPA_FEATURE_ASYNC_DESTROY); /* * Calculate the max number of in-flight bytes for pool-wide * scanning operations (minimum 1MB). Limits for the issuing * phase are done per top-level vdev and are handled separately. */ scn->scn_maxinflight_bytes = MAX(zfs_scan_vdev_limit * dsl_scan_count_leaves(spa->spa_root_vdev), 1ULL << 20); avl_create(&scn->scn_queue, scan_ds_queue_compare, sizeof (scan_ds_t), offsetof(scan_ds_t, sds_node)); avl_create(&scn->scn_prefetch_queue, scan_prefetch_queue_compare, sizeof (scan_prefetch_issue_ctx_t), offsetof(scan_prefetch_issue_ctx_t, spic_avl_node)); err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, "scrub_func", sizeof (uint64_t), 1, &f); if (err == 0) { /* * There was an old-style scrub in progress. Restart a * new-style scrub from the beginning. */ scn->scn_restart_txg = txg; zfs_dbgmsg("old-style scrub was in progress; " "restarting new-style scrub in txg %llu", (longlong_t)scn->scn_restart_txg); /* * Load the queue obj from the old location so that it * can be freed by dsl_scan_done(). */ (void) zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, "scrub_queue", sizeof (uint64_t), 1, &scn->scn_phys.scn_queue_obj); } else { err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SCAN, sizeof (uint64_t), SCAN_PHYS_NUMINTS, &scn->scn_phys); /* * Detect if the pool contains the signature of #2094. If it * does properly update the scn->scn_phys structure and notify * the administrator by setting an errata for the pool. */ if (err == EOVERFLOW) { uint64_t zaptmp[SCAN_PHYS_NUMINTS + 1]; VERIFY3S(SCAN_PHYS_NUMINTS, ==, 24); VERIFY3S(offsetof(dsl_scan_phys_t, scn_flags), ==, (23 * sizeof (uint64_t))); err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SCAN, sizeof (uint64_t), SCAN_PHYS_NUMINTS + 1, &zaptmp); if (err == 0) { uint64_t overflow = zaptmp[SCAN_PHYS_NUMINTS]; if (overflow & ~DSL_SCAN_FLAGS_MASK || scn->scn_async_destroying) { spa->spa_errata = ZPOOL_ERRATA_ZOL_2094_ASYNC_DESTROY; return (EOVERFLOW); } bcopy(zaptmp, &scn->scn_phys, SCAN_PHYS_NUMINTS * sizeof (uint64_t)); scn->scn_phys.scn_flags = overflow; /* Required scrub already in progress. */ if (scn->scn_phys.scn_state == DSS_FINISHED || scn->scn_phys.scn_state == DSS_CANCELED) spa->spa_errata = ZPOOL_ERRATA_ZOL_2094_SCRUB; } } if (err == ENOENT) return (0); else if (err) return (err); /* * We might be restarting after a reboot, so jump the issued * counter to how far we've scanned. We know we're consistent * up to here. */ scn->scn_issued_before_pass = scn->scn_phys.scn_examined; if (dsl_scan_is_running(scn) && spa_prev_software_version(dp->dp_spa) < SPA_VERSION_SCAN) { /* * A new-type scrub was in progress on an old * pool, and the pool was accessed by old * software. Restart from the beginning, since * the old software may have changed the pool in * the meantime. */ scn->scn_restart_txg = txg; zfs_dbgmsg("new-style scrub was modified " "by old software; restarting in txg %llu", (longlong_t)scn->scn_restart_txg); } else if (dsl_scan_resilvering(dp)) { /* * If a resilver is in progress and there are already * errors, restart it instead of finishing this scan and * then restarting it. If there haven't been any errors * then remember that the incore DTL is valid. */ if (scn->scn_phys.scn_errors > 0) { scn->scn_restart_txg = txg; zfs_dbgmsg("resilver can't excise DTL_MISSING " "when finished; restarting in txg %llu", (u_longlong_t)scn->scn_restart_txg); } else { /* it's safe to excise DTL when finished */ spa->spa_scrub_started = B_TRUE; } } } bcopy(&scn->scn_phys, &scn->scn_phys_cached, sizeof (scn->scn_phys)); /* reload the queue into the in-core state */ if (scn->scn_phys.scn_queue_obj != 0) { zap_cursor_t zc; zap_attribute_t za; for (zap_cursor_init(&zc, dp->dp_meta_objset, scn->scn_phys.scn_queue_obj); zap_cursor_retrieve(&zc, &za) == 0; (void) zap_cursor_advance(&zc)) { scan_ds_queue_insert(scn, zfs_strtonum(za.za_name, NULL), za.za_first_integer); } zap_cursor_fini(&zc); } spa_scan_stat_init(spa); return (0); } void dsl_scan_fini(dsl_pool_t *dp) { if (dp->dp_scan != NULL) { dsl_scan_t *scn = dp->dp_scan; if (scn->scn_taskq != NULL) taskq_destroy(scn->scn_taskq); scan_ds_queue_clear(scn); avl_destroy(&scn->scn_queue); scan_ds_prefetch_queue_clear(scn); avl_destroy(&scn->scn_prefetch_queue); kmem_free(dp->dp_scan, sizeof (dsl_scan_t)); dp->dp_scan = NULL; } } static boolean_t dsl_scan_restarting(dsl_scan_t *scn, dmu_tx_t *tx) { return (scn->scn_restart_txg != 0 && scn->scn_restart_txg <= tx->tx_txg); } boolean_t dsl_scan_resilver_scheduled(dsl_pool_t *dp) { return ((dp->dp_scan && dp->dp_scan->scn_restart_txg != 0) || (spa_async_tasks(dp->dp_spa) & SPA_ASYNC_RESILVER)); } boolean_t dsl_scan_scrubbing(const dsl_pool_t *dp) { dsl_scan_phys_t *scn_phys = &dp->dp_scan->scn_phys; return (scn_phys->scn_state == DSS_SCANNING && scn_phys->scn_func == POOL_SCAN_SCRUB); } boolean_t dsl_scan_is_paused_scrub(const dsl_scan_t *scn) { return (dsl_scan_scrubbing(scn->scn_dp) && scn->scn_phys.scn_flags & DSF_SCRUB_PAUSED); } /* * Writes out a persistent dsl_scan_phys_t record to the pool directory. * Because we can be running in the block sorting algorithm, we do not always * want to write out the record, only when it is "safe" to do so. This safety * condition is achieved by making sure that the sorting queues are empty * (scn_bytes_pending == 0). When this condition is not true, the sync'd state * is inconsistent with how much actual scanning progress has been made. The * kind of sync to be performed is specified by the sync_type argument. If the * sync is optional, we only sync if the queues are empty. If the sync is * mandatory, we do a hard ASSERT to make sure that the queues are empty. The * third possible state is a "cached" sync. This is done in response to: * 1) The dataset that was in the last sync'd dsl_scan_phys_t having been * destroyed, so we wouldn't be able to restart scanning from it. * 2) The snapshot that was in the last sync'd dsl_scan_phys_t having been * superseded by a newer snapshot. * 3) The dataset that was in the last sync'd dsl_scan_phys_t having been * swapped with its clone. * In all cases, a cached sync simply rewrites the last record we've written, * just slightly modified. For the modifications that are performed to the * last written dsl_scan_phys_t, see dsl_scan_ds_destroyed, * dsl_scan_ds_snapshotted and dsl_scan_ds_clone_swapped. */ static void dsl_scan_sync_state(dsl_scan_t *scn, dmu_tx_t *tx, state_sync_type_t sync_type) { int i; spa_t *spa = scn->scn_dp->dp_spa; ASSERT(sync_type != SYNC_MANDATORY || scn->scn_bytes_pending == 0); if (scn->scn_bytes_pending == 0) { for (i = 0; i < spa->spa_root_vdev->vdev_children; i++) { vdev_t *vd = spa->spa_root_vdev->vdev_child[i]; dsl_scan_io_queue_t *q = vd->vdev_scan_io_queue; if (q == NULL) continue; mutex_enter(&vd->vdev_scan_io_queue_lock); ASSERT3P(avl_first(&q->q_sios_by_addr), ==, NULL); ASSERT3P(zfs_btree_first(&q->q_exts_by_size, NULL), ==, NULL); ASSERT3P(range_tree_first(q->q_exts_by_addr), ==, NULL); mutex_exit(&vd->vdev_scan_io_queue_lock); } if (scn->scn_phys.scn_queue_obj != 0) scan_ds_queue_sync(scn, tx); VERIFY0(zap_update(scn->scn_dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SCAN, sizeof (uint64_t), SCAN_PHYS_NUMINTS, &scn->scn_phys, tx)); bcopy(&scn->scn_phys, &scn->scn_phys_cached, sizeof (scn->scn_phys)); if (scn->scn_checkpointing) zfs_dbgmsg("finish scan checkpoint"); scn->scn_checkpointing = B_FALSE; scn->scn_last_checkpoint = ddi_get_lbolt(); } else if (sync_type == SYNC_CACHED) { VERIFY0(zap_update(scn->scn_dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SCAN, sizeof (uint64_t), SCAN_PHYS_NUMINTS, &scn->scn_phys_cached, tx)); } } /* ARGSUSED */ static int dsl_scan_setup_check(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; vdev_t *rvd = scn->scn_dp->dp_spa->spa_root_vdev; if (dsl_scan_is_running(scn) || vdev_rebuild_active(rvd)) return (SET_ERROR(EBUSY)); return (0); } static void dsl_scan_setup_sync(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; pool_scan_func_t *funcp = arg; dmu_object_type_t ot = 0; dsl_pool_t *dp = scn->scn_dp; spa_t *spa = dp->dp_spa; ASSERT(!dsl_scan_is_running(scn)); ASSERT(*funcp > POOL_SCAN_NONE && *funcp < POOL_SCAN_FUNCS); bzero(&scn->scn_phys, sizeof (scn->scn_phys)); scn->scn_phys.scn_func = *funcp; scn->scn_phys.scn_state = DSS_SCANNING; scn->scn_phys.scn_min_txg = 0; scn->scn_phys.scn_max_txg = tx->tx_txg; scn->scn_phys.scn_ddt_class_max = DDT_CLASSES - 1; /* the entire DDT */ scn->scn_phys.scn_start_time = gethrestime_sec(); scn->scn_phys.scn_errors = 0; scn->scn_phys.scn_to_examine = spa->spa_root_vdev->vdev_stat.vs_alloc; scn->scn_issued_before_pass = 0; scn->scn_restart_txg = 0; scn->scn_done_txg = 0; scn->scn_last_checkpoint = 0; scn->scn_checkpointing = B_FALSE; spa_scan_stat_init(spa); if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) { scn->scn_phys.scn_ddt_class_max = zfs_scrub_ddt_class_max; /* rewrite all disk labels */ vdev_config_dirty(spa->spa_root_vdev); if (vdev_resilver_needed(spa->spa_root_vdev, &scn->scn_phys.scn_min_txg, &scn->scn_phys.scn_max_txg)) { nvlist_t *aux = fnvlist_alloc(); fnvlist_add_string(aux, ZFS_EV_RESILVER_TYPE, "healing"); spa_event_notify(spa, NULL, aux, ESC_ZFS_RESILVER_START); nvlist_free(aux); } else { spa_event_notify(spa, NULL, NULL, ESC_ZFS_SCRUB_START); } spa->spa_scrub_started = B_TRUE; /* * If this is an incremental scrub, limit the DDT scrub phase * to just the auto-ditto class (for correctness); the rest * of the scrub should go faster using top-down pruning. */ if (scn->scn_phys.scn_min_txg > TXG_INITIAL) scn->scn_phys.scn_ddt_class_max = DDT_CLASS_DITTO; /* * When starting a resilver clear any existing rebuild state. * This is required to prevent stale rebuild status from * being reported when a rebuild is run, then a resilver and * finally a scrub. In which case only the scrub status * should be reported by 'zpool status'. */ if (scn->scn_phys.scn_func == POOL_SCAN_RESILVER) { vdev_t *rvd = spa->spa_root_vdev; for (uint64_t i = 0; i < rvd->vdev_children; i++) { vdev_t *vd = rvd->vdev_child[i]; vdev_rebuild_clear_sync( (void *)(uintptr_t)vd->vdev_id, tx); } } } /* back to the generic stuff */ if (dp->dp_blkstats == NULL) { dp->dp_blkstats = vmem_alloc(sizeof (zfs_all_blkstats_t), KM_SLEEP); mutex_init(&dp->dp_blkstats->zab_lock, NULL, MUTEX_DEFAULT, NULL); } bzero(&dp->dp_blkstats->zab_type, sizeof (dp->dp_blkstats->zab_type)); if (spa_version(spa) < SPA_VERSION_DSL_SCRUB) ot = DMU_OT_ZAP_OTHER; scn->scn_phys.scn_queue_obj = zap_create(dp->dp_meta_objset, ot ? ot : DMU_OT_SCAN_QUEUE, DMU_OT_NONE, 0, tx); bcopy(&scn->scn_phys, &scn->scn_phys_cached, sizeof (scn->scn_phys)); dsl_scan_sync_state(scn, tx, SYNC_MANDATORY); spa_history_log_internal(spa, "scan setup", tx, "func=%u mintxg=%llu maxtxg=%llu", *funcp, (u_longlong_t)scn->scn_phys.scn_min_txg, (u_longlong_t)scn->scn_phys.scn_max_txg); } /* * Called by the ZFS_IOC_POOL_SCAN ioctl to start a scrub or resilver. * Can also be called to resume a paused scrub. */ int dsl_scan(dsl_pool_t *dp, pool_scan_func_t func) { spa_t *spa = dp->dp_spa; dsl_scan_t *scn = dp->dp_scan; /* * Purge all vdev caches and probe all devices. We do this here * rather than in sync context because this requires a writer lock * on the spa_config lock, which we can't do from sync context. The * spa_scrub_reopen flag indicates that vdev_open() should not * attempt to start another scrub. */ spa_vdev_state_enter(spa, SCL_NONE); spa->spa_scrub_reopen = B_TRUE; vdev_reopen(spa->spa_root_vdev); spa->spa_scrub_reopen = B_FALSE; (void) spa_vdev_state_exit(spa, NULL, 0); if (func == POOL_SCAN_RESILVER) { dsl_scan_restart_resilver(spa->spa_dsl_pool, 0); return (0); } if (func == POOL_SCAN_SCRUB && dsl_scan_is_paused_scrub(scn)) { /* got scrub start cmd, resume paused scrub */ int err = dsl_scrub_set_pause_resume(scn->scn_dp, POOL_SCRUB_NORMAL); if (err == 0) { spa_event_notify(spa, NULL, NULL, ESC_ZFS_SCRUB_RESUME); return (SET_ERROR(ECANCELED)); } return (SET_ERROR(err)); } return (dsl_sync_task(spa_name(spa), dsl_scan_setup_check, dsl_scan_setup_sync, &func, 0, ZFS_SPACE_CHECK_EXTRA_RESERVED)); } /* ARGSUSED */ static void dsl_scan_done(dsl_scan_t *scn, boolean_t complete, dmu_tx_t *tx) { static const char *old_names[] = { "scrub_bookmark", "scrub_ddt_bookmark", "scrub_ddt_class_max", "scrub_queue", "scrub_min_txg", "scrub_max_txg", "scrub_func", "scrub_errors", NULL }; dsl_pool_t *dp = scn->scn_dp; spa_t *spa = dp->dp_spa; int i; /* Remove any remnants of an old-style scrub. */ for (i = 0; old_names[i]; i++) { (void) zap_remove(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, old_names[i], tx); } if (scn->scn_phys.scn_queue_obj != 0) { VERIFY0(dmu_object_free(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, tx)); scn->scn_phys.scn_queue_obj = 0; } scan_ds_queue_clear(scn); scan_ds_prefetch_queue_clear(scn); scn->scn_phys.scn_flags &= ~DSF_SCRUB_PAUSED; /* * If we were "restarted" from a stopped state, don't bother * with anything else. */ if (!dsl_scan_is_running(scn)) { ASSERT(!scn->scn_is_sorted); return; } if (scn->scn_is_sorted) { scan_io_queues_destroy(scn); scn->scn_is_sorted = B_FALSE; if (scn->scn_taskq != NULL) { taskq_destroy(scn->scn_taskq); scn->scn_taskq = NULL; } } scn->scn_phys.scn_state = complete ? DSS_FINISHED : DSS_CANCELED; spa_notify_waiters(spa); if (dsl_scan_restarting(scn, tx)) spa_history_log_internal(spa, "scan aborted, restarting", tx, "errors=%llu", (u_longlong_t)spa_get_errlog_size(spa)); else if (!complete) spa_history_log_internal(spa, "scan cancelled", tx, "errors=%llu", (u_longlong_t)spa_get_errlog_size(spa)); else spa_history_log_internal(spa, "scan done", tx, "errors=%llu", (u_longlong_t)spa_get_errlog_size(spa)); if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) { spa->spa_scrub_active = B_FALSE; /* * If the scrub/resilver completed, update all DTLs to * reflect this. Whether it succeeded or not, vacate * all temporary scrub DTLs. * * As the scrub does not currently support traversing * data that have been freed but are part of a checkpoint, * we don't mark the scrub as done in the DTLs as faults * may still exist in those vdevs. */ if (complete && !spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) { vdev_dtl_reassess(spa->spa_root_vdev, tx->tx_txg, scn->scn_phys.scn_max_txg, B_TRUE, B_FALSE); if (scn->scn_phys.scn_min_txg) { nvlist_t *aux = fnvlist_alloc(); fnvlist_add_string(aux, ZFS_EV_RESILVER_TYPE, "healing"); spa_event_notify(spa, NULL, aux, ESC_ZFS_RESILVER_FINISH); nvlist_free(aux); } else { spa_event_notify(spa, NULL, NULL, ESC_ZFS_SCRUB_FINISH); } } else { vdev_dtl_reassess(spa->spa_root_vdev, tx->tx_txg, 0, B_TRUE, B_FALSE); } spa_errlog_rotate(spa); /* * Don't clear flag until after vdev_dtl_reassess to ensure that * DTL_MISSING will get updated when possible. */ spa->spa_scrub_started = B_FALSE; /* * We may have finished replacing a device. * Let the async thread assess this and handle the detach. */ spa_async_request(spa, SPA_ASYNC_RESILVER_DONE); /* * Clear any resilver_deferred flags in the config. * If there are drives that need resilvering, kick * off an asynchronous request to start resilver. * vdev_clear_resilver_deferred() may update the config * before the resilver can restart. In the event of * a crash during this period, the spa loading code * will find the drives that need to be resilvered * and start the resilver then. */ if (spa_feature_is_enabled(spa, SPA_FEATURE_RESILVER_DEFER) && vdev_clear_resilver_deferred(spa->spa_root_vdev, tx)) { spa_history_log_internal(spa, "starting deferred resilver", tx, "errors=%llu", (u_longlong_t)spa_get_errlog_size(spa)); spa_async_request(spa, SPA_ASYNC_RESILVER); } } scn->scn_phys.scn_end_time = gethrestime_sec(); if (spa->spa_errata == ZPOOL_ERRATA_ZOL_2094_SCRUB) spa->spa_errata = 0; ASSERT(!dsl_scan_is_running(scn)); } /* ARGSUSED */ static int dsl_scan_cancel_check(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; if (!dsl_scan_is_running(scn)) return (SET_ERROR(ENOENT)); return (0); } /* ARGSUSED */ static void dsl_scan_cancel_sync(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; dsl_scan_done(scn, B_FALSE, tx); dsl_scan_sync_state(scn, tx, SYNC_MANDATORY); spa_event_notify(scn->scn_dp->dp_spa, NULL, NULL, ESC_ZFS_SCRUB_ABORT); } int dsl_scan_cancel(dsl_pool_t *dp) { return (dsl_sync_task(spa_name(dp->dp_spa), dsl_scan_cancel_check, dsl_scan_cancel_sync, NULL, 3, ZFS_SPACE_CHECK_RESERVED)); } static int dsl_scrub_pause_resume_check(void *arg, dmu_tx_t *tx) { pool_scrub_cmd_t *cmd = arg; dsl_pool_t *dp = dmu_tx_pool(tx); dsl_scan_t *scn = dp->dp_scan; if (*cmd == POOL_SCRUB_PAUSE) { /* can't pause a scrub when there is no in-progress scrub */ if (!dsl_scan_scrubbing(dp)) return (SET_ERROR(ENOENT)); /* can't pause a paused scrub */ if (dsl_scan_is_paused_scrub(scn)) return (SET_ERROR(EBUSY)); } else if (*cmd != POOL_SCRUB_NORMAL) { return (SET_ERROR(ENOTSUP)); } return (0); } static void dsl_scrub_pause_resume_sync(void *arg, dmu_tx_t *tx) { pool_scrub_cmd_t *cmd = arg; dsl_pool_t *dp = dmu_tx_pool(tx); spa_t *spa = dp->dp_spa; dsl_scan_t *scn = dp->dp_scan; if (*cmd == POOL_SCRUB_PAUSE) { /* can't pause a scrub when there is no in-progress scrub */ spa->spa_scan_pass_scrub_pause = gethrestime_sec(); scn->scn_phys.scn_flags |= DSF_SCRUB_PAUSED; scn->scn_phys_cached.scn_flags |= DSF_SCRUB_PAUSED; dsl_scan_sync_state(scn, tx, SYNC_CACHED); spa_event_notify(spa, NULL, NULL, ESC_ZFS_SCRUB_PAUSED); spa_notify_waiters(spa); } else { ASSERT3U(*cmd, ==, POOL_SCRUB_NORMAL); if (dsl_scan_is_paused_scrub(scn)) { /* * We need to keep track of how much time we spend * paused per pass so that we can adjust the scrub rate * shown in the output of 'zpool status' */ spa->spa_scan_pass_scrub_spent_paused += gethrestime_sec() - spa->spa_scan_pass_scrub_pause; spa->spa_scan_pass_scrub_pause = 0; scn->scn_phys.scn_flags &= ~DSF_SCRUB_PAUSED; scn->scn_phys_cached.scn_flags &= ~DSF_SCRUB_PAUSED; dsl_scan_sync_state(scn, tx, SYNC_CACHED); } } } /* * Set scrub pause/resume state if it makes sense to do so */ int dsl_scrub_set_pause_resume(const dsl_pool_t *dp, pool_scrub_cmd_t cmd) { return (dsl_sync_task(spa_name(dp->dp_spa), dsl_scrub_pause_resume_check, dsl_scrub_pause_resume_sync, &cmd, 3, ZFS_SPACE_CHECK_RESERVED)); } /* start a new scan, or restart an existing one. */ void dsl_scan_restart_resilver(dsl_pool_t *dp, uint64_t txg) { if (txg == 0) { dmu_tx_t *tx; tx = dmu_tx_create_dd(dp->dp_mos_dir); VERIFY(0 == dmu_tx_assign(tx, TXG_WAIT)); txg = dmu_tx_get_txg(tx); dp->dp_scan->scn_restart_txg = txg; dmu_tx_commit(tx); } else { dp->dp_scan->scn_restart_txg = txg; } zfs_dbgmsg("restarting resilver txg=%llu", (longlong_t)txg); } void dsl_free(dsl_pool_t *dp, uint64_t txg, const blkptr_t *bp) { zio_free(dp->dp_spa, txg, bp); } void dsl_free_sync(zio_t *pio, dsl_pool_t *dp, uint64_t txg, const blkptr_t *bpp) { ASSERT(dsl_pool_sync_context(dp)); zio_nowait(zio_free_sync(pio, dp->dp_spa, txg, bpp, pio->io_flags)); } static int scan_ds_queue_compare(const void *a, const void *b) { const scan_ds_t *sds_a = a, *sds_b = b; if (sds_a->sds_dsobj < sds_b->sds_dsobj) return (-1); if (sds_a->sds_dsobj == sds_b->sds_dsobj) return (0); return (1); } static void scan_ds_queue_clear(dsl_scan_t *scn) { void *cookie = NULL; scan_ds_t *sds; while ((sds = avl_destroy_nodes(&scn->scn_queue, &cookie)) != NULL) { kmem_free(sds, sizeof (*sds)); } } static boolean_t scan_ds_queue_contains(dsl_scan_t *scn, uint64_t dsobj, uint64_t *txg) { scan_ds_t srch, *sds; srch.sds_dsobj = dsobj; sds = avl_find(&scn->scn_queue, &srch, NULL); if (sds != NULL && txg != NULL) *txg = sds->sds_txg; return (sds != NULL); } static void scan_ds_queue_insert(dsl_scan_t *scn, uint64_t dsobj, uint64_t txg) { scan_ds_t *sds; avl_index_t where; sds = kmem_zalloc(sizeof (*sds), KM_SLEEP); sds->sds_dsobj = dsobj; sds->sds_txg = txg; VERIFY3P(avl_find(&scn->scn_queue, sds, &where), ==, NULL); avl_insert(&scn->scn_queue, sds, where); } static void scan_ds_queue_remove(dsl_scan_t *scn, uint64_t dsobj) { scan_ds_t srch, *sds; srch.sds_dsobj = dsobj; sds = avl_find(&scn->scn_queue, &srch, NULL); VERIFY(sds != NULL); avl_remove(&scn->scn_queue, sds); kmem_free(sds, sizeof (*sds)); } static void scan_ds_queue_sync(dsl_scan_t *scn, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; spa_t *spa = dp->dp_spa; dmu_object_type_t ot = (spa_version(spa) >= SPA_VERSION_DSL_SCRUB) ? DMU_OT_SCAN_QUEUE : DMU_OT_ZAP_OTHER; ASSERT0(scn->scn_bytes_pending); ASSERT(scn->scn_phys.scn_queue_obj != 0); VERIFY0(dmu_object_free(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, tx)); scn->scn_phys.scn_queue_obj = zap_create(dp->dp_meta_objset, ot, DMU_OT_NONE, 0, tx); for (scan_ds_t *sds = avl_first(&scn->scn_queue); sds != NULL; sds = AVL_NEXT(&scn->scn_queue, sds)) { VERIFY0(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, sds->sds_dsobj, sds->sds_txg, tx)); } } /* * Computes the memory limit state that we're currently in. A sorted scan * needs quite a bit of memory to hold the sorting queue, so we need to * reasonably constrain the size so it doesn't impact overall system * performance. We compute two limits: * 1) Hard memory limit: if the amount of memory used by the sorting * queues on a pool gets above this value, we stop the metadata * scanning portion and start issuing the queued up and sorted * I/Os to reduce memory usage. * This limit is calculated as a fraction of physmem (by default 5%). * We constrain the lower bound of the hard limit to an absolute * minimum of zfs_scan_mem_lim_min (default: 16 MiB). We also constrain * the upper bound to 5% of the total pool size - no chance we'll * ever need that much memory, but just to keep the value in check. * 2) Soft memory limit: once we hit the hard memory limit, we start * issuing I/O to reduce queue memory usage, but we don't want to * completely empty out the queues, since we might be able to find I/Os * that will fill in the gaps of our non-sequential IOs at some point * in the future. So we stop the issuing of I/Os once the amount of * memory used drops below the soft limit (at which point we stop issuing * I/O and start scanning metadata again). * * This limit is calculated by subtracting a fraction of the hard * limit from the hard limit. By default this fraction is 5%, so * the soft limit is 95% of the hard limit. We cap the size of the * difference between the hard and soft limits at an absolute * maximum of zfs_scan_mem_lim_soft_max (default: 128 MiB) - this is * sufficient to not cause too frequent switching between the * metadata scan and I/O issue (even at 2k recordsize, 128 MiB's * worth of queues is about 1.2 GiB of on-pool data, so scanning * that should take at least a decent fraction of a second). */ static boolean_t dsl_scan_should_clear(dsl_scan_t *scn) { spa_t *spa = scn->scn_dp->dp_spa; vdev_t *rvd = scn->scn_dp->dp_spa->spa_root_vdev; uint64_t alloc, mlim_hard, mlim_soft, mused; alloc = metaslab_class_get_alloc(spa_normal_class(spa)); alloc += metaslab_class_get_alloc(spa_special_class(spa)); alloc += metaslab_class_get_alloc(spa_dedup_class(spa)); mlim_hard = MAX((physmem / zfs_scan_mem_lim_fact) * PAGESIZE, zfs_scan_mem_lim_min); mlim_hard = MIN(mlim_hard, alloc / 20); mlim_soft = mlim_hard - MIN(mlim_hard / zfs_scan_mem_lim_soft_fact, zfs_scan_mem_lim_soft_max); mused = 0; for (uint64_t i = 0; i < rvd->vdev_children; i++) { vdev_t *tvd = rvd->vdev_child[i]; dsl_scan_io_queue_t *queue; mutex_enter(&tvd->vdev_scan_io_queue_lock); queue = tvd->vdev_scan_io_queue; if (queue != NULL) { /* # extents in exts_by_size = # in exts_by_addr */ mused += zfs_btree_numnodes(&queue->q_exts_by_size) * sizeof (range_seg_gap_t) + queue->q_sio_memused; } mutex_exit(&tvd->vdev_scan_io_queue_lock); } dprintf("current scan memory usage: %llu bytes\n", (longlong_t)mused); if (mused == 0) ASSERT0(scn->scn_bytes_pending); /* * If we are above our hard limit, we need to clear out memory. * If we are below our soft limit, we need to accumulate sequential IOs. * Otherwise, we should keep doing whatever we are currently doing. */ if (mused >= mlim_hard) return (B_TRUE); else if (mused < mlim_soft) return (B_FALSE); else return (scn->scn_clearing); } static boolean_t dsl_scan_check_suspend(dsl_scan_t *scn, const zbookmark_phys_t *zb) { /* we never skip user/group accounting objects */ if (zb && (int64_t)zb->zb_object < 0) return (B_FALSE); if (scn->scn_suspending) return (B_TRUE); /* we're already suspending */ if (!ZB_IS_ZERO(&scn->scn_phys.scn_bookmark)) return (B_FALSE); /* we're resuming */ /* We only know how to resume from level-0 and objset blocks. */ if (zb && (zb->zb_level != 0 && zb->zb_level != ZB_ROOT_LEVEL)) return (B_FALSE); /* * We suspend if: * - we have scanned for at least the minimum time (default 1 sec * for scrub, 3 sec for resilver), and either we have sufficient * dirty data that we are starting to write more quickly * (default 30%), someone is explicitly waiting for this txg * to complete, or we have used up all of the time in the txg * timeout (default 5 sec). * or * - the spa is shutting down because this pool is being exported * or the machine is rebooting. * or * - the scan queue has reached its memory use limit */ uint64_t curr_time_ns = gethrtime(); uint64_t scan_time_ns = curr_time_ns - scn->scn_sync_start_time; uint64_t sync_time_ns = curr_time_ns - scn->scn_dp->dp_spa->spa_sync_starttime; int dirty_pct = scn->scn_dp->dp_dirty_total * 100 / zfs_dirty_data_max; int mintime = (scn->scn_phys.scn_func == POOL_SCAN_RESILVER) ? zfs_resilver_min_time_ms : zfs_scrub_min_time_ms; if ((NSEC2MSEC(scan_time_ns) > mintime && (dirty_pct >= zfs_vdev_async_write_active_min_dirty_percent || txg_sync_waiting(scn->scn_dp) || NSEC2SEC(sync_time_ns) >= zfs_txg_timeout)) || spa_shutting_down(scn->scn_dp->dp_spa) || (zfs_scan_strict_mem_lim && dsl_scan_should_clear(scn))) { if (zb && zb->zb_level == ZB_ROOT_LEVEL) { dprintf("suspending at first available bookmark " "%llx/%llx/%llx/%llx\n", (longlong_t)zb->zb_objset, (longlong_t)zb->zb_object, (longlong_t)zb->zb_level, (longlong_t)zb->zb_blkid); SET_BOOKMARK(&scn->scn_phys.scn_bookmark, zb->zb_objset, 0, 0, 0); } else if (zb != NULL) { dprintf("suspending at bookmark %llx/%llx/%llx/%llx\n", (longlong_t)zb->zb_objset, (longlong_t)zb->zb_object, (longlong_t)zb->zb_level, (longlong_t)zb->zb_blkid); scn->scn_phys.scn_bookmark = *zb; } else { #ifdef ZFS_DEBUG dsl_scan_phys_t *scnp = &scn->scn_phys; dprintf("suspending at at DDT bookmark " "%llx/%llx/%llx/%llx\n", (longlong_t)scnp->scn_ddt_bookmark.ddb_class, (longlong_t)scnp->scn_ddt_bookmark.ddb_type, (longlong_t)scnp->scn_ddt_bookmark.ddb_checksum, (longlong_t)scnp->scn_ddt_bookmark.ddb_cursor); #endif } scn->scn_suspending = B_TRUE; return (B_TRUE); } return (B_FALSE); } typedef struct zil_scan_arg { dsl_pool_t *zsa_dp; zil_header_t *zsa_zh; } zil_scan_arg_t; /* ARGSUSED */ static int dsl_scan_zil_block(zilog_t *zilog, const blkptr_t *bp, void *arg, uint64_t claim_txg) { zil_scan_arg_t *zsa = arg; dsl_pool_t *dp = zsa->zsa_dp; dsl_scan_t *scn = dp->dp_scan; zil_header_t *zh = zsa->zsa_zh; zbookmark_phys_t zb; ASSERT(!BP_IS_REDACTED(bp)); if (BP_IS_HOLE(bp) || bp->blk_birth <= scn->scn_phys.scn_cur_min_txg) return (0); /* * One block ("stubby") can be allocated a long time ago; we * want to visit that one because it has been allocated * (on-disk) even if it hasn't been claimed (even though for * scrub there's nothing to do to it). */ if (claim_txg == 0 && bp->blk_birth >= spa_min_claim_txg(dp->dp_spa)) return (0); SET_BOOKMARK(&zb, zh->zh_log.blk_cksum.zc_word[ZIL_ZC_OBJSET], ZB_ZIL_OBJECT, ZB_ZIL_LEVEL, bp->blk_cksum.zc_word[ZIL_ZC_SEQ]); VERIFY(0 == scan_funcs[scn->scn_phys.scn_func](dp, bp, &zb)); return (0); } /* ARGSUSED */ static int dsl_scan_zil_record(zilog_t *zilog, const lr_t *lrc, void *arg, uint64_t claim_txg) { if (lrc->lrc_txtype == TX_WRITE) { zil_scan_arg_t *zsa = arg; dsl_pool_t *dp = zsa->zsa_dp; dsl_scan_t *scn = dp->dp_scan; zil_header_t *zh = zsa->zsa_zh; const lr_write_t *lr = (const lr_write_t *)lrc; const blkptr_t *bp = &lr->lr_blkptr; zbookmark_phys_t zb; ASSERT(!BP_IS_REDACTED(bp)); if (BP_IS_HOLE(bp) || bp->blk_birth <= scn->scn_phys.scn_cur_min_txg) return (0); /* * birth can be < claim_txg if this record's txg is * already txg sync'ed (but this log block contains * other records that are not synced) */ if (claim_txg == 0 || bp->blk_birth < claim_txg) return (0); SET_BOOKMARK(&zb, zh->zh_log.blk_cksum.zc_word[ZIL_ZC_OBJSET], lr->lr_foid, ZB_ZIL_LEVEL, lr->lr_offset / BP_GET_LSIZE(bp)); VERIFY(0 == scan_funcs[scn->scn_phys.scn_func](dp, bp, &zb)); } return (0); } static void dsl_scan_zil(dsl_pool_t *dp, zil_header_t *zh) { uint64_t claim_txg = zh->zh_claim_txg; zil_scan_arg_t zsa = { dp, zh }; zilog_t *zilog; ASSERT(spa_writeable(dp->dp_spa)); /* * We only want to visit blocks that have been claimed but not yet * replayed (or, in read-only mode, blocks that *would* be claimed). */ if (claim_txg == 0) return; zilog = zil_alloc(dp->dp_meta_objset, zh); (void) zil_parse(zilog, dsl_scan_zil_block, dsl_scan_zil_record, &zsa, claim_txg, B_FALSE); zil_free(zilog); } /* * We compare scan_prefetch_issue_ctx_t's based on their bookmarks. The idea * here is to sort the AVL tree by the order each block will be needed. */ static int scan_prefetch_queue_compare(const void *a, const void *b) { const scan_prefetch_issue_ctx_t *spic_a = a, *spic_b = b; const scan_prefetch_ctx_t *spc_a = spic_a->spic_spc; const scan_prefetch_ctx_t *spc_b = spic_b->spic_spc; return (zbookmark_compare(spc_a->spc_datablkszsec, spc_a->spc_indblkshift, spc_b->spc_datablkszsec, spc_b->spc_indblkshift, &spic_a->spic_zb, &spic_b->spic_zb)); } static void scan_prefetch_ctx_rele(scan_prefetch_ctx_t *spc, void *tag) { if (zfs_refcount_remove(&spc->spc_refcnt, tag) == 0) { zfs_refcount_destroy(&spc->spc_refcnt); kmem_free(spc, sizeof (scan_prefetch_ctx_t)); } } static scan_prefetch_ctx_t * scan_prefetch_ctx_create(dsl_scan_t *scn, dnode_phys_t *dnp, void *tag) { scan_prefetch_ctx_t *spc; spc = kmem_alloc(sizeof (scan_prefetch_ctx_t), KM_SLEEP); zfs_refcount_create(&spc->spc_refcnt); zfs_refcount_add(&spc->spc_refcnt, tag); spc->spc_scn = scn; if (dnp != NULL) { spc->spc_datablkszsec = dnp->dn_datablkszsec; spc->spc_indblkshift = dnp->dn_indblkshift; spc->spc_root = B_FALSE; } else { spc->spc_datablkszsec = 0; spc->spc_indblkshift = 0; spc->spc_root = B_TRUE; } return (spc); } static void scan_prefetch_ctx_add_ref(scan_prefetch_ctx_t *spc, void *tag) { zfs_refcount_add(&spc->spc_refcnt, tag); } static void scan_ds_prefetch_queue_clear(dsl_scan_t *scn) { spa_t *spa = scn->scn_dp->dp_spa; void *cookie = NULL; scan_prefetch_issue_ctx_t *spic = NULL; mutex_enter(&spa->spa_scrub_lock); while ((spic = avl_destroy_nodes(&scn->scn_prefetch_queue, &cookie)) != NULL) { scan_prefetch_ctx_rele(spic->spic_spc, scn); kmem_free(spic, sizeof (scan_prefetch_issue_ctx_t)); } mutex_exit(&spa->spa_scrub_lock); } static boolean_t dsl_scan_check_prefetch_resume(scan_prefetch_ctx_t *spc, const zbookmark_phys_t *zb) { zbookmark_phys_t *last_zb = &spc->spc_scn->scn_prefetch_bookmark; dnode_phys_t tmp_dnp; dnode_phys_t *dnp = (spc->spc_root) ? NULL : &tmp_dnp; if (zb->zb_objset != last_zb->zb_objset) return (B_TRUE); if ((int64_t)zb->zb_object < 0) return (B_FALSE); tmp_dnp.dn_datablkszsec = spc->spc_datablkszsec; tmp_dnp.dn_indblkshift = spc->spc_indblkshift; if (zbookmark_subtree_completed(dnp, zb, last_zb)) return (B_TRUE); return (B_FALSE); } static void dsl_scan_prefetch(scan_prefetch_ctx_t *spc, blkptr_t *bp, zbookmark_phys_t *zb) { avl_index_t idx; dsl_scan_t *scn = spc->spc_scn; spa_t *spa = scn->scn_dp->dp_spa; scan_prefetch_issue_ctx_t *spic; if (zfs_no_scrub_prefetch || BP_IS_REDACTED(bp)) return; if (BP_IS_HOLE(bp) || bp->blk_birth <= scn->scn_phys.scn_cur_min_txg || (BP_GET_LEVEL(bp) == 0 && BP_GET_TYPE(bp) != DMU_OT_DNODE && BP_GET_TYPE(bp) != DMU_OT_OBJSET)) return; if (dsl_scan_check_prefetch_resume(spc, zb)) return; scan_prefetch_ctx_add_ref(spc, scn); spic = kmem_alloc(sizeof (scan_prefetch_issue_ctx_t), KM_SLEEP); spic->spic_spc = spc; spic->spic_bp = *bp; spic->spic_zb = *zb; /* * Add the IO to the queue of blocks to prefetch. This allows us to * prioritize blocks that we will need first for the main traversal * thread. */ mutex_enter(&spa->spa_scrub_lock); if (avl_find(&scn->scn_prefetch_queue, spic, &idx) != NULL) { /* this block is already queued for prefetch */ kmem_free(spic, sizeof (scan_prefetch_issue_ctx_t)); scan_prefetch_ctx_rele(spc, scn); mutex_exit(&spa->spa_scrub_lock); return; } avl_insert(&scn->scn_prefetch_queue, spic, idx); cv_broadcast(&spa->spa_scrub_io_cv); mutex_exit(&spa->spa_scrub_lock); } static void dsl_scan_prefetch_dnode(dsl_scan_t *scn, dnode_phys_t *dnp, uint64_t objset, uint64_t object) { int i; zbookmark_phys_t zb; scan_prefetch_ctx_t *spc; if (dnp->dn_nblkptr == 0 && !(dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR)) return; SET_BOOKMARK(&zb, objset, object, 0, 0); spc = scan_prefetch_ctx_create(scn, dnp, FTAG); for (i = 0; i < dnp->dn_nblkptr; i++) { zb.zb_level = BP_GET_LEVEL(&dnp->dn_blkptr[i]); zb.zb_blkid = i; dsl_scan_prefetch(spc, &dnp->dn_blkptr[i], &zb); } if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) { zb.zb_level = 0; zb.zb_blkid = DMU_SPILL_BLKID; dsl_scan_prefetch(spc, DN_SPILL_BLKPTR(dnp), &zb); } scan_prefetch_ctx_rele(spc, FTAG); } static void dsl_scan_prefetch_cb(zio_t *zio, const zbookmark_phys_t *zb, const blkptr_t *bp, arc_buf_t *buf, void *private) { scan_prefetch_ctx_t *spc = private; dsl_scan_t *scn = spc->spc_scn; spa_t *spa = scn->scn_dp->dp_spa; /* broadcast that the IO has completed for rate limiting purposes */ mutex_enter(&spa->spa_scrub_lock); ASSERT3U(spa->spa_scrub_inflight, >=, BP_GET_PSIZE(bp)); spa->spa_scrub_inflight -= BP_GET_PSIZE(bp); cv_broadcast(&spa->spa_scrub_io_cv); mutex_exit(&spa->spa_scrub_lock); /* if there was an error or we are done prefetching, just cleanup */ if (buf == NULL || scn->scn_prefetch_stop) goto out; if (BP_GET_LEVEL(bp) > 0) { int i; blkptr_t *cbp; int epb = BP_GET_LSIZE(bp) >> SPA_BLKPTRSHIFT; zbookmark_phys_t czb; for (i = 0, cbp = buf->b_data; i < epb; i++, cbp++) { SET_BOOKMARK(&czb, zb->zb_objset, zb->zb_object, zb->zb_level - 1, zb->zb_blkid * epb + i); dsl_scan_prefetch(spc, cbp, &czb); } } else if (BP_GET_TYPE(bp) == DMU_OT_DNODE) { dnode_phys_t *cdnp; int i; int epb = BP_GET_LSIZE(bp) >> DNODE_SHIFT; for (i = 0, cdnp = buf->b_data; i < epb; i += cdnp->dn_extra_slots + 1, cdnp += cdnp->dn_extra_slots + 1) { dsl_scan_prefetch_dnode(scn, cdnp, zb->zb_objset, zb->zb_blkid * epb + i); } } else if (BP_GET_TYPE(bp) == DMU_OT_OBJSET) { objset_phys_t *osp = buf->b_data; dsl_scan_prefetch_dnode(scn, &osp->os_meta_dnode, zb->zb_objset, DMU_META_DNODE_OBJECT); if (OBJSET_BUF_HAS_USERUSED(buf)) { dsl_scan_prefetch_dnode(scn, &osp->os_groupused_dnode, zb->zb_objset, DMU_GROUPUSED_OBJECT); dsl_scan_prefetch_dnode(scn, &osp->os_userused_dnode, zb->zb_objset, DMU_USERUSED_OBJECT); } } out: if (buf != NULL) arc_buf_destroy(buf, private); scan_prefetch_ctx_rele(spc, scn); } /* ARGSUSED */ static void dsl_scan_prefetch_thread(void *arg) { dsl_scan_t *scn = arg; spa_t *spa = scn->scn_dp->dp_spa; scan_prefetch_issue_ctx_t *spic; /* loop until we are told to stop */ while (!scn->scn_prefetch_stop) { arc_flags_t flags = ARC_FLAG_NOWAIT | ARC_FLAG_PRESCIENT_PREFETCH | ARC_FLAG_PREFETCH; int zio_flags = ZIO_FLAG_CANFAIL | ZIO_FLAG_SCAN_THREAD; mutex_enter(&spa->spa_scrub_lock); /* * Wait until we have an IO to issue and are not above our * maximum in flight limit. */ while (!scn->scn_prefetch_stop && (avl_numnodes(&scn->scn_prefetch_queue) == 0 || spa->spa_scrub_inflight >= scn->scn_maxinflight_bytes)) { cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock); } /* recheck if we should stop since we waited for the cv */ if (scn->scn_prefetch_stop) { mutex_exit(&spa->spa_scrub_lock); break; } /* remove the prefetch IO from the tree */ spic = avl_first(&scn->scn_prefetch_queue); spa->spa_scrub_inflight += BP_GET_PSIZE(&spic->spic_bp); avl_remove(&scn->scn_prefetch_queue, spic); mutex_exit(&spa->spa_scrub_lock); if (BP_IS_PROTECTED(&spic->spic_bp)) { ASSERT(BP_GET_TYPE(&spic->spic_bp) == DMU_OT_DNODE || BP_GET_TYPE(&spic->spic_bp) == DMU_OT_OBJSET); ASSERT3U(BP_GET_LEVEL(&spic->spic_bp), ==, 0); zio_flags |= ZIO_FLAG_RAW; } /* issue the prefetch asynchronously */ (void) arc_read(scn->scn_zio_root, scn->scn_dp->dp_spa, &spic->spic_bp, dsl_scan_prefetch_cb, spic->spic_spc, ZIO_PRIORITY_SCRUB, zio_flags, &flags, &spic->spic_zb); kmem_free(spic, sizeof (scan_prefetch_issue_ctx_t)); } ASSERT(scn->scn_prefetch_stop); /* free any prefetches we didn't get to complete */ mutex_enter(&spa->spa_scrub_lock); while ((spic = avl_first(&scn->scn_prefetch_queue)) != NULL) { avl_remove(&scn->scn_prefetch_queue, spic); scan_prefetch_ctx_rele(spic->spic_spc, scn); kmem_free(spic, sizeof (scan_prefetch_issue_ctx_t)); } ASSERT0(avl_numnodes(&scn->scn_prefetch_queue)); mutex_exit(&spa->spa_scrub_lock); } static boolean_t dsl_scan_check_resume(dsl_scan_t *scn, const dnode_phys_t *dnp, const zbookmark_phys_t *zb) { /* * We never skip over user/group accounting objects (obj<0) */ if (!ZB_IS_ZERO(&scn->scn_phys.scn_bookmark) && (int64_t)zb->zb_object >= 0) { /* * If we already visited this bp & everything below (in * a prior txg sync), don't bother doing it again. */ if (zbookmark_subtree_completed(dnp, zb, &scn->scn_phys.scn_bookmark)) return (B_TRUE); /* * If we found the block we're trying to resume from, or * we went past it to a different object, zero it out to * indicate that it's OK to start checking for suspending * again. */ if (bcmp(zb, &scn->scn_phys.scn_bookmark, sizeof (*zb)) == 0 || zb->zb_object > scn->scn_phys.scn_bookmark.zb_object) { dprintf("resuming at %llx/%llx/%llx/%llx\n", (longlong_t)zb->zb_objset, (longlong_t)zb->zb_object, (longlong_t)zb->zb_level, (longlong_t)zb->zb_blkid); bzero(&scn->scn_phys.scn_bookmark, sizeof (*zb)); } } return (B_FALSE); } static void dsl_scan_visitbp(blkptr_t *bp, const zbookmark_phys_t *zb, dnode_phys_t *dnp, dsl_dataset_t *ds, dsl_scan_t *scn, dmu_objset_type_t ostype, dmu_tx_t *tx); inline __attribute__((always_inline)) static void dsl_scan_visitdnode( dsl_scan_t *, dsl_dataset_t *ds, dmu_objset_type_t ostype, dnode_phys_t *dnp, uint64_t object, dmu_tx_t *tx); /* * Return nonzero on i/o error. * Return new buf to write out in *bufp. */ inline __attribute__((always_inline)) static int dsl_scan_recurse(dsl_scan_t *scn, dsl_dataset_t *ds, dmu_objset_type_t ostype, dnode_phys_t *dnp, const blkptr_t *bp, const zbookmark_phys_t *zb, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; int zio_flags = ZIO_FLAG_CANFAIL | ZIO_FLAG_SCAN_THREAD; int err; ASSERT(!BP_IS_REDACTED(bp)); if (BP_GET_LEVEL(bp) > 0) { arc_flags_t flags = ARC_FLAG_WAIT; int i; blkptr_t *cbp; int epb = BP_GET_LSIZE(bp) >> SPA_BLKPTRSHIFT; arc_buf_t *buf; err = arc_read(NULL, dp->dp_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_SCRUB, zio_flags, &flags, zb); if (err) { scn->scn_phys.scn_errors++; return (err); } for (i = 0, cbp = buf->b_data; i < epb; i++, cbp++) { zbookmark_phys_t czb; SET_BOOKMARK(&czb, zb->zb_objset, zb->zb_object, zb->zb_level - 1, zb->zb_blkid * epb + i); dsl_scan_visitbp(cbp, &czb, dnp, ds, scn, ostype, tx); } arc_buf_destroy(buf, &buf); } else if (BP_GET_TYPE(bp) == DMU_OT_DNODE) { arc_flags_t flags = ARC_FLAG_WAIT; dnode_phys_t *cdnp; int i; int epb = BP_GET_LSIZE(bp) >> DNODE_SHIFT; arc_buf_t *buf; if (BP_IS_PROTECTED(bp)) { ASSERT3U(BP_GET_COMPRESS(bp), ==, ZIO_COMPRESS_OFF); zio_flags |= ZIO_FLAG_RAW; } err = arc_read(NULL, dp->dp_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_SCRUB, zio_flags, &flags, zb); if (err) { scn->scn_phys.scn_errors++; return (err); } for (i = 0, cdnp = buf->b_data; i < epb; i += cdnp->dn_extra_slots + 1, cdnp += cdnp->dn_extra_slots + 1) { dsl_scan_visitdnode(scn, ds, ostype, cdnp, zb->zb_blkid * epb + i, tx); } arc_buf_destroy(buf, &buf); } else if (BP_GET_TYPE(bp) == DMU_OT_OBJSET) { arc_flags_t flags = ARC_FLAG_WAIT; objset_phys_t *osp; arc_buf_t *buf; err = arc_read(NULL, dp->dp_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_SCRUB, zio_flags, &flags, zb); if (err) { scn->scn_phys.scn_errors++; return (err); } osp = buf->b_data; dsl_scan_visitdnode(scn, ds, osp->os_type, &osp->os_meta_dnode, DMU_META_DNODE_OBJECT, tx); if (OBJSET_BUF_HAS_USERUSED(buf)) { /* * We also always visit user/group/project accounting * objects, and never skip them, even if we are * suspending. This is necessary so that the * space deltas from this txg get integrated. */ if (OBJSET_BUF_HAS_PROJECTUSED(buf)) dsl_scan_visitdnode(scn, ds, osp->os_type, &osp->os_projectused_dnode, DMU_PROJECTUSED_OBJECT, tx); dsl_scan_visitdnode(scn, ds, osp->os_type, &osp->os_groupused_dnode, DMU_GROUPUSED_OBJECT, tx); dsl_scan_visitdnode(scn, ds, osp->os_type, &osp->os_userused_dnode, DMU_USERUSED_OBJECT, tx); } arc_buf_destroy(buf, &buf); } return (0); } inline __attribute__((always_inline)) static void dsl_scan_visitdnode(dsl_scan_t *scn, dsl_dataset_t *ds, dmu_objset_type_t ostype, dnode_phys_t *dnp, uint64_t object, dmu_tx_t *tx) { int j; for (j = 0; j < dnp->dn_nblkptr; j++) { zbookmark_phys_t czb; SET_BOOKMARK(&czb, ds ? ds->ds_object : 0, object, dnp->dn_nlevels - 1, j); dsl_scan_visitbp(&dnp->dn_blkptr[j], &czb, dnp, ds, scn, ostype, tx); } if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) { zbookmark_phys_t czb; SET_BOOKMARK(&czb, ds ? ds->ds_object : 0, object, 0, DMU_SPILL_BLKID); dsl_scan_visitbp(DN_SPILL_BLKPTR(dnp), &czb, dnp, ds, scn, ostype, tx); } } /* * The arguments are in this order because mdb can only print the * first 5; we want them to be useful. */ static void dsl_scan_visitbp(blkptr_t *bp, const zbookmark_phys_t *zb, dnode_phys_t *dnp, dsl_dataset_t *ds, dsl_scan_t *scn, dmu_objset_type_t ostype, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; blkptr_t *bp_toread = NULL; if (dsl_scan_check_suspend(scn, zb)) return; if (dsl_scan_check_resume(scn, dnp, zb)) return; scn->scn_visited_this_txg++; /* * This debugging is commented out to conserve stack space. This * function is called recursively and the debugging adds several * bytes to the stack for each call. It can be commented back in * if required to debug an issue in dsl_scan_visitbp(). * * dprintf_bp(bp, * "visiting ds=%p/%llu zb=%llx/%llx/%llx/%llx bp=%p", * ds, ds ? ds->ds_object : 0, * zb->zb_objset, zb->zb_object, zb->zb_level, zb->zb_blkid, * bp); */ if (BP_IS_HOLE(bp)) { scn->scn_holes_this_txg++; return; } if (BP_IS_REDACTED(bp)) { ASSERT(dsl_dataset_feature_is_active(ds, SPA_FEATURE_REDACTED_DATASETS)); return; } if (bp->blk_birth <= scn->scn_phys.scn_cur_min_txg) { scn->scn_lt_min_this_txg++; return; } bp_toread = kmem_alloc(sizeof (blkptr_t), KM_SLEEP); *bp_toread = *bp; if (dsl_scan_recurse(scn, ds, ostype, dnp, bp_toread, zb, tx) != 0) goto out; /* * If dsl_scan_ddt() has already visited this block, it will have * already done any translations or scrubbing, so don't call the * callback again. */ if (ddt_class_contains(dp->dp_spa, scn->scn_phys.scn_ddt_class_max, bp)) { scn->scn_ddt_contained_this_txg++; goto out; } /* * If this block is from the future (after cur_max_txg), then we * are doing this on behalf of a deleted snapshot, and we will * revisit the future block on the next pass of this dataset. * Don't scan it now unless we need to because something * under it was modified. */ if (BP_PHYSICAL_BIRTH(bp) > scn->scn_phys.scn_cur_max_txg) { scn->scn_gt_max_this_txg++; goto out; } scan_funcs[scn->scn_phys.scn_func](dp, bp, zb); out: kmem_free(bp_toread, sizeof (blkptr_t)); } static void dsl_scan_visit_rootbp(dsl_scan_t *scn, dsl_dataset_t *ds, blkptr_t *bp, dmu_tx_t *tx) { zbookmark_phys_t zb; scan_prefetch_ctx_t *spc; SET_BOOKMARK(&zb, ds ? ds->ds_object : DMU_META_OBJSET, ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID); if (ZB_IS_ZERO(&scn->scn_phys.scn_bookmark)) { SET_BOOKMARK(&scn->scn_prefetch_bookmark, zb.zb_objset, 0, 0, 0); } else { scn->scn_prefetch_bookmark = scn->scn_phys.scn_bookmark; } scn->scn_objsets_visited_this_txg++; spc = scan_prefetch_ctx_create(scn, NULL, FTAG); dsl_scan_prefetch(spc, bp, &zb); scan_prefetch_ctx_rele(spc, FTAG); dsl_scan_visitbp(bp, &zb, NULL, ds, scn, DMU_OST_NONE, tx); dprintf_ds(ds, "finished scan%s", ""); } static void ds_destroyed_scn_phys(dsl_dataset_t *ds, dsl_scan_phys_t *scn_phys) { if (scn_phys->scn_bookmark.zb_objset == ds->ds_object) { if (ds->ds_is_snapshot) { /* * Note: * - scn_cur_{min,max}_txg stays the same. * - Setting the flag is not really necessary if * scn_cur_max_txg == scn_max_txg, because there * is nothing after this snapshot that we care * about. However, we set it anyway and then * ignore it when we retraverse it in * dsl_scan_visitds(). */ scn_phys->scn_bookmark.zb_objset = dsl_dataset_phys(ds)->ds_next_snap_obj; zfs_dbgmsg("destroying ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)dsl_dataset_phys(ds)-> ds_next_snap_obj); scn_phys->scn_flags |= DSF_VISIT_DS_AGAIN; } else { SET_BOOKMARK(&scn_phys->scn_bookmark, ZB_DESTROYED_OBJSET, 0, 0, 0); zfs_dbgmsg("destroying ds %llu; currently traversing; " "reset bookmark to -1,0,0,0", (u_longlong_t)ds->ds_object); } } } /* * Invoked when a dataset is destroyed. We need to make sure that: * * 1) If it is the dataset that was currently being scanned, we write * a new dsl_scan_phys_t and marking the objset reference in it * as destroyed. * 2) Remove it from the work queue, if it was present. * * If the dataset was actually a snapshot, instead of marking the dataset * as destroyed, we instead substitute the next snapshot in line. */ void dsl_scan_ds_destroyed(dsl_dataset_t *ds, dmu_tx_t *tx) { dsl_pool_t *dp = ds->ds_dir->dd_pool; dsl_scan_t *scn = dp->dp_scan; uint64_t mintxg; if (!dsl_scan_is_running(scn)) return; ds_destroyed_scn_phys(ds, &scn->scn_phys); ds_destroyed_scn_phys(ds, &scn->scn_phys_cached); if (scan_ds_queue_contains(scn, ds->ds_object, &mintxg)) { scan_ds_queue_remove(scn, ds->ds_object); if (ds->ds_is_snapshot) scan_ds_queue_insert(scn, dsl_dataset_phys(ds)->ds_next_snap_obj, mintxg); } if (zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, &mintxg) == 0) { ASSERT3U(dsl_dataset_phys(ds)->ds_num_children, <=, 1); VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, tx)); if (ds->ds_is_snapshot) { /* * We keep the same mintxg; it could be > * ds_creation_txg if the previous snapshot was * deleted too. */ VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, dsl_dataset_phys(ds)->ds_next_snap_obj, mintxg, tx) == 0); zfs_dbgmsg("destroying ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)dsl_dataset_phys(ds)-> ds_next_snap_obj); } else { zfs_dbgmsg("destroying ds %llu; in queue; removing", (u_longlong_t)ds->ds_object); } } /* * dsl_scan_sync() should be called after this, and should sync * out our changed state, but just to be safe, do it here. */ dsl_scan_sync_state(scn, tx, SYNC_CACHED); } static void ds_snapshotted_bookmark(dsl_dataset_t *ds, zbookmark_phys_t *scn_bookmark) { if (scn_bookmark->zb_objset == ds->ds_object) { scn_bookmark->zb_objset = dsl_dataset_phys(ds)->ds_prev_snap_obj; zfs_dbgmsg("snapshotting ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)dsl_dataset_phys(ds)->ds_prev_snap_obj); } } /* * Called when a dataset is snapshotted. If we were currently traversing * this snapshot, we reset our bookmark to point at the newly created * snapshot. We also modify our work queue to remove the old snapshot and * replace with the new one. */ void dsl_scan_ds_snapshotted(dsl_dataset_t *ds, dmu_tx_t *tx) { dsl_pool_t *dp = ds->ds_dir->dd_pool; dsl_scan_t *scn = dp->dp_scan; uint64_t mintxg; if (!dsl_scan_is_running(scn)) return; ASSERT(dsl_dataset_phys(ds)->ds_prev_snap_obj != 0); ds_snapshotted_bookmark(ds, &scn->scn_phys.scn_bookmark); ds_snapshotted_bookmark(ds, &scn->scn_phys_cached.scn_bookmark); if (scan_ds_queue_contains(scn, ds->ds_object, &mintxg)) { scan_ds_queue_remove(scn, ds->ds_object); scan_ds_queue_insert(scn, dsl_dataset_phys(ds)->ds_prev_snap_obj, mintxg); } if (zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, &mintxg) == 0) { VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, tx)); VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, dsl_dataset_phys(ds)->ds_prev_snap_obj, mintxg, tx) == 0); zfs_dbgmsg("snapshotting ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)dsl_dataset_phys(ds)->ds_prev_snap_obj); } dsl_scan_sync_state(scn, tx, SYNC_CACHED); } static void ds_clone_swapped_bookmark(dsl_dataset_t *ds1, dsl_dataset_t *ds2, zbookmark_phys_t *scn_bookmark) { if (scn_bookmark->zb_objset == ds1->ds_object) { scn_bookmark->zb_objset = ds2->ds_object; zfs_dbgmsg("clone_swap ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds1->ds_object, (u_longlong_t)ds2->ds_object); } else if (scn_bookmark->zb_objset == ds2->ds_object) { scn_bookmark->zb_objset = ds1->ds_object; zfs_dbgmsg("clone_swap ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds2->ds_object, (u_longlong_t)ds1->ds_object); } } /* * Called when an origin dataset and its clone are swapped. If we were * currently traversing the dataset, we need to switch to traversing the * newly promoted clone. */ void dsl_scan_ds_clone_swapped(dsl_dataset_t *ds1, dsl_dataset_t *ds2, dmu_tx_t *tx) { dsl_pool_t *dp = ds1->ds_dir->dd_pool; dsl_scan_t *scn = dp->dp_scan; uint64_t mintxg1, mintxg2; boolean_t ds1_queued, ds2_queued; if (!dsl_scan_is_running(scn)) return; ds_clone_swapped_bookmark(ds1, ds2, &scn->scn_phys.scn_bookmark); ds_clone_swapped_bookmark(ds1, ds2, &scn->scn_phys_cached.scn_bookmark); /* * Handle the in-memory scan queue. */ ds1_queued = scan_ds_queue_contains(scn, ds1->ds_object, &mintxg1); ds2_queued = scan_ds_queue_contains(scn, ds2->ds_object, &mintxg2); /* Sanity checking. */ if (ds1_queued) { ASSERT3U(mintxg1, ==, dsl_dataset_phys(ds1)->ds_prev_snap_txg); ASSERT3U(mintxg1, ==, dsl_dataset_phys(ds2)->ds_prev_snap_txg); } if (ds2_queued) { ASSERT3U(mintxg2, ==, dsl_dataset_phys(ds1)->ds_prev_snap_txg); ASSERT3U(mintxg2, ==, dsl_dataset_phys(ds2)->ds_prev_snap_txg); } if (ds1_queued && ds2_queued) { /* * If both are queued, we don't need to do anything. * The swapping code below would not handle this case correctly, * since we can't insert ds2 if it is already there. That's * because scan_ds_queue_insert() prohibits a duplicate insert * and panics. */ } else if (ds1_queued) { scan_ds_queue_remove(scn, ds1->ds_object); scan_ds_queue_insert(scn, ds2->ds_object, mintxg1); } else if (ds2_queued) { scan_ds_queue_remove(scn, ds2->ds_object); scan_ds_queue_insert(scn, ds1->ds_object, mintxg2); } /* * Handle the on-disk scan queue. * The on-disk state is an out-of-date version of the in-memory state, * so the in-memory and on-disk values for ds1_queued and ds2_queued may * be different. Therefore we need to apply the swap logic to the * on-disk state independently of the in-memory state. */ ds1_queued = zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, &mintxg1) == 0; ds2_queued = zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds2->ds_object, &mintxg2) == 0; /* Sanity checking. */ if (ds1_queued) { ASSERT3U(mintxg1, ==, dsl_dataset_phys(ds1)->ds_prev_snap_txg); ASSERT3U(mintxg1, ==, dsl_dataset_phys(ds2)->ds_prev_snap_txg); } if (ds2_queued) { ASSERT3U(mintxg2, ==, dsl_dataset_phys(ds1)->ds_prev_snap_txg); ASSERT3U(mintxg2, ==, dsl_dataset_phys(ds2)->ds_prev_snap_txg); } if (ds1_queued && ds2_queued) { /* * If both are queued, we don't need to do anything. * Alternatively, we could check for EEXIST from * zap_add_int_key() and back out to the original state, but * that would be more work than checking for this case upfront. */ } else if (ds1_queued) { VERIFY3S(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, tx)); VERIFY3S(0, ==, zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds2->ds_object, mintxg1, tx)); zfs_dbgmsg("clone_swap ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds1->ds_object, (u_longlong_t)ds2->ds_object); } else if (ds2_queued) { VERIFY3S(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds2->ds_object, tx)); VERIFY3S(0, ==, zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, mintxg2, tx)); zfs_dbgmsg("clone_swap ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds2->ds_object, (u_longlong_t)ds1->ds_object); } dsl_scan_sync_state(scn, tx, SYNC_CACHED); } /* ARGSUSED */ static int enqueue_clones_cb(dsl_pool_t *dp, dsl_dataset_t *hds, void *arg) { uint64_t originobj = *(uint64_t *)arg; dsl_dataset_t *ds; int err; dsl_scan_t *scn = dp->dp_scan; if (dsl_dir_phys(hds->ds_dir)->dd_origin_obj != originobj) return (0); err = dsl_dataset_hold_obj(dp, hds->ds_object, FTAG, &ds); if (err) return (err); while (dsl_dataset_phys(ds)->ds_prev_snap_obj != originobj) { dsl_dataset_t *prev; err = dsl_dataset_hold_obj(dp, dsl_dataset_phys(ds)->ds_prev_snap_obj, FTAG, &prev); dsl_dataset_rele(ds, FTAG); if (err) return (err); ds = prev; } scan_ds_queue_insert(scn, ds->ds_object, dsl_dataset_phys(ds)->ds_prev_snap_txg); dsl_dataset_rele(ds, FTAG); return (0); } static void dsl_scan_visitds(dsl_scan_t *scn, uint64_t dsobj, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; dsl_dataset_t *ds; VERIFY3U(0, ==, dsl_dataset_hold_obj(dp, dsobj, FTAG, &ds)); if (scn->scn_phys.scn_cur_min_txg >= scn->scn_phys.scn_max_txg) { /* * This can happen if this snapshot was created after the * scan started, and we already completed a previous snapshot * that was created after the scan started. This snapshot * only references blocks with: * * birth < our ds_creation_txg * cur_min_txg is no less than ds_creation_txg. * We have already visited these blocks. * or * birth > scn_max_txg * The scan requested not to visit these blocks. * * Subsequent snapshots (and clones) can reference our * blocks, or blocks with even higher birth times. * Therefore we do not need to visit them either, * so we do not add them to the work queue. * * Note that checking for cur_min_txg >= cur_max_txg * is not sufficient, because in that case we may need to * visit subsequent snapshots. This happens when min_txg > 0, * which raises cur_min_txg. In this case we will visit * this dataset but skip all of its blocks, because the * rootbp's birth time is < cur_min_txg. Then we will * add the next snapshots/clones to the work queue. */ char *dsname = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP); dsl_dataset_name(ds, dsname); zfs_dbgmsg("scanning dataset %llu (%s) is unnecessary because " "cur_min_txg (%llu) >= max_txg (%llu)", (longlong_t)dsobj, dsname, (longlong_t)scn->scn_phys.scn_cur_min_txg, (longlong_t)scn->scn_phys.scn_max_txg); kmem_free(dsname, MAXNAMELEN); goto out; } /* * Only the ZIL in the head (non-snapshot) is valid. Even though * snapshots can have ZIL block pointers (which may be the same * BP as in the head), they must be ignored. In addition, $ORIGIN * doesn't have a objset (i.e. its ds_bp is a hole) so we don't * need to look for a ZIL in it either. So we traverse the ZIL here, * rather than in scan_recurse(), because the regular snapshot * block-sharing rules don't apply to it. */ if (!dsl_dataset_is_snapshot(ds) && (dp->dp_origin_snap == NULL || ds->ds_dir != dp->dp_origin_snap->ds_dir)) { objset_t *os; if (dmu_objset_from_ds(ds, &os) != 0) { goto out; } dsl_scan_zil(dp, &os->os_zil_header); } /* * Iterate over the bps in this ds. */ dmu_buf_will_dirty(ds->ds_dbuf, tx); rrw_enter(&ds->ds_bp_rwlock, RW_READER, FTAG); dsl_scan_visit_rootbp(scn, ds, &dsl_dataset_phys(ds)->ds_bp, tx); rrw_exit(&ds->ds_bp_rwlock, FTAG); char *dsname = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP); dsl_dataset_name(ds, dsname); zfs_dbgmsg("scanned dataset %llu (%s) with min=%llu max=%llu; " "suspending=%u", (longlong_t)dsobj, dsname, (longlong_t)scn->scn_phys.scn_cur_min_txg, (longlong_t)scn->scn_phys.scn_cur_max_txg, (int)scn->scn_suspending); kmem_free(dsname, ZFS_MAX_DATASET_NAME_LEN); if (scn->scn_suspending) goto out; /* * We've finished this pass over this dataset. */ /* * If we did not completely visit this dataset, do another pass. */ if (scn->scn_phys.scn_flags & DSF_VISIT_DS_AGAIN) { zfs_dbgmsg("incomplete pass; visiting again"); scn->scn_phys.scn_flags &= ~DSF_VISIT_DS_AGAIN; scan_ds_queue_insert(scn, ds->ds_object, scn->scn_phys.scn_cur_max_txg); goto out; } /* * Add descendant datasets to work queue. */ if (dsl_dataset_phys(ds)->ds_next_snap_obj != 0) { scan_ds_queue_insert(scn, dsl_dataset_phys(ds)->ds_next_snap_obj, dsl_dataset_phys(ds)->ds_creation_txg); } if (dsl_dataset_phys(ds)->ds_num_children > 1) { boolean_t usenext = B_FALSE; if (dsl_dataset_phys(ds)->ds_next_clones_obj != 0) { uint64_t count; /* * A bug in a previous version of the code could * cause upgrade_clones_cb() to not set * ds_next_snap_obj when it should, leading to a * missing entry. Therefore we can only use the * next_clones_obj when its count is correct. */ int err = zap_count(dp->dp_meta_objset, dsl_dataset_phys(ds)->ds_next_clones_obj, &count); if (err == 0 && count == dsl_dataset_phys(ds)->ds_num_children - 1) usenext = B_TRUE; } if (usenext) { zap_cursor_t zc; zap_attribute_t za; for (zap_cursor_init(&zc, dp->dp_meta_objset, dsl_dataset_phys(ds)->ds_next_clones_obj); zap_cursor_retrieve(&zc, &za) == 0; (void) zap_cursor_advance(&zc)) { scan_ds_queue_insert(scn, zfs_strtonum(za.za_name, NULL), dsl_dataset_phys(ds)->ds_creation_txg); } zap_cursor_fini(&zc); } else { VERIFY0(dmu_objset_find_dp(dp, dp->dp_root_dir_obj, enqueue_clones_cb, &ds->ds_object, DS_FIND_CHILDREN)); } } out: dsl_dataset_rele(ds, FTAG); } /* ARGSUSED */ static int enqueue_cb(dsl_pool_t *dp, dsl_dataset_t *hds, void *arg) { dsl_dataset_t *ds; int err; dsl_scan_t *scn = dp->dp_scan; err = dsl_dataset_hold_obj(dp, hds->ds_object, FTAG, &ds); if (err) return (err); while (dsl_dataset_phys(ds)->ds_prev_snap_obj != 0) { dsl_dataset_t *prev; err = dsl_dataset_hold_obj(dp, dsl_dataset_phys(ds)->ds_prev_snap_obj, FTAG, &prev); if (err) { dsl_dataset_rele(ds, FTAG); return (err); } /* * If this is a clone, we don't need to worry about it for now. */ if (dsl_dataset_phys(prev)->ds_next_snap_obj != ds->ds_object) { dsl_dataset_rele(ds, FTAG); dsl_dataset_rele(prev, FTAG); return (0); } dsl_dataset_rele(ds, FTAG); ds = prev; } scan_ds_queue_insert(scn, ds->ds_object, dsl_dataset_phys(ds)->ds_prev_snap_txg); dsl_dataset_rele(ds, FTAG); return (0); } /* ARGSUSED */ void dsl_scan_ddt_entry(dsl_scan_t *scn, enum zio_checksum checksum, ddt_entry_t *dde, dmu_tx_t *tx) { const ddt_key_t *ddk = &dde->dde_key; ddt_phys_t *ddp = dde->dde_phys; blkptr_t bp; zbookmark_phys_t zb = { 0 }; int p; if (!dsl_scan_is_running(scn)) return; /* * This function is special because it is the only thing * that can add scan_io_t's to the vdev scan queues from * outside dsl_scan_sync(). For the most part this is ok * as long as it is called from within syncing context. * However, dsl_scan_sync() expects that no new sio's will * be added between when all the work for a scan is done * and the next txg when the scan is actually marked as * completed. This check ensures we do not issue new sio's * during this period. */ if (scn->scn_done_txg != 0) return; for (p = 0; p < DDT_PHYS_TYPES; p++, ddp++) { if (ddp->ddp_phys_birth == 0 || ddp->ddp_phys_birth > scn->scn_phys.scn_max_txg) continue; ddt_bp_create(checksum, ddk, ddp, &bp); scn->scn_visited_this_txg++; scan_funcs[scn->scn_phys.scn_func](scn->scn_dp, &bp, &zb); } } /* * Scrub/dedup interaction. * * If there are N references to a deduped block, we don't want to scrub it * N times -- ideally, we should scrub it exactly once. * * We leverage the fact that the dde's replication class (enum ddt_class) * is ordered from highest replication class (DDT_CLASS_DITTO) to lowest * (DDT_CLASS_UNIQUE) so that we may walk the DDT in that order. * * To prevent excess scrubbing, the scrub begins by walking the DDT * to find all blocks with refcnt > 1, and scrubs each of these once. * Since there are two replication classes which contain blocks with * refcnt > 1, we scrub the highest replication class (DDT_CLASS_DITTO) first. * Finally the top-down scrub begins, only visiting blocks with refcnt == 1. * * There would be nothing more to say if a block's refcnt couldn't change * during a scrub, but of course it can so we must account for changes * in a block's replication class. * * Here's an example of what can occur: * * If a block has refcnt > 1 during the DDT scrub phase, but has refcnt == 1 * when visited during the top-down scrub phase, it will be scrubbed twice. * This negates our scrub optimization, but is otherwise harmless. * * If a block has refcnt == 1 during the DDT scrub phase, but has refcnt > 1 * on each visit during the top-down scrub phase, it will never be scrubbed. * To catch this, ddt_sync_entry() notifies the scrub code whenever a block's * reference class transitions to a higher level (i.e DDT_CLASS_UNIQUE to * DDT_CLASS_DUPLICATE); if it transitions from refcnt == 1 to refcnt > 1 * while a scrub is in progress, it scrubs the block right then. */ static void dsl_scan_ddt(dsl_scan_t *scn, dmu_tx_t *tx) { ddt_bookmark_t *ddb = &scn->scn_phys.scn_ddt_bookmark; ddt_entry_t dde; int error; uint64_t n = 0; bzero(&dde, sizeof (ddt_entry_t)); while ((error = ddt_walk(scn->scn_dp->dp_spa, ddb, &dde)) == 0) { ddt_t *ddt; if (ddb->ddb_class > scn->scn_phys.scn_ddt_class_max) break; dprintf("visiting ddb=%llu/%llu/%llu/%llx\n", (longlong_t)ddb->ddb_class, (longlong_t)ddb->ddb_type, (longlong_t)ddb->ddb_checksum, (longlong_t)ddb->ddb_cursor); /* There should be no pending changes to the dedup table */ ddt = scn->scn_dp->dp_spa->spa_ddt[ddb->ddb_checksum]; ASSERT(avl_first(&ddt->ddt_tree) == NULL); dsl_scan_ddt_entry(scn, ddb->ddb_checksum, &dde, tx); n++; if (dsl_scan_check_suspend(scn, NULL)) break; } zfs_dbgmsg("scanned %llu ddt entries with class_max = %u; " "suspending=%u", (longlong_t)n, (int)scn->scn_phys.scn_ddt_class_max, (int)scn->scn_suspending); ASSERT(error == 0 || error == ENOENT); ASSERT(error != ENOENT || ddb->ddb_class > scn->scn_phys.scn_ddt_class_max); } static uint64_t dsl_scan_ds_maxtxg(dsl_dataset_t *ds) { uint64_t smt = ds->ds_dir->dd_pool->dp_scan->scn_phys.scn_max_txg; if (ds->ds_is_snapshot) return (MIN(smt, dsl_dataset_phys(ds)->ds_creation_txg)); return (smt); } static void dsl_scan_visit(dsl_scan_t *scn, dmu_tx_t *tx) { scan_ds_t *sds; dsl_pool_t *dp = scn->scn_dp; if (scn->scn_phys.scn_ddt_bookmark.ddb_class <= scn->scn_phys.scn_ddt_class_max) { scn->scn_phys.scn_cur_min_txg = scn->scn_phys.scn_min_txg; scn->scn_phys.scn_cur_max_txg = scn->scn_phys.scn_max_txg; dsl_scan_ddt(scn, tx); if (scn->scn_suspending) return; } if (scn->scn_phys.scn_bookmark.zb_objset == DMU_META_OBJSET) { /* First do the MOS & ORIGIN */ scn->scn_phys.scn_cur_min_txg = scn->scn_phys.scn_min_txg; scn->scn_phys.scn_cur_max_txg = scn->scn_phys.scn_max_txg; dsl_scan_visit_rootbp(scn, NULL, &dp->dp_meta_rootbp, tx); spa_set_rootblkptr(dp->dp_spa, &dp->dp_meta_rootbp); if (scn->scn_suspending) return; if (spa_version(dp->dp_spa) < SPA_VERSION_DSL_SCRUB) { VERIFY0(dmu_objset_find_dp(dp, dp->dp_root_dir_obj, enqueue_cb, NULL, DS_FIND_CHILDREN)); } else { dsl_scan_visitds(scn, dp->dp_origin_snap->ds_object, tx); } ASSERT(!scn->scn_suspending); } else if (scn->scn_phys.scn_bookmark.zb_objset != ZB_DESTROYED_OBJSET) { uint64_t dsobj = scn->scn_phys.scn_bookmark.zb_objset; /* * If we were suspended, continue from here. Note if the * ds we were suspended on was deleted, the zb_objset may * be -1, so we will skip this and find a new objset * below. */ dsl_scan_visitds(scn, dsobj, tx); if (scn->scn_suspending) return; } /* * In case we suspended right at the end of the ds, zero the * bookmark so we don't think that we're still trying to resume. */ bzero(&scn->scn_phys.scn_bookmark, sizeof (zbookmark_phys_t)); /* * Keep pulling things out of the dataset avl queue. Updates to the * persistent zap-object-as-queue happen only at checkpoints. */ while ((sds = avl_first(&scn->scn_queue)) != NULL) { dsl_dataset_t *ds; uint64_t dsobj = sds->sds_dsobj; uint64_t txg = sds->sds_txg; /* dequeue and free the ds from the queue */ scan_ds_queue_remove(scn, dsobj); sds = NULL; /* set up min / max txg */ VERIFY3U(0, ==, dsl_dataset_hold_obj(dp, dsobj, FTAG, &ds)); if (txg != 0) { scn->scn_phys.scn_cur_min_txg = MAX(scn->scn_phys.scn_min_txg, txg); } else { scn->scn_phys.scn_cur_min_txg = MAX(scn->scn_phys.scn_min_txg, dsl_dataset_phys(ds)->ds_prev_snap_txg); } scn->scn_phys.scn_cur_max_txg = dsl_scan_ds_maxtxg(ds); dsl_dataset_rele(ds, FTAG); dsl_scan_visitds(scn, dsobj, tx); if (scn->scn_suspending) return; } /* No more objsets to fetch, we're done */ scn->scn_phys.scn_bookmark.zb_objset = ZB_DESTROYED_OBJSET; ASSERT0(scn->scn_suspending); } static uint64_t dsl_scan_count_leaves(vdev_t *vd) { uint64_t i, leaves = 0; /* we only count leaves that belong to the main pool and are readable */ if (vd->vdev_islog || vd->vdev_isspare || vd->vdev_isl2cache || !vdev_readable(vd)) return (0); if (vd->vdev_ops->vdev_op_leaf) return (1); for (i = 0; i < vd->vdev_children; i++) { leaves += dsl_scan_count_leaves(vd->vdev_child[i]); } return (leaves); } static void scan_io_queues_update_zio_stats(dsl_scan_io_queue_t *q, const blkptr_t *bp) { int i; uint64_t cur_size = 0; for (i = 0; i < BP_GET_NDVAS(bp); i++) { cur_size += DVA_GET_ASIZE(&bp->blk_dva[i]); } q->q_total_zio_size_this_txg += cur_size; q->q_zios_this_txg++; } static void scan_io_queues_update_seg_stats(dsl_scan_io_queue_t *q, uint64_t start, uint64_t end) { q->q_total_seg_size_this_txg += end - start; q->q_segs_this_txg++; } static boolean_t scan_io_queue_check_suspend(dsl_scan_t *scn) { /* See comment in dsl_scan_check_suspend() */ uint64_t curr_time_ns = gethrtime(); uint64_t scan_time_ns = curr_time_ns - scn->scn_sync_start_time; uint64_t sync_time_ns = curr_time_ns - scn->scn_dp->dp_spa->spa_sync_starttime; int dirty_pct = scn->scn_dp->dp_dirty_total * 100 / zfs_dirty_data_max; int mintime = (scn->scn_phys.scn_func == POOL_SCAN_RESILVER) ? zfs_resilver_min_time_ms : zfs_scrub_min_time_ms; return ((NSEC2MSEC(scan_time_ns) > mintime && (dirty_pct >= zfs_vdev_async_write_active_min_dirty_percent || txg_sync_waiting(scn->scn_dp) || NSEC2SEC(sync_time_ns) >= zfs_txg_timeout)) || spa_shutting_down(scn->scn_dp->dp_spa)); } /* * Given a list of scan_io_t's in io_list, this issues the I/Os out to * disk. This consumes the io_list and frees the scan_io_t's. This is * called when emptying queues, either when we're up against the memory * limit or when we have finished scanning. Returns B_TRUE if we stopped * processing the list before we finished. Any sios that were not issued * will remain in the io_list. */ static boolean_t scan_io_queue_issue(dsl_scan_io_queue_t *queue, list_t *io_list) { dsl_scan_t *scn = queue->q_scn; scan_io_t *sio; int64_t bytes_issued = 0; boolean_t suspended = B_FALSE; while ((sio = list_head(io_list)) != NULL) { blkptr_t bp; if (scan_io_queue_check_suspend(scn)) { suspended = B_TRUE; break; } sio2bp(sio, &bp); bytes_issued += SIO_GET_ASIZE(sio); scan_exec_io(scn->scn_dp, &bp, sio->sio_flags, &sio->sio_zb, queue); (void) list_remove_head(io_list); scan_io_queues_update_zio_stats(queue, &bp); sio_free(sio); } atomic_add_64(&scn->scn_bytes_pending, -bytes_issued); return (suspended); } /* * This function removes sios from an IO queue which reside within a given * range_seg_t and inserts them (in offset order) into a list. Note that * we only ever return a maximum of 32 sios at once. If there are more sios * to process within this segment that did not make it onto the list we * return B_TRUE and otherwise B_FALSE. */ static boolean_t scan_io_queue_gather(dsl_scan_io_queue_t *queue, range_seg_t *rs, list_t *list) { scan_io_t *srch_sio, *sio, *next_sio; avl_index_t idx; uint_t num_sios = 0; int64_t bytes_issued = 0; ASSERT(rs != NULL); ASSERT(MUTEX_HELD(&queue->q_vd->vdev_scan_io_queue_lock)); srch_sio = sio_alloc(1); srch_sio->sio_nr_dvas = 1; SIO_SET_OFFSET(srch_sio, rs_get_start(rs, queue->q_exts_by_addr)); /* * The exact start of the extent might not contain any matching zios, * so if that's the case, examine the next one in the tree. */ sio = avl_find(&queue->q_sios_by_addr, srch_sio, &idx); sio_free(srch_sio); if (sio == NULL) sio = avl_nearest(&queue->q_sios_by_addr, idx, AVL_AFTER); while (sio != NULL && SIO_GET_OFFSET(sio) < rs_get_end(rs, queue->q_exts_by_addr) && num_sios <= 32) { ASSERT3U(SIO_GET_OFFSET(sio), >=, rs_get_start(rs, queue->q_exts_by_addr)); ASSERT3U(SIO_GET_END_OFFSET(sio), <=, rs_get_end(rs, queue->q_exts_by_addr)); next_sio = AVL_NEXT(&queue->q_sios_by_addr, sio); avl_remove(&queue->q_sios_by_addr, sio); queue->q_sio_memused -= SIO_GET_MUSED(sio); bytes_issued += SIO_GET_ASIZE(sio); num_sios++; list_insert_tail(list, sio); sio = next_sio; } /* * We limit the number of sios we process at once to 32 to avoid * biting off more than we can chew. If we didn't take everything * in the segment we update it to reflect the work we were able to * complete. Otherwise, we remove it from the range tree entirely. */ if (sio != NULL && SIO_GET_OFFSET(sio) < rs_get_end(rs, queue->q_exts_by_addr)) { range_tree_adjust_fill(queue->q_exts_by_addr, rs, -bytes_issued); range_tree_resize_segment(queue->q_exts_by_addr, rs, SIO_GET_OFFSET(sio), rs_get_end(rs, queue->q_exts_by_addr) - SIO_GET_OFFSET(sio)); return (B_TRUE); } else { uint64_t rstart = rs_get_start(rs, queue->q_exts_by_addr); uint64_t rend = rs_get_end(rs, queue->q_exts_by_addr); range_tree_remove(queue->q_exts_by_addr, rstart, rend - rstart); return (B_FALSE); } } /* * This is called from the queue emptying thread and selects the next * extent from which we are to issue I/Os. The behavior of this function * depends on the state of the scan, the current memory consumption and * whether or not we are performing a scan shutdown. * 1) We select extents in an elevator algorithm (LBA-order) if the scan * needs to perform a checkpoint * 2) We select the largest available extent if we are up against the * memory limit. * 3) Otherwise we don't select any extents. */ static range_seg_t * scan_io_queue_fetch_ext(dsl_scan_io_queue_t *queue) { dsl_scan_t *scn = queue->q_scn; range_tree_t *rt = queue->q_exts_by_addr; ASSERT(MUTEX_HELD(&queue->q_vd->vdev_scan_io_queue_lock)); ASSERT(scn->scn_is_sorted); /* handle tunable overrides */ if (scn->scn_checkpointing || scn->scn_clearing) { if (zfs_scan_issue_strategy == 1) { return (range_tree_first(rt)); } else if (zfs_scan_issue_strategy == 2) { /* * We need to get the original entry in the by_addr * tree so we can modify it. */ range_seg_t *size_rs = zfs_btree_first(&queue->q_exts_by_size, NULL); if (size_rs == NULL) return (NULL); uint64_t start = rs_get_start(size_rs, rt); uint64_t size = rs_get_end(size_rs, rt) - start; range_seg_t *addr_rs = range_tree_find(rt, start, size); ASSERT3P(addr_rs, !=, NULL); ASSERT3U(rs_get_start(size_rs, rt), ==, rs_get_start(addr_rs, rt)); ASSERT3U(rs_get_end(size_rs, rt), ==, rs_get_end(addr_rs, rt)); return (addr_rs); } } /* * During normal clearing, we want to issue our largest segments * first, keeping IO as sequential as possible, and leaving the * smaller extents for later with the hope that they might eventually * grow to larger sequential segments. However, when the scan is * checkpointing, no new extents will be added to the sorting queue, * so the way we are sorted now is as good as it will ever get. * In this case, we instead switch to issuing extents in LBA order. */ if (scn->scn_checkpointing) { return (range_tree_first(rt)); } else if (scn->scn_clearing) { /* * We need to get the original entry in the by_addr * tree so we can modify it. */ range_seg_t *size_rs = zfs_btree_first(&queue->q_exts_by_size, NULL); if (size_rs == NULL) return (NULL); uint64_t start = rs_get_start(size_rs, rt); uint64_t size = rs_get_end(size_rs, rt) - start; range_seg_t *addr_rs = range_tree_find(rt, start, size); ASSERT3P(addr_rs, !=, NULL); ASSERT3U(rs_get_start(size_rs, rt), ==, rs_get_start(addr_rs, rt)); ASSERT3U(rs_get_end(size_rs, rt), ==, rs_get_end(addr_rs, rt)); return (addr_rs); } else { return (NULL); } } static void scan_io_queues_run_one(void *arg) { dsl_scan_io_queue_t *queue = arg; kmutex_t *q_lock = &queue->q_vd->vdev_scan_io_queue_lock; boolean_t suspended = B_FALSE; range_seg_t *rs = NULL; scan_io_t *sio = NULL; list_t sio_list; uint64_t bytes_per_leaf = zfs_scan_vdev_limit; uint64_t nr_leaves = dsl_scan_count_leaves(queue->q_vd); ASSERT(queue->q_scn->scn_is_sorted); list_create(&sio_list, sizeof (scan_io_t), offsetof(scan_io_t, sio_nodes.sio_list_node)); mutex_enter(q_lock); /* calculate maximum in-flight bytes for this txg (min 1MB) */ queue->q_maxinflight_bytes = MAX(nr_leaves * bytes_per_leaf, 1ULL << 20); /* reset per-queue scan statistics for this txg */ queue->q_total_seg_size_this_txg = 0; queue->q_segs_this_txg = 0; queue->q_total_zio_size_this_txg = 0; queue->q_zios_this_txg = 0; /* loop until we run out of time or sios */ while ((rs = scan_io_queue_fetch_ext(queue)) != NULL) { uint64_t seg_start = 0, seg_end = 0; boolean_t more_left = B_TRUE; ASSERT(list_is_empty(&sio_list)); /* loop while we still have sios left to process in this rs */ while (more_left) { scan_io_t *first_sio, *last_sio; /* * We have selected which extent needs to be * processed next. Gather up the corresponding sios. */ more_left = scan_io_queue_gather(queue, rs, &sio_list); ASSERT(!list_is_empty(&sio_list)); first_sio = list_head(&sio_list); last_sio = list_tail(&sio_list); seg_end = SIO_GET_END_OFFSET(last_sio); if (seg_start == 0) seg_start = SIO_GET_OFFSET(first_sio); /* * Issuing sios can take a long time so drop the * queue lock. The sio queue won't be updated by * other threads since we're in syncing context so * we can be sure that our trees will remain exactly * as we left them. */ mutex_exit(q_lock); suspended = scan_io_queue_issue(queue, &sio_list); mutex_enter(q_lock); if (suspended) break; } /* update statistics for debugging purposes */ scan_io_queues_update_seg_stats(queue, seg_start, seg_end); if (suspended) break; } /* * If we were suspended in the middle of processing, * requeue any unfinished sios and exit. */ while ((sio = list_head(&sio_list)) != NULL) { list_remove(&sio_list, sio); scan_io_queue_insert_impl(queue, sio); } mutex_exit(q_lock); list_destroy(&sio_list); } /* * Performs an emptying run on all scan queues in the pool. This just * punches out one thread per top-level vdev, each of which processes * only that vdev's scan queue. We can parallelize the I/O here because * we know that each queue's I/Os only affect its own top-level vdev. * * This function waits for the queue runs to complete, and must be * called from dsl_scan_sync (or in general, syncing context). */ static void scan_io_queues_run(dsl_scan_t *scn) { spa_t *spa = scn->scn_dp->dp_spa; ASSERT(scn->scn_is_sorted); ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER)); if (scn->scn_bytes_pending == 0) return; if (scn->scn_taskq == NULL) { int nthreads = spa->spa_root_vdev->vdev_children; /* * We need to make this taskq *always* execute as many * threads in parallel as we have top-level vdevs and no * less, otherwise strange serialization of the calls to * scan_io_queues_run_one can occur during spa_sync runs * and that significantly impacts performance. */ scn->scn_taskq = taskq_create("dsl_scan_iss", nthreads, minclsyspri, nthreads, nthreads, TASKQ_PREPOPULATE); } for (uint64_t i = 0; i < spa->spa_root_vdev->vdev_children; i++) { vdev_t *vd = spa->spa_root_vdev->vdev_child[i]; mutex_enter(&vd->vdev_scan_io_queue_lock); if (vd->vdev_scan_io_queue != NULL) { VERIFY(taskq_dispatch(scn->scn_taskq, scan_io_queues_run_one, vd->vdev_scan_io_queue, TQ_SLEEP) != TASKQID_INVALID); } mutex_exit(&vd->vdev_scan_io_queue_lock); } /* * Wait for the queues to finish issuing their IOs for this run * before we return. There may still be IOs in flight at this * point. */ taskq_wait(scn->scn_taskq); } static boolean_t dsl_scan_async_block_should_pause(dsl_scan_t *scn) { uint64_t elapsed_nanosecs; if (zfs_recover) return (B_FALSE); if (zfs_async_block_max_blocks != 0 && scn->scn_visited_this_txg >= zfs_async_block_max_blocks) { return (B_TRUE); } if (zfs_max_async_dedup_frees != 0 && scn->scn_dedup_frees_this_txg >= zfs_max_async_dedup_frees) { return (B_TRUE); } elapsed_nanosecs = gethrtime() - scn->scn_sync_start_time; return (elapsed_nanosecs / NANOSEC > zfs_txg_timeout || (NSEC2MSEC(elapsed_nanosecs) > scn->scn_async_block_min_time_ms && txg_sync_waiting(scn->scn_dp)) || spa_shutting_down(scn->scn_dp->dp_spa)); } static int dsl_scan_free_block_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx) { dsl_scan_t *scn = arg; if (!scn->scn_is_bptree || (BP_GET_LEVEL(bp) == 0 && BP_GET_TYPE(bp) != DMU_OT_OBJSET)) { if (dsl_scan_async_block_should_pause(scn)) return (SET_ERROR(ERESTART)); } zio_nowait(zio_free_sync(scn->scn_zio_root, scn->scn_dp->dp_spa, dmu_tx_get_txg(tx), bp, 0)); dsl_dir_diduse_space(tx->tx_pool->dp_free_dir, DD_USED_HEAD, -bp_get_dsize_sync(scn->scn_dp->dp_spa, bp), -BP_GET_PSIZE(bp), -BP_GET_UCSIZE(bp), tx); scn->scn_visited_this_txg++; if (BP_GET_DEDUP(bp)) scn->scn_dedup_frees_this_txg++; return (0); } static void dsl_scan_update_stats(dsl_scan_t *scn) { spa_t *spa = scn->scn_dp->dp_spa; uint64_t i; uint64_t seg_size_total = 0, zio_size_total = 0; uint64_t seg_count_total = 0, zio_count_total = 0; for (i = 0; i < spa->spa_root_vdev->vdev_children; i++) { vdev_t *vd = spa->spa_root_vdev->vdev_child[i]; dsl_scan_io_queue_t *queue = vd->vdev_scan_io_queue; if (queue == NULL) continue; seg_size_total += queue->q_total_seg_size_this_txg; zio_size_total += queue->q_total_zio_size_this_txg; seg_count_total += queue->q_segs_this_txg; zio_count_total += queue->q_zios_this_txg; } if (seg_count_total == 0 || zio_count_total == 0) { scn->scn_avg_seg_size_this_txg = 0; scn->scn_avg_zio_size_this_txg = 0; scn->scn_segs_this_txg = 0; scn->scn_zios_this_txg = 0; return; } scn->scn_avg_seg_size_this_txg = seg_size_total / seg_count_total; scn->scn_avg_zio_size_this_txg = zio_size_total / zio_count_total; scn->scn_segs_this_txg = seg_count_total; scn->scn_zios_this_txg = zio_count_total; } static int bpobj_dsl_scan_free_block_cb(void *arg, const blkptr_t *bp, boolean_t bp_freed, dmu_tx_t *tx) { ASSERT(!bp_freed); return (dsl_scan_free_block_cb(arg, bp, tx)); } static int dsl_scan_obsolete_block_cb(void *arg, const blkptr_t *bp, boolean_t bp_freed, dmu_tx_t *tx) { ASSERT(!bp_freed); dsl_scan_t *scn = arg; const dva_t *dva = &bp->blk_dva[0]; if (dsl_scan_async_block_should_pause(scn)) return (SET_ERROR(ERESTART)); spa_vdev_indirect_mark_obsolete(scn->scn_dp->dp_spa, DVA_GET_VDEV(dva), DVA_GET_OFFSET(dva), DVA_GET_ASIZE(dva), tx); scn->scn_visited_this_txg++; return (0); } boolean_t dsl_scan_active(dsl_scan_t *scn) { spa_t *spa = scn->scn_dp->dp_spa; uint64_t used = 0, comp, uncomp; boolean_t clones_left; if (spa->spa_load_state != SPA_LOAD_NONE) return (B_FALSE); if (spa_shutting_down(spa)) return (B_FALSE); if ((dsl_scan_is_running(scn) && !dsl_scan_is_paused_scrub(scn)) || (scn->scn_async_destroying && !scn->scn_async_stalled)) return (B_TRUE); if (spa_version(scn->scn_dp->dp_spa) >= SPA_VERSION_DEADLISTS) { (void) bpobj_space(&scn->scn_dp->dp_free_bpobj, &used, &comp, &uncomp); } clones_left = spa_livelist_delete_check(spa); return ((used != 0) || (clones_left)); } static boolean_t dsl_scan_check_deferred(vdev_t *vd) { boolean_t need_resilver = B_FALSE; for (int c = 0; c < vd->vdev_children; c++) { need_resilver |= dsl_scan_check_deferred(vd->vdev_child[c]); } if (!vdev_is_concrete(vd) || vd->vdev_aux || !vd->vdev_ops->vdev_op_leaf) return (need_resilver); if (!vd->vdev_resilver_deferred) need_resilver = B_TRUE; return (need_resilver); } static boolean_t dsl_scan_need_resilver(spa_t *spa, const dva_t *dva, size_t psize, uint64_t phys_birth) { vdev_t *vd; vd = vdev_lookup_top(spa, DVA_GET_VDEV(dva)); if (vd->vdev_ops == &vdev_indirect_ops) { /* * The indirect vdev can point to multiple * vdevs. For simplicity, always create * the resilver zio_t. zio_vdev_io_start() * will bypass the child resilver i/o's if * they are on vdevs that don't have DTL's. */ return (B_TRUE); } if (DVA_GET_GANG(dva)) { /* * Gang members may be spread across multiple * vdevs, so the best estimate we have is the * scrub range, which has already been checked. * XXX -- it would be better to change our * allocation policy to ensure that all * gang members reside on the same vdev. */ return (B_TRUE); } /* * Check if the txg falls within the range which must be * resilvered. DVAs outside this range can always be skipped. */ if (!vdev_dtl_contains(vd, DTL_PARTIAL, phys_birth, 1)) return (B_FALSE); /* * Check if the top-level vdev must resilver this offset. * When the offset does not intersect with a dirty leaf DTL * then it may be possible to skip the resilver IO. The psize * is provided instead of asize to simplify the check for RAIDZ. */ if (!vdev_dtl_need_resilver(vd, DVA_GET_OFFSET(dva), psize)) return (B_FALSE); /* * Check that this top-level vdev has a device under it which * is resilvering and is not deferred. */ if (!dsl_scan_check_deferred(vd)) return (B_FALSE); return (B_TRUE); } static int dsl_process_async_destroys(dsl_pool_t *dp, dmu_tx_t *tx) { dsl_scan_t *scn = dp->dp_scan; spa_t *spa = dp->dp_spa; int err = 0; if (spa_suspend_async_destroy(spa)) return (0); if (zfs_free_bpobj_enabled && spa_version(spa) >= SPA_VERSION_DEADLISTS) { scn->scn_is_bptree = B_FALSE; scn->scn_async_block_min_time_ms = zfs_free_min_time_ms; scn->scn_zio_root = zio_root(spa, NULL, NULL, ZIO_FLAG_MUSTSUCCEED); err = bpobj_iterate(&dp->dp_free_bpobj, bpobj_dsl_scan_free_block_cb, scn, tx); VERIFY0(zio_wait(scn->scn_zio_root)); scn->scn_zio_root = NULL; if (err != 0 && err != ERESTART) zfs_panic_recover("error %u from bpobj_iterate()", err); } if (err == 0 && spa_feature_is_active(spa, SPA_FEATURE_ASYNC_DESTROY)) { ASSERT(scn->scn_async_destroying); scn->scn_is_bptree = B_TRUE; scn->scn_zio_root = zio_root(spa, NULL, NULL, ZIO_FLAG_MUSTSUCCEED); err = bptree_iterate(dp->dp_meta_objset, dp->dp_bptree_obj, B_TRUE, dsl_scan_free_block_cb, scn, tx); VERIFY0(zio_wait(scn->scn_zio_root)); scn->scn_zio_root = NULL; if (err == EIO || err == ECKSUM) { err = 0; } else if (err != 0 && err != ERESTART) { zfs_panic_recover("error %u from " "traverse_dataset_destroyed()", err); } if (bptree_is_empty(dp->dp_meta_objset, dp->dp_bptree_obj)) { /* finished; deactivate async destroy feature */ spa_feature_decr(spa, SPA_FEATURE_ASYNC_DESTROY, tx); ASSERT(!spa_feature_is_active(spa, SPA_FEATURE_ASYNC_DESTROY)); VERIFY0(zap_remove(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_BPTREE_OBJ, tx)); VERIFY0(bptree_free(dp->dp_meta_objset, dp->dp_bptree_obj, tx)); dp->dp_bptree_obj = 0; scn->scn_async_destroying = B_FALSE; scn->scn_async_stalled = B_FALSE; } else { /* * If we didn't make progress, mark the async * destroy as stalled, so that we will not initiate * a spa_sync() on its behalf. Note that we only * check this if we are not finished, because if the * bptree had no blocks for us to visit, we can * finish without "making progress". */ scn->scn_async_stalled = (scn->scn_visited_this_txg == 0); } } if (scn->scn_visited_this_txg) { zfs_dbgmsg("freed %llu blocks in %llums from " "free_bpobj/bptree txg %llu; err=%u", (longlong_t)scn->scn_visited_this_txg, (longlong_t) NSEC2MSEC(gethrtime() - scn->scn_sync_start_time), (longlong_t)tx->tx_txg, err); scn->scn_visited_this_txg = 0; scn->scn_dedup_frees_this_txg = 0; /* * Write out changes to the DDT that may be required as a * result of the blocks freed. This ensures that the DDT * is clean when a scrub/resilver runs. */ ddt_sync(spa, tx->tx_txg); } if (err != 0) return (err); if (dp->dp_free_dir != NULL && !scn->scn_async_destroying && zfs_free_leak_on_eio && (dsl_dir_phys(dp->dp_free_dir)->dd_used_bytes != 0 || dsl_dir_phys(dp->dp_free_dir)->dd_compressed_bytes != 0 || dsl_dir_phys(dp->dp_free_dir)->dd_uncompressed_bytes != 0)) { /* * We have finished background destroying, but there is still * some space left in the dp_free_dir. Transfer this leaked * space to the dp_leak_dir. */ if (dp->dp_leak_dir == NULL) { rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG); (void) dsl_dir_create_sync(dp, dp->dp_root_dir, LEAK_DIR_NAME, tx); VERIFY0(dsl_pool_open_special_dir(dp, LEAK_DIR_NAME, &dp->dp_leak_dir)); rrw_exit(&dp->dp_config_rwlock, FTAG); } dsl_dir_diduse_space(dp->dp_leak_dir, DD_USED_HEAD, dsl_dir_phys(dp->dp_free_dir)->dd_used_bytes, dsl_dir_phys(dp->dp_free_dir)->dd_compressed_bytes, dsl_dir_phys(dp->dp_free_dir)->dd_uncompressed_bytes, tx); dsl_dir_diduse_space(dp->dp_free_dir, DD_USED_HEAD, -dsl_dir_phys(dp->dp_free_dir)->dd_used_bytes, -dsl_dir_phys(dp->dp_free_dir)->dd_compressed_bytes, -dsl_dir_phys(dp->dp_free_dir)->dd_uncompressed_bytes, tx); } if (dp->dp_free_dir != NULL && !scn->scn_async_destroying && !spa_livelist_delete_check(spa)) { /* finished; verify that space accounting went to zero */ ASSERT0(dsl_dir_phys(dp->dp_free_dir)->dd_used_bytes); ASSERT0(dsl_dir_phys(dp->dp_free_dir)->dd_compressed_bytes); ASSERT0(dsl_dir_phys(dp->dp_free_dir)->dd_uncompressed_bytes); } spa_notify_waiters(spa); EQUIV(bpobj_is_open(&dp->dp_obsolete_bpobj), 0 == zap_contains(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_OBSOLETE_BPOBJ)); if (err == 0 && bpobj_is_open(&dp->dp_obsolete_bpobj)) { ASSERT(spa_feature_is_active(dp->dp_spa, SPA_FEATURE_OBSOLETE_COUNTS)); scn->scn_is_bptree = B_FALSE; scn->scn_async_block_min_time_ms = zfs_obsolete_min_time_ms; err = bpobj_iterate(&dp->dp_obsolete_bpobj, dsl_scan_obsolete_block_cb, scn, tx); if (err != 0 && err != ERESTART) zfs_panic_recover("error %u from bpobj_iterate()", err); if (bpobj_is_empty(&dp->dp_obsolete_bpobj)) dsl_pool_destroy_obsolete_bpobj(dp, tx); } return (0); } /* * This is the primary entry point for scans that is called from syncing * context. Scans must happen entirely during syncing context so that we * can guarantee that blocks we are currently scanning will not change out * from under us. While a scan is active, this function controls how quickly * transaction groups proceed, instead of the normal handling provided by * txg_sync_thread(). */ void dsl_scan_sync(dsl_pool_t *dp, dmu_tx_t *tx) { int err = 0; dsl_scan_t *scn = dp->dp_scan; spa_t *spa = dp->dp_spa; state_sync_type_t sync_type = SYNC_OPTIONAL; if (spa->spa_resilver_deferred && !spa_feature_is_active(dp->dp_spa, SPA_FEATURE_RESILVER_DEFER)) spa_feature_incr(spa, SPA_FEATURE_RESILVER_DEFER, tx); /* * Check for scn_restart_txg before checking spa_load_state, so * that we can restart an old-style scan while the pool is being * imported (see dsl_scan_init). We also restart scans if there * is a deferred resilver and the user has manually disabled * deferred resilvers via the tunable. */ if (dsl_scan_restarting(scn, tx) || (spa->spa_resilver_deferred && zfs_resilver_disable_defer)) { pool_scan_func_t func = POOL_SCAN_SCRUB; dsl_scan_done(scn, B_FALSE, tx); if (vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL)) func = POOL_SCAN_RESILVER; zfs_dbgmsg("restarting scan func=%u txg=%llu", func, (longlong_t)tx->tx_txg); dsl_scan_setup_sync(&func, tx); } /* * Only process scans in sync pass 1. */ if (spa_sync_pass(spa) > 1) return; /* * If the spa is shutting down, then stop scanning. This will * ensure that the scan does not dirty any new data during the * shutdown phase. */ if (spa_shutting_down(spa)) return; /* * If the scan is inactive due to a stalled async destroy, try again. */ if (!scn->scn_async_stalled && !dsl_scan_active(scn)) return; /* reset scan statistics */ scn->scn_visited_this_txg = 0; scn->scn_dedup_frees_this_txg = 0; scn->scn_holes_this_txg = 0; scn->scn_lt_min_this_txg = 0; scn->scn_gt_max_this_txg = 0; scn->scn_ddt_contained_this_txg = 0; scn->scn_objsets_visited_this_txg = 0; scn->scn_avg_seg_size_this_txg = 0; scn->scn_segs_this_txg = 0; scn->scn_avg_zio_size_this_txg = 0; scn->scn_zios_this_txg = 0; scn->scn_suspending = B_FALSE; scn->scn_sync_start_time = gethrtime(); spa->spa_scrub_active = B_TRUE; /* * First process the async destroys. If we suspend, don't do * any scrubbing or resilvering. This ensures that there are no * async destroys while we are scanning, so the scan code doesn't * have to worry about traversing it. It is also faster to free the * blocks than to scrub them. */ err = dsl_process_async_destroys(dp, tx); if (err != 0) return; if (!dsl_scan_is_running(scn) || dsl_scan_is_paused_scrub(scn)) return; /* * Wait a few txgs after importing to begin scanning so that * we can get the pool imported quickly. */ if (spa->spa_syncing_txg < spa->spa_first_txg + SCAN_IMPORT_WAIT_TXGS) return; /* * zfs_scan_suspend_progress can be set to disable scan progress. * We don't want to spin the txg_sync thread, so we add a delay * here to simulate the time spent doing a scan. This is mostly * useful for testing and debugging. */ if (zfs_scan_suspend_progress) { uint64_t scan_time_ns = gethrtime() - scn->scn_sync_start_time; int mintime = (scn->scn_phys.scn_func == POOL_SCAN_RESILVER) ? zfs_resilver_min_time_ms : zfs_scrub_min_time_ms; while (zfs_scan_suspend_progress && !txg_sync_waiting(scn->scn_dp) && !spa_shutting_down(scn->scn_dp->dp_spa) && NSEC2MSEC(scan_time_ns) < mintime) { delay(hz); scan_time_ns = gethrtime() - scn->scn_sync_start_time; } return; } /* * It is possible to switch from unsorted to sorted at any time, * but afterwards the scan will remain sorted unless reloaded from * a checkpoint after a reboot. */ if (!zfs_scan_legacy) { scn->scn_is_sorted = B_TRUE; if (scn->scn_last_checkpoint == 0) scn->scn_last_checkpoint = ddi_get_lbolt(); } /* * For sorted scans, determine what kind of work we will be doing * this txg based on our memory limitations and whether or not we * need to perform a checkpoint. */ if (scn->scn_is_sorted) { /* * If we are over our checkpoint interval, set scn_clearing * so that we can begin checkpointing immediately. The * checkpoint allows us to save a consistent bookmark * representing how much data we have scrubbed so far. * Otherwise, use the memory limit to determine if we should * scan for metadata or start issue scrub IOs. We accumulate * metadata until we hit our hard memory limit at which point * we issue scrub IOs until we are at our soft memory limit. */ if (scn->scn_checkpointing || ddi_get_lbolt() - scn->scn_last_checkpoint > SEC_TO_TICK(zfs_scan_checkpoint_intval)) { if (!scn->scn_checkpointing) zfs_dbgmsg("begin scan checkpoint"); scn->scn_checkpointing = B_TRUE; scn->scn_clearing = B_TRUE; } else { boolean_t should_clear = dsl_scan_should_clear(scn); if (should_clear && !scn->scn_clearing) { zfs_dbgmsg("begin scan clearing"); scn->scn_clearing = B_TRUE; } else if (!should_clear && scn->scn_clearing) { zfs_dbgmsg("finish scan clearing"); scn->scn_clearing = B_FALSE; } } } else { ASSERT0(scn->scn_checkpointing); ASSERT0(scn->scn_clearing); } if (!scn->scn_clearing && scn->scn_done_txg == 0) { /* Need to scan metadata for more blocks to scrub */ dsl_scan_phys_t *scnp = &scn->scn_phys; taskqid_t prefetch_tqid; uint64_t bytes_per_leaf = zfs_scan_vdev_limit; uint64_t nr_leaves = dsl_scan_count_leaves(spa->spa_root_vdev); /* * Recalculate the max number of in-flight bytes for pool-wide * scanning operations (minimum 1MB). Limits for the issuing * phase are done per top-level vdev and are handled separately. */ scn->scn_maxinflight_bytes = MAX(nr_leaves * bytes_per_leaf, 1ULL << 20); if (scnp->scn_ddt_bookmark.ddb_class <= scnp->scn_ddt_class_max) { ASSERT(ZB_IS_ZERO(&scnp->scn_bookmark)); zfs_dbgmsg("doing scan sync txg %llu; " "ddt bm=%llu/%llu/%llu/%llx", (longlong_t)tx->tx_txg, (longlong_t)scnp->scn_ddt_bookmark.ddb_class, (longlong_t)scnp->scn_ddt_bookmark.ddb_type, (longlong_t)scnp->scn_ddt_bookmark.ddb_checksum, (longlong_t)scnp->scn_ddt_bookmark.ddb_cursor); } else { zfs_dbgmsg("doing scan sync txg %llu; " "bm=%llu/%llu/%llu/%llu", (longlong_t)tx->tx_txg, (longlong_t)scnp->scn_bookmark.zb_objset, (longlong_t)scnp->scn_bookmark.zb_object, (longlong_t)scnp->scn_bookmark.zb_level, (longlong_t)scnp->scn_bookmark.zb_blkid); } scn->scn_zio_root = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_CANFAIL); scn->scn_prefetch_stop = B_FALSE; prefetch_tqid = taskq_dispatch(dp->dp_sync_taskq, dsl_scan_prefetch_thread, scn, TQ_SLEEP); ASSERT(prefetch_tqid != TASKQID_INVALID); dsl_pool_config_enter(dp, FTAG); dsl_scan_visit(scn, tx); dsl_pool_config_exit(dp, FTAG); mutex_enter(&dp->dp_spa->spa_scrub_lock); scn->scn_prefetch_stop = B_TRUE; cv_broadcast(&spa->spa_scrub_io_cv); mutex_exit(&dp->dp_spa->spa_scrub_lock); taskq_wait_id(dp->dp_sync_taskq, prefetch_tqid); (void) zio_wait(scn->scn_zio_root); scn->scn_zio_root = NULL; zfs_dbgmsg("scan visited %llu blocks in %llums " "(%llu os's, %llu holes, %llu < mintxg, " "%llu in ddt, %llu > maxtxg)", (longlong_t)scn->scn_visited_this_txg, (longlong_t)NSEC2MSEC(gethrtime() - scn->scn_sync_start_time), (longlong_t)scn->scn_objsets_visited_this_txg, (longlong_t)scn->scn_holes_this_txg, (longlong_t)scn->scn_lt_min_this_txg, (longlong_t)scn->scn_ddt_contained_this_txg, (longlong_t)scn->scn_gt_max_this_txg); if (!scn->scn_suspending) { ASSERT0(avl_numnodes(&scn->scn_queue)); scn->scn_done_txg = tx->tx_txg + 1; if (scn->scn_is_sorted) { scn->scn_checkpointing = B_TRUE; scn->scn_clearing = B_TRUE; } zfs_dbgmsg("scan complete txg %llu", (longlong_t)tx->tx_txg); } } else if (scn->scn_is_sorted && scn->scn_bytes_pending != 0) { ASSERT(scn->scn_clearing); /* need to issue scrubbing IOs from per-vdev queues */ scn->scn_zio_root = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_CANFAIL); scan_io_queues_run(scn); (void) zio_wait(scn->scn_zio_root); scn->scn_zio_root = NULL; /* calculate and dprintf the current memory usage */ (void) dsl_scan_should_clear(scn); dsl_scan_update_stats(scn); zfs_dbgmsg("scan issued %llu blocks (%llu segs) in %llums " "(avg_block_size = %llu, avg_seg_size = %llu)", (longlong_t)scn->scn_zios_this_txg, (longlong_t)scn->scn_segs_this_txg, (longlong_t)NSEC2MSEC(gethrtime() - scn->scn_sync_start_time), (longlong_t)scn->scn_avg_zio_size_this_txg, (longlong_t)scn->scn_avg_seg_size_this_txg); } else if (scn->scn_done_txg != 0 && scn->scn_done_txg <= tx->tx_txg) { /* Finished with everything. Mark the scrub as complete */ zfs_dbgmsg("scan issuing complete txg %llu", (longlong_t)tx->tx_txg); ASSERT3U(scn->scn_done_txg, !=, 0); ASSERT0(spa->spa_scrub_inflight); ASSERT0(scn->scn_bytes_pending); dsl_scan_done(scn, B_TRUE, tx); sync_type = SYNC_MANDATORY; } dsl_scan_sync_state(scn, tx, sync_type); } static void count_block(dsl_scan_t *scn, zfs_all_blkstats_t *zab, const blkptr_t *bp) { int i; /* * Don't count embedded bp's, since we already did the work of * scanning these when we scanned the containing block. */ if (BP_IS_EMBEDDED(bp)) return; /* * Update the spa's stats on how many bytes we have issued. * Sequential scrubs create a zio for each DVA of the bp. Each * of these will include all DVAs for repair purposes, but the * zio code will only try the first one unless there is an issue. * Therefore, we should only count the first DVA for these IOs. */ if (scn->scn_is_sorted) { atomic_add_64(&scn->scn_dp->dp_spa->spa_scan_pass_issued, DVA_GET_ASIZE(&bp->blk_dva[0])); } else { spa_t *spa = scn->scn_dp->dp_spa; for (i = 0; i < BP_GET_NDVAS(bp); i++) { atomic_add_64(&spa->spa_scan_pass_issued, DVA_GET_ASIZE(&bp->blk_dva[i])); } } /* * If we resume after a reboot, zab will be NULL; don't record * incomplete stats in that case. */ if (zab == NULL) return; mutex_enter(&zab->zab_lock); for (i = 0; i < 4; i++) { int l = (i < 2) ? BP_GET_LEVEL(bp) : DN_MAX_LEVELS; int t = (i & 1) ? BP_GET_TYPE(bp) : DMU_OT_TOTAL; if (t & DMU_OT_NEWTYPE) t = DMU_OT_OTHER; zfs_blkstat_t *zb = &zab->zab_type[l][t]; int equal; zb->zb_count++; zb->zb_asize += BP_GET_ASIZE(bp); zb->zb_lsize += BP_GET_LSIZE(bp); zb->zb_psize += BP_GET_PSIZE(bp); zb->zb_gangs += BP_COUNT_GANG(bp); switch (BP_GET_NDVAS(bp)) { case 2: if (DVA_GET_VDEV(&bp->blk_dva[0]) == DVA_GET_VDEV(&bp->blk_dva[1])) zb->zb_ditto_2_of_2_samevdev++; break; case 3: equal = (DVA_GET_VDEV(&bp->blk_dva[0]) == DVA_GET_VDEV(&bp->blk_dva[1])) + (DVA_GET_VDEV(&bp->blk_dva[0]) == DVA_GET_VDEV(&bp->blk_dva[2])) + (DVA_GET_VDEV(&bp->blk_dva[1]) == DVA_GET_VDEV(&bp->blk_dva[2])); if (equal == 1) zb->zb_ditto_2_of_3_samevdev++; else if (equal == 3) zb->zb_ditto_3_of_3_samevdev++; break; } } mutex_exit(&zab->zab_lock); } static void scan_io_queue_insert_impl(dsl_scan_io_queue_t *queue, scan_io_t *sio) { avl_index_t idx; int64_t asize = SIO_GET_ASIZE(sio); dsl_scan_t *scn = queue->q_scn; ASSERT(MUTEX_HELD(&queue->q_vd->vdev_scan_io_queue_lock)); if (avl_find(&queue->q_sios_by_addr, sio, &idx) != NULL) { /* block is already scheduled for reading */ atomic_add_64(&scn->scn_bytes_pending, -asize); sio_free(sio); return; } avl_insert(&queue->q_sios_by_addr, sio, idx); queue->q_sio_memused += SIO_GET_MUSED(sio); range_tree_add(queue->q_exts_by_addr, SIO_GET_OFFSET(sio), asize); } /* * Given all the info we got from our metadata scanning process, we * construct a scan_io_t and insert it into the scan sorting queue. The * I/O must already be suitable for us to process. This is controlled * by dsl_scan_enqueue(). */ static void scan_io_queue_insert(dsl_scan_io_queue_t *queue, const blkptr_t *bp, int dva_i, int zio_flags, const zbookmark_phys_t *zb) { dsl_scan_t *scn = queue->q_scn; scan_io_t *sio = sio_alloc(BP_GET_NDVAS(bp)); ASSERT0(BP_IS_GANG(bp)); ASSERT(MUTEX_HELD(&queue->q_vd->vdev_scan_io_queue_lock)); bp2sio(bp, sio, dva_i); sio->sio_flags = zio_flags; sio->sio_zb = *zb; /* * Increment the bytes pending counter now so that we can't * get an integer underflow in case the worker processes the * zio before we get to incrementing this counter. */ atomic_add_64(&scn->scn_bytes_pending, SIO_GET_ASIZE(sio)); scan_io_queue_insert_impl(queue, sio); } /* * Given a set of I/O parameters as discovered by the metadata traversal * process, attempts to place the I/O into the sorted queues (if allowed), * or immediately executes the I/O. */ static void dsl_scan_enqueue(dsl_pool_t *dp, const blkptr_t *bp, int zio_flags, const zbookmark_phys_t *zb) { spa_t *spa = dp->dp_spa; ASSERT(!BP_IS_EMBEDDED(bp)); /* * Gang blocks are hard to issue sequentially, so we just issue them * here immediately instead of queuing them. */ if (!dp->dp_scan->scn_is_sorted || BP_IS_GANG(bp)) { scan_exec_io(dp, bp, zio_flags, zb, NULL); return; } for (int i = 0; i < BP_GET_NDVAS(bp); i++) { dva_t dva; vdev_t *vdev; dva = bp->blk_dva[i]; vdev = vdev_lookup_top(spa, DVA_GET_VDEV(&dva)); ASSERT(vdev != NULL); mutex_enter(&vdev->vdev_scan_io_queue_lock); if (vdev->vdev_scan_io_queue == NULL) vdev->vdev_scan_io_queue = scan_io_queue_create(vdev); ASSERT(dp->dp_scan != NULL); scan_io_queue_insert(vdev->vdev_scan_io_queue, bp, i, zio_flags, zb); mutex_exit(&vdev->vdev_scan_io_queue_lock); } } static int dsl_scan_scrub_cb(dsl_pool_t *dp, const blkptr_t *bp, const zbookmark_phys_t *zb) { dsl_scan_t *scn = dp->dp_scan; spa_t *spa = dp->dp_spa; uint64_t phys_birth = BP_PHYSICAL_BIRTH(bp); size_t psize = BP_GET_PSIZE(bp); boolean_t needs_io = B_FALSE; int zio_flags = ZIO_FLAG_SCAN_THREAD | ZIO_FLAG_RAW | ZIO_FLAG_CANFAIL; if (phys_birth <= scn->scn_phys.scn_min_txg || phys_birth >= scn->scn_phys.scn_max_txg) { count_block(scn, dp->dp_blkstats, bp); return (0); } /* Embedded BP's have phys_birth==0, so we reject them above. */ ASSERT(!BP_IS_EMBEDDED(bp)); ASSERT(DSL_SCAN_IS_SCRUB_RESILVER(scn)); if (scn->scn_phys.scn_func == POOL_SCAN_SCRUB) { zio_flags |= ZIO_FLAG_SCRUB; needs_io = B_TRUE; } else { ASSERT3U(scn->scn_phys.scn_func, ==, POOL_SCAN_RESILVER); zio_flags |= ZIO_FLAG_RESILVER; needs_io = B_FALSE; } /* If it's an intent log block, failure is expected. */ if (zb->zb_level == ZB_ZIL_LEVEL) zio_flags |= ZIO_FLAG_SPECULATIVE; for (int d = 0; d < BP_GET_NDVAS(bp); d++) { const dva_t *dva = &bp->blk_dva[d]; /* * Keep track of how much data we've examined so that - * zpool(1M) status can make useful progress reports. + * zpool(8) status can make useful progress reports. */ scn->scn_phys.scn_examined += DVA_GET_ASIZE(dva); spa->spa_scan_pass_exam += DVA_GET_ASIZE(dva); /* if it's a resilver, this may not be in the target range */ if (!needs_io) needs_io = dsl_scan_need_resilver(spa, dva, psize, phys_birth); } if (needs_io && !zfs_no_scrub_io) { dsl_scan_enqueue(dp, bp, zio_flags, zb); } else { count_block(scn, dp->dp_blkstats, bp); } /* do not relocate this block */ return (0); } static void dsl_scan_scrub_done(zio_t *zio) { spa_t *spa = zio->io_spa; blkptr_t *bp = zio->io_bp; dsl_scan_io_queue_t *queue = zio->io_private; abd_free(zio->io_abd); if (queue == NULL) { mutex_enter(&spa->spa_scrub_lock); ASSERT3U(spa->spa_scrub_inflight, >=, BP_GET_PSIZE(bp)); spa->spa_scrub_inflight -= BP_GET_PSIZE(bp); cv_broadcast(&spa->spa_scrub_io_cv); mutex_exit(&spa->spa_scrub_lock); } else { mutex_enter(&queue->q_vd->vdev_scan_io_queue_lock); ASSERT3U(queue->q_inflight_bytes, >=, BP_GET_PSIZE(bp)); queue->q_inflight_bytes -= BP_GET_PSIZE(bp); cv_broadcast(&queue->q_zio_cv); mutex_exit(&queue->q_vd->vdev_scan_io_queue_lock); } if (zio->io_error && (zio->io_error != ECKSUM || !(zio->io_flags & ZIO_FLAG_SPECULATIVE))) { atomic_inc_64(&spa->spa_dsl_pool->dp_scan->scn_phys.scn_errors); } } /* * Given a scanning zio's information, executes the zio. The zio need * not necessarily be only sortable, this function simply executes the * zio, no matter what it is. The optional queue argument allows the * caller to specify that they want per top level vdev IO rate limiting * instead of the legacy global limiting. */ static void scan_exec_io(dsl_pool_t *dp, const blkptr_t *bp, int zio_flags, const zbookmark_phys_t *zb, dsl_scan_io_queue_t *queue) { spa_t *spa = dp->dp_spa; dsl_scan_t *scn = dp->dp_scan; size_t size = BP_GET_PSIZE(bp); abd_t *data = abd_alloc_for_io(size, B_FALSE); ASSERT3U(scn->scn_maxinflight_bytes, >, 0); if (queue == NULL) { mutex_enter(&spa->spa_scrub_lock); while (spa->spa_scrub_inflight >= scn->scn_maxinflight_bytes) cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock); spa->spa_scrub_inflight += BP_GET_PSIZE(bp); mutex_exit(&spa->spa_scrub_lock); } else { kmutex_t *q_lock = &queue->q_vd->vdev_scan_io_queue_lock; mutex_enter(q_lock); while (queue->q_inflight_bytes >= queue->q_maxinflight_bytes) cv_wait(&queue->q_zio_cv, q_lock); queue->q_inflight_bytes += BP_GET_PSIZE(bp); mutex_exit(q_lock); } count_block(scn, dp->dp_blkstats, bp); zio_nowait(zio_read(scn->scn_zio_root, spa, bp, data, size, dsl_scan_scrub_done, queue, ZIO_PRIORITY_SCRUB, zio_flags, zb)); } /* * This is the primary extent sorting algorithm. We balance two parameters: * 1) how many bytes of I/O are in an extent * 2) how well the extent is filled with I/O (as a fraction of its total size) * Since we allow extents to have gaps between their constituent I/Os, it's * possible to have a fairly large extent that contains the same amount of * I/O bytes than a much smaller extent, which just packs the I/O more tightly. * The algorithm sorts based on a score calculated from the extent's size, * the relative fill volume (in %) and a "fill weight" parameter that controls * the split between whether we prefer larger extents or more well populated * extents: * * SCORE = FILL_IN_BYTES + (FILL_IN_PERCENT * FILL_IN_BYTES * FILL_WEIGHT) * * Example: * 1) assume extsz = 64 MiB * 2) assume fill = 32 MiB (extent is half full) * 3) assume fill_weight = 3 * 4) SCORE = 32M + (((32M * 100) / 64M) * 3 * 32M) / 100 * SCORE = 32M + (50 * 3 * 32M) / 100 * SCORE = 32M + (4800M / 100) * SCORE = 32M + 48M * ^ ^ * | +--- final total relative fill-based score * +--------- final total fill-based score * SCORE = 80M * * As can be seen, at fill_ratio=3, the algorithm is slightly biased towards * extents that are more completely filled (in a 3:2 ratio) vs just larger. * Note that as an optimization, we replace multiplication and division by * 100 with bitshifting by 7 (which effectively multiplies and divides by 128). */ static int ext_size_compare(const void *x, const void *y) { const range_seg_gap_t *rsa = x, *rsb = y; uint64_t sa = rsa->rs_end - rsa->rs_start; uint64_t sb = rsb->rs_end - rsb->rs_start; uint64_t score_a, score_b; score_a = rsa->rs_fill + ((((rsa->rs_fill << 7) / sa) * fill_weight * rsa->rs_fill) >> 7); score_b = rsb->rs_fill + ((((rsb->rs_fill << 7) / sb) * fill_weight * rsb->rs_fill) >> 7); if (score_a > score_b) return (-1); if (score_a == score_b) { if (rsa->rs_start < rsb->rs_start) return (-1); if (rsa->rs_start == rsb->rs_start) return (0); return (1); } return (1); } /* * Comparator for the q_sios_by_addr tree. Sorting is simply performed * based on LBA-order (from lowest to highest). */ static int sio_addr_compare(const void *x, const void *y) { const scan_io_t *a = x, *b = y; return (TREE_CMP(SIO_GET_OFFSET(a), SIO_GET_OFFSET(b))); } /* IO queues are created on demand when they are needed. */ static dsl_scan_io_queue_t * scan_io_queue_create(vdev_t *vd) { dsl_scan_t *scn = vd->vdev_spa->spa_dsl_pool->dp_scan; dsl_scan_io_queue_t *q = kmem_zalloc(sizeof (*q), KM_SLEEP); q->q_scn = scn; q->q_vd = vd; q->q_sio_memused = 0; cv_init(&q->q_zio_cv, NULL, CV_DEFAULT, NULL); q->q_exts_by_addr = range_tree_create_impl(&rt_btree_ops, RANGE_SEG_GAP, &q->q_exts_by_size, 0, 0, ext_size_compare, zfs_scan_max_ext_gap); avl_create(&q->q_sios_by_addr, sio_addr_compare, sizeof (scan_io_t), offsetof(scan_io_t, sio_nodes.sio_addr_node)); return (q); } /* * Destroys a scan queue and all segments and scan_io_t's contained in it. * No further execution of I/O occurs, anything pending in the queue is * simply freed without being executed. */ void dsl_scan_io_queue_destroy(dsl_scan_io_queue_t *queue) { dsl_scan_t *scn = queue->q_scn; scan_io_t *sio; void *cookie = NULL; int64_t bytes_dequeued = 0; ASSERT(MUTEX_HELD(&queue->q_vd->vdev_scan_io_queue_lock)); while ((sio = avl_destroy_nodes(&queue->q_sios_by_addr, &cookie)) != NULL) { ASSERT(range_tree_contains(queue->q_exts_by_addr, SIO_GET_OFFSET(sio), SIO_GET_ASIZE(sio))); bytes_dequeued += SIO_GET_ASIZE(sio); queue->q_sio_memused -= SIO_GET_MUSED(sio); sio_free(sio); } ASSERT0(queue->q_sio_memused); atomic_add_64(&scn->scn_bytes_pending, -bytes_dequeued); range_tree_vacate(queue->q_exts_by_addr, NULL, queue); range_tree_destroy(queue->q_exts_by_addr); avl_destroy(&queue->q_sios_by_addr); cv_destroy(&queue->q_zio_cv); kmem_free(queue, sizeof (*queue)); } /* * Properly transfers a dsl_scan_queue_t from `svd' to `tvd'. This is * called on behalf of vdev_top_transfer when creating or destroying * a mirror vdev due to zpool attach/detach. */ void dsl_scan_io_queue_vdev_xfer(vdev_t *svd, vdev_t *tvd) { mutex_enter(&svd->vdev_scan_io_queue_lock); mutex_enter(&tvd->vdev_scan_io_queue_lock); VERIFY3P(tvd->vdev_scan_io_queue, ==, NULL); tvd->vdev_scan_io_queue = svd->vdev_scan_io_queue; svd->vdev_scan_io_queue = NULL; if (tvd->vdev_scan_io_queue != NULL) tvd->vdev_scan_io_queue->q_vd = tvd; mutex_exit(&tvd->vdev_scan_io_queue_lock); mutex_exit(&svd->vdev_scan_io_queue_lock); } static void scan_io_queues_destroy(dsl_scan_t *scn) { vdev_t *rvd = scn->scn_dp->dp_spa->spa_root_vdev; for (uint64_t i = 0; i < rvd->vdev_children; i++) { vdev_t *tvd = rvd->vdev_child[i]; mutex_enter(&tvd->vdev_scan_io_queue_lock); if (tvd->vdev_scan_io_queue != NULL) dsl_scan_io_queue_destroy(tvd->vdev_scan_io_queue); tvd->vdev_scan_io_queue = NULL; mutex_exit(&tvd->vdev_scan_io_queue_lock); } } static void dsl_scan_freed_dva(spa_t *spa, const blkptr_t *bp, int dva_i) { dsl_pool_t *dp = spa->spa_dsl_pool; dsl_scan_t *scn = dp->dp_scan; vdev_t *vdev; kmutex_t *q_lock; dsl_scan_io_queue_t *queue; scan_io_t *srch_sio, *sio; avl_index_t idx; uint64_t start, size; vdev = vdev_lookup_top(spa, DVA_GET_VDEV(&bp->blk_dva[dva_i])); ASSERT(vdev != NULL); q_lock = &vdev->vdev_scan_io_queue_lock; queue = vdev->vdev_scan_io_queue; mutex_enter(q_lock); if (queue == NULL) { mutex_exit(q_lock); return; } srch_sio = sio_alloc(BP_GET_NDVAS(bp)); bp2sio(bp, srch_sio, dva_i); start = SIO_GET_OFFSET(srch_sio); size = SIO_GET_ASIZE(srch_sio); /* * We can find the zio in two states: * 1) Cold, just sitting in the queue of zio's to be issued at * some point in the future. In this case, all we do is * remove the zio from the q_sios_by_addr tree, decrement * its data volume from the containing range_seg_t and * resort the q_exts_by_size tree to reflect that the * range_seg_t has lost some of its 'fill'. We don't shorten * the range_seg_t - this is usually rare enough not to be * worth the extra hassle of trying keep track of precise * extent boundaries. * 2) Hot, where the zio is currently in-flight in * dsl_scan_issue_ios. In this case, we can't simply * reach in and stop the in-flight zio's, so we instead * block the caller. Eventually, dsl_scan_issue_ios will * be done with issuing the zio's it gathered and will * signal us. */ sio = avl_find(&queue->q_sios_by_addr, srch_sio, &idx); sio_free(srch_sio); if (sio != NULL) { int64_t asize = SIO_GET_ASIZE(sio); blkptr_t tmpbp; /* Got it while it was cold in the queue */ ASSERT3U(start, ==, SIO_GET_OFFSET(sio)); ASSERT3U(size, ==, asize); avl_remove(&queue->q_sios_by_addr, sio); queue->q_sio_memused -= SIO_GET_MUSED(sio); ASSERT(range_tree_contains(queue->q_exts_by_addr, start, size)); range_tree_remove_fill(queue->q_exts_by_addr, start, size); /* * We only update scn_bytes_pending in the cold path, * otherwise it will already have been accounted for as * part of the zio's execution. */ atomic_add_64(&scn->scn_bytes_pending, -asize); /* count the block as though we issued it */ sio2bp(sio, &tmpbp); count_block(scn, dp->dp_blkstats, &tmpbp); sio_free(sio); } mutex_exit(q_lock); } /* * Callback invoked when a zio_free() zio is executing. This needs to be * intercepted to prevent the zio from deallocating a particular portion * of disk space and it then getting reallocated and written to, while we * still have it queued up for processing. */ void dsl_scan_freed(spa_t *spa, const blkptr_t *bp) { dsl_pool_t *dp = spa->spa_dsl_pool; dsl_scan_t *scn = dp->dp_scan; ASSERT(!BP_IS_EMBEDDED(bp)); ASSERT(scn != NULL); if (!dsl_scan_is_running(scn)) return; for (int i = 0; i < BP_GET_NDVAS(bp); i++) dsl_scan_freed_dva(spa, bp, i); } /* * Check if a vdev needs resilvering (non-empty DTL), if so, and resilver has * not started, start it. Otherwise, only restart if max txg in DTL range is * greater than the max txg in the current scan. If the DTL max is less than * the scan max, then the vdev has not missed any new data since the resilver * started, so a restart is not needed. */ void dsl_scan_assess_vdev(dsl_pool_t *dp, vdev_t *vd) { uint64_t min, max; if (!vdev_resilver_needed(vd, &min, &max)) return; if (!dsl_scan_resilvering(dp)) { spa_async_request(dp->dp_spa, SPA_ASYNC_RESILVER); return; } if (max <= dp->dp_scan->scn_phys.scn_max_txg) return; /* restart is needed, check if it can be deferred */ if (spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_RESILVER_DEFER)) vdev_defer_resilver(vd); else spa_async_request(dp->dp_spa, SPA_ASYNC_RESILVER); } /* BEGIN CSTYLED */ ZFS_MODULE_PARAM(zfs, zfs_, scan_vdev_limit, ULONG, ZMOD_RW, "Max bytes in flight per leaf vdev for scrubs and resilvers"); ZFS_MODULE_PARAM(zfs, zfs_, scrub_min_time_ms, INT, ZMOD_RW, "Min millisecs to scrub per txg"); ZFS_MODULE_PARAM(zfs, zfs_, obsolete_min_time_ms, INT, ZMOD_RW, "Min millisecs to obsolete per txg"); ZFS_MODULE_PARAM(zfs, zfs_, free_min_time_ms, INT, ZMOD_RW, "Min millisecs to free per txg"); ZFS_MODULE_PARAM(zfs, zfs_, resilver_min_time_ms, INT, ZMOD_RW, "Min millisecs to resilver per txg"); ZFS_MODULE_PARAM(zfs, zfs_, scan_suspend_progress, INT, ZMOD_RW, "Set to prevent scans from progressing"); ZFS_MODULE_PARAM(zfs, zfs_, no_scrub_io, INT, ZMOD_RW, "Set to disable scrub I/O"); ZFS_MODULE_PARAM(zfs, zfs_, no_scrub_prefetch, INT, ZMOD_RW, "Set to disable scrub prefetching"); ZFS_MODULE_PARAM(zfs, zfs_, async_block_max_blocks, ULONG, ZMOD_RW, "Max number of blocks freed in one txg"); ZFS_MODULE_PARAM(zfs, zfs_, max_async_dedup_frees, ULONG, ZMOD_RW, "Max number of dedup blocks freed in one txg"); ZFS_MODULE_PARAM(zfs, zfs_, free_bpobj_enabled, INT, ZMOD_RW, "Enable processing of the free_bpobj"); ZFS_MODULE_PARAM(zfs, zfs_, scan_mem_lim_fact, INT, ZMOD_RW, "Fraction of RAM for scan hard limit"); ZFS_MODULE_PARAM(zfs, zfs_, scan_issue_strategy, INT, ZMOD_RW, "IO issuing strategy during scrubbing. " "0 = default, 1 = LBA, 2 = size"); ZFS_MODULE_PARAM(zfs, zfs_, scan_legacy, INT, ZMOD_RW, "Scrub using legacy non-sequential method"); ZFS_MODULE_PARAM(zfs, zfs_, scan_checkpoint_intval, INT, ZMOD_RW, "Scan progress on-disk checkpointing interval"); ZFS_MODULE_PARAM(zfs, zfs_, scan_max_ext_gap, ULONG, ZMOD_RW, "Max gap in bytes between sequential scrub / resilver I/Os"); ZFS_MODULE_PARAM(zfs, zfs_, scan_mem_lim_soft_fact, INT, ZMOD_RW, "Fraction of hard limit used as soft limit"); ZFS_MODULE_PARAM(zfs, zfs_, scan_strict_mem_lim, INT, ZMOD_RW, "Tunable to attempt to reduce lock contention"); ZFS_MODULE_PARAM(zfs, zfs_, scan_fill_weight, INT, ZMOD_RW, "Tunable to adjust bias towards more filled segments during scans"); ZFS_MODULE_PARAM(zfs, zfs_, resilver_disable_defer, INT, ZMOD_RW, "Process all resilvers immediately"); /* END CSTYLED */ diff --git a/module/zfs/metaslab.c b/module/zfs/metaslab.c index 133005b227e5..325f505b72d1 100644 --- a/module/zfs/metaslab.c +++ b/module/zfs/metaslab.c @@ -1,6237 +1,6237 @@ /* * 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) 2011, 2019 by Delphix. All rights reserved. * Copyright (c) 2013 by Saso Kiselkov. All rights reserved. * Copyright (c) 2015, Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2017, Intel Corporation. */ #include #include #include #include #include #include #include #include #include #include #include #include #define WITH_DF_BLOCK_ALLOCATOR #define GANG_ALLOCATION(flags) \ ((flags) & (METASLAB_GANG_CHILD | METASLAB_GANG_HEADER)) /* * Metaslab granularity, in bytes. This is roughly similar to what would be * referred to as the "stripe size" in traditional RAID arrays. In normal * operation, we will try to write this amount of data to a top-level vdev * before moving on to the next one. */ unsigned long metaslab_aliquot = 512 << 10; /* * For testing, make some blocks above a certain size be gang blocks. */ unsigned long metaslab_force_ganging = SPA_MAXBLOCKSIZE + 1; /* * In pools where the log space map feature is not enabled we touch * multiple metaslabs (and their respective space maps) with each * transaction group. Thus, we benefit from having a small space map * block size since it allows us to issue more I/O operations scattered * around the disk. So a sane default for the space map block size * is 8~16K. */ int zfs_metaslab_sm_blksz_no_log = (1 << 14); /* * When the log space map feature is enabled, we accumulate a lot of * changes per metaslab that are flushed once in a while so we benefit * from a bigger block size like 128K for the metaslab space maps. */ int zfs_metaslab_sm_blksz_with_log = (1 << 17); /* * The in-core space map representation is more compact than its on-disk form. * The zfs_condense_pct determines how much more compact the in-core * space map representation must be before we compact it on-disk. * Values should be greater than or equal to 100. */ int zfs_condense_pct = 200; /* * Condensing a metaslab is not guaranteed to actually reduce the amount of * space used on disk. In particular, a space map uses data in increments of * MAX(1 << ashift, space_map_blksz), so a metaslab might use the * same number of blocks after condensing. Since the goal of condensing is to * reduce the number of IOPs required to read the space map, we only want to * condense when we can be sure we will reduce the number of blocks used by the * space map. Unfortunately, we cannot precisely compute whether or not this is * the case in metaslab_should_condense since we are holding ms_lock. Instead, * we apply the following heuristic: do not condense a spacemap unless the * uncondensed size consumes greater than zfs_metaslab_condense_block_threshold * blocks. */ int zfs_metaslab_condense_block_threshold = 4; /* * The zfs_mg_noalloc_threshold defines which metaslab groups should * be eligible for allocation. The value is defined as a percentage of * free space. Metaslab groups that have more free space than * zfs_mg_noalloc_threshold are always eligible for allocations. Once * a metaslab group's free space is less than or equal to the * zfs_mg_noalloc_threshold the allocator will avoid allocating to that * group unless all groups in the pool have reached zfs_mg_noalloc_threshold. * Once all groups in the pool reach zfs_mg_noalloc_threshold then all * groups are allowed to accept allocations. Gang blocks are always * eligible to allocate on any metaslab group. The default value of 0 means * no metaslab group will be excluded based on this criterion. */ int zfs_mg_noalloc_threshold = 0; /* * Metaslab groups are considered eligible for allocations if their * fragmentation metric (measured as a percentage) is less than or * equal to zfs_mg_fragmentation_threshold. If a metaslab group * exceeds this threshold then it will be skipped unless all metaslab * groups within the metaslab class have also crossed this threshold. * * This tunable was introduced to avoid edge cases where we continue * allocating from very fragmented disks in our pool while other, less * fragmented disks, exists. On the other hand, if all disks in the * pool are uniformly approaching the threshold, the threshold can * be a speed bump in performance, where we keep switching the disks * that we allocate from (e.g. we allocate some segments from disk A * making it bypassing the threshold while freeing segments from disk * B getting its fragmentation below the threshold). * * Empirically, we've seen that our vdev selection for allocations is * good enough that fragmentation increases uniformly across all vdevs * the majority of the time. Thus we set the threshold percentage high * enough to avoid hitting the speed bump on pools that are being pushed * to the edge. */ int zfs_mg_fragmentation_threshold = 95; /* * Allow metaslabs to keep their active state as long as their fragmentation * percentage is less than or equal to zfs_metaslab_fragmentation_threshold. An * active metaslab that exceeds this threshold will no longer keep its active * status allowing better metaslabs to be selected. */ int zfs_metaslab_fragmentation_threshold = 70; /* * When set will load all metaslabs when pool is first opened. */ int metaslab_debug_load = 0; /* * When set will prevent metaslabs from being unloaded. */ int metaslab_debug_unload = 0; /* * Minimum size which forces the dynamic allocator to change * it's allocation strategy. Once the space map cannot satisfy * an allocation of this size then it switches to using more * aggressive strategy (i.e search by size rather than offset). */ uint64_t metaslab_df_alloc_threshold = SPA_OLD_MAXBLOCKSIZE; /* * The minimum free space, in percent, which must be available * in a space map to continue allocations in a first-fit fashion. * Once the space map's free space drops below this level we dynamically * switch to using best-fit allocations. */ int metaslab_df_free_pct = 4; /* * Maximum distance to search forward from the last offset. Without this * limit, fragmented pools can see >100,000 iterations and * metaslab_block_picker() becomes the performance limiting factor on * high-performance storage. * * With the default setting of 16MB, we typically see less than 500 * iterations, even with very fragmented, ashift=9 pools. The maximum number * of iterations possible is: * metaslab_df_max_search / (2 * (1<ks_data = &metaslab_stats; kstat_install(metaslab_ksp); } } void metaslab_stat_fini(void) { if (metaslab_ksp != NULL) { kstat_delete(metaslab_ksp); metaslab_ksp = NULL; } kmem_cache_destroy(metaslab_alloc_trace_cache); metaslab_alloc_trace_cache = NULL; } #else void metaslab_stat_init(void) { } void metaslab_stat_fini(void) { } #endif /* * ========================================================================== * Metaslab classes * ========================================================================== */ metaslab_class_t * metaslab_class_create(spa_t *spa, metaslab_ops_t *ops) { metaslab_class_t *mc; mc = kmem_zalloc(sizeof (metaslab_class_t), KM_SLEEP); mc->mc_spa = spa; mc->mc_rotor = NULL; mc->mc_ops = ops; mutex_init(&mc->mc_lock, NULL, MUTEX_DEFAULT, NULL); mc->mc_metaslab_txg_list = multilist_create(sizeof (metaslab_t), offsetof(metaslab_t, ms_class_txg_node), metaslab_idx_func); mc->mc_alloc_slots = kmem_zalloc(spa->spa_alloc_count * sizeof (zfs_refcount_t), KM_SLEEP); mc->mc_alloc_max_slots = kmem_zalloc(spa->spa_alloc_count * sizeof (uint64_t), KM_SLEEP); for (int i = 0; i < spa->spa_alloc_count; i++) zfs_refcount_create_tracked(&mc->mc_alloc_slots[i]); return (mc); } void metaslab_class_destroy(metaslab_class_t *mc) { ASSERT(mc->mc_rotor == NULL); ASSERT(mc->mc_alloc == 0); ASSERT(mc->mc_deferred == 0); ASSERT(mc->mc_space == 0); ASSERT(mc->mc_dspace == 0); for (int i = 0; i < mc->mc_spa->spa_alloc_count; i++) zfs_refcount_destroy(&mc->mc_alloc_slots[i]); kmem_free(mc->mc_alloc_slots, mc->mc_spa->spa_alloc_count * sizeof (zfs_refcount_t)); kmem_free(mc->mc_alloc_max_slots, mc->mc_spa->spa_alloc_count * sizeof (uint64_t)); mutex_destroy(&mc->mc_lock); multilist_destroy(mc->mc_metaslab_txg_list); kmem_free(mc, sizeof (metaslab_class_t)); } int metaslab_class_validate(metaslab_class_t *mc) { metaslab_group_t *mg; vdev_t *vd; /* * Must hold one of the spa_config locks. */ ASSERT(spa_config_held(mc->mc_spa, SCL_ALL, RW_READER) || spa_config_held(mc->mc_spa, SCL_ALL, RW_WRITER)); if ((mg = mc->mc_rotor) == NULL) return (0); do { vd = mg->mg_vd; ASSERT(vd->vdev_mg != NULL); ASSERT3P(vd->vdev_top, ==, vd); ASSERT3P(mg->mg_class, ==, mc); ASSERT3P(vd->vdev_ops, !=, &vdev_hole_ops); } while ((mg = mg->mg_next) != mc->mc_rotor); return (0); } static void metaslab_class_space_update(metaslab_class_t *mc, int64_t alloc_delta, int64_t defer_delta, int64_t space_delta, int64_t dspace_delta) { atomic_add_64(&mc->mc_alloc, alloc_delta); atomic_add_64(&mc->mc_deferred, defer_delta); atomic_add_64(&mc->mc_space, space_delta); atomic_add_64(&mc->mc_dspace, dspace_delta); } uint64_t metaslab_class_get_alloc(metaslab_class_t *mc) { return (mc->mc_alloc); } uint64_t metaslab_class_get_deferred(metaslab_class_t *mc) { return (mc->mc_deferred); } uint64_t metaslab_class_get_space(metaslab_class_t *mc) { return (mc->mc_space); } uint64_t metaslab_class_get_dspace(metaslab_class_t *mc) { return (spa_deflate(mc->mc_spa) ? mc->mc_dspace : mc->mc_space); } void metaslab_class_histogram_verify(metaslab_class_t *mc) { spa_t *spa = mc->mc_spa; vdev_t *rvd = spa->spa_root_vdev; uint64_t *mc_hist; int i; if ((zfs_flags & ZFS_DEBUG_HISTOGRAM_VERIFY) == 0) return; mc_hist = kmem_zalloc(sizeof (uint64_t) * RANGE_TREE_HISTOGRAM_SIZE, KM_SLEEP); for (int c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; metaslab_group_t *mg = tvd->vdev_mg; /* * Skip any holes, uninitialized top-levels, or * vdevs that are not in this metalab class. */ if (!vdev_is_concrete(tvd) || tvd->vdev_ms_shift == 0 || mg->mg_class != mc) { continue; } for (i = 0; i < RANGE_TREE_HISTOGRAM_SIZE; i++) mc_hist[i] += mg->mg_histogram[i]; } for (i = 0; i < RANGE_TREE_HISTOGRAM_SIZE; i++) VERIFY3U(mc_hist[i], ==, mc->mc_histogram[i]); kmem_free(mc_hist, sizeof (uint64_t) * RANGE_TREE_HISTOGRAM_SIZE); } /* * Calculate the metaslab class's fragmentation metric. The metric * is weighted based on the space contribution of each metaslab group. * The return value will be a number between 0 and 100 (inclusive), or * ZFS_FRAG_INVALID if the metric has not been set. See comment above the * zfs_frag_table for more information about the metric. */ uint64_t metaslab_class_fragmentation(metaslab_class_t *mc) { vdev_t *rvd = mc->mc_spa->spa_root_vdev; uint64_t fragmentation = 0; spa_config_enter(mc->mc_spa, SCL_VDEV, FTAG, RW_READER); for (int c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; metaslab_group_t *mg = tvd->vdev_mg; /* * Skip any holes, uninitialized top-levels, * or vdevs that are not in this metalab class. */ if (!vdev_is_concrete(tvd) || tvd->vdev_ms_shift == 0 || mg->mg_class != mc) { continue; } /* * If a metaslab group does not contain a fragmentation * metric then just bail out. */ if (mg->mg_fragmentation == ZFS_FRAG_INVALID) { spa_config_exit(mc->mc_spa, SCL_VDEV, FTAG); return (ZFS_FRAG_INVALID); } /* * Determine how much this metaslab_group is contributing * to the overall pool fragmentation metric. */ fragmentation += mg->mg_fragmentation * metaslab_group_get_space(mg); } fragmentation /= metaslab_class_get_space(mc); ASSERT3U(fragmentation, <=, 100); spa_config_exit(mc->mc_spa, SCL_VDEV, FTAG); return (fragmentation); } /* * Calculate the amount of expandable space that is available in * this metaslab class. If a device is expanded then its expandable * space will be the amount of allocatable space that is currently not * part of this metaslab class. */ uint64_t metaslab_class_expandable_space(metaslab_class_t *mc) { vdev_t *rvd = mc->mc_spa->spa_root_vdev; uint64_t space = 0; spa_config_enter(mc->mc_spa, SCL_VDEV, FTAG, RW_READER); for (int c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; metaslab_group_t *mg = tvd->vdev_mg; if (!vdev_is_concrete(tvd) || tvd->vdev_ms_shift == 0 || mg->mg_class != mc) { continue; } /* * Calculate if we have enough space to add additional * metaslabs. We report the expandable space in terms * of the metaslab size since that's the unit of expansion. */ space += P2ALIGN(tvd->vdev_max_asize - tvd->vdev_asize, 1ULL << tvd->vdev_ms_shift); } spa_config_exit(mc->mc_spa, SCL_VDEV, FTAG); return (space); } void metaslab_class_evict_old(metaslab_class_t *mc, uint64_t txg) { multilist_t *ml = mc->mc_metaslab_txg_list; for (int i = 0; i < multilist_get_num_sublists(ml); i++) { multilist_sublist_t *mls = multilist_sublist_lock(ml, i); metaslab_t *msp = multilist_sublist_head(mls); multilist_sublist_unlock(mls); while (msp != NULL) { mutex_enter(&msp->ms_lock); /* * If the metaslab has been removed from the list * (which could happen if we were at the memory limit * and it was evicted during this loop), then we can't * proceed and we should restart the sublist. */ if (!multilist_link_active(&msp->ms_class_txg_node)) { mutex_exit(&msp->ms_lock); i--; break; } mls = multilist_sublist_lock(ml, i); metaslab_t *next_msp = multilist_sublist_next(mls, msp); multilist_sublist_unlock(mls); if (txg > msp->ms_selected_txg + metaslab_unload_delay && gethrtime() > msp->ms_selected_time + (uint64_t)MSEC2NSEC(metaslab_unload_delay_ms)) { metaslab_evict(msp, txg); } else { /* * Once we've hit a metaslab selected too * recently to evict, we're done evicting for * now. */ mutex_exit(&msp->ms_lock); break; } mutex_exit(&msp->ms_lock); msp = next_msp; } } } static int metaslab_compare(const void *x1, const void *x2) { const metaslab_t *m1 = (const metaslab_t *)x1; const metaslab_t *m2 = (const metaslab_t *)x2; int sort1 = 0; int sort2 = 0; if (m1->ms_allocator != -1 && m1->ms_primary) sort1 = 1; else if (m1->ms_allocator != -1 && !m1->ms_primary) sort1 = 2; if (m2->ms_allocator != -1 && m2->ms_primary) sort2 = 1; else if (m2->ms_allocator != -1 && !m2->ms_primary) sort2 = 2; /* * Sort inactive metaslabs first, then primaries, then secondaries. When * selecting a metaslab to allocate from, an allocator first tries its * primary, then secondary active metaslab. If it doesn't have active * metaslabs, or can't allocate from them, it searches for an inactive * metaslab to activate. If it can't find a suitable one, it will steal * a primary or secondary metaslab from another allocator. */ if (sort1 < sort2) return (-1); if (sort1 > sort2) return (1); int cmp = TREE_CMP(m2->ms_weight, m1->ms_weight); if (likely(cmp)) return (cmp); IMPLY(TREE_CMP(m1->ms_start, m2->ms_start) == 0, m1 == m2); return (TREE_CMP(m1->ms_start, m2->ms_start)); } /* * ========================================================================== * Metaslab groups * ========================================================================== */ /* * Update the allocatable flag and the metaslab group's capacity. * The allocatable flag is set to true if the capacity is below * the zfs_mg_noalloc_threshold or has a fragmentation value that is * greater than zfs_mg_fragmentation_threshold. If a metaslab group * transitions from allocatable to non-allocatable or vice versa then the * metaslab group's class is updated to reflect the transition. */ static void metaslab_group_alloc_update(metaslab_group_t *mg) { vdev_t *vd = mg->mg_vd; metaslab_class_t *mc = mg->mg_class; vdev_stat_t *vs = &vd->vdev_stat; boolean_t was_allocatable; boolean_t was_initialized; ASSERT(vd == vd->vdev_top); ASSERT3U(spa_config_held(mc->mc_spa, SCL_ALLOC, RW_READER), ==, SCL_ALLOC); mutex_enter(&mg->mg_lock); was_allocatable = mg->mg_allocatable; was_initialized = mg->mg_initialized; mg->mg_free_capacity = ((vs->vs_space - vs->vs_alloc) * 100) / (vs->vs_space + 1); mutex_enter(&mc->mc_lock); /* * If the metaslab group was just added then it won't * have any space until we finish syncing out this txg. * At that point we will consider it initialized and available * for allocations. We also don't consider non-activated * metaslab groups (e.g. vdevs that are in the middle of being removed) * to be initialized, because they can't be used for allocation. */ mg->mg_initialized = metaslab_group_initialized(mg); if (!was_initialized && mg->mg_initialized) { mc->mc_groups++; } else if (was_initialized && !mg->mg_initialized) { ASSERT3U(mc->mc_groups, >, 0); mc->mc_groups--; } if (mg->mg_initialized) mg->mg_no_free_space = B_FALSE; /* * A metaslab group is considered allocatable if it has plenty * of free space or is not heavily fragmented. We only take * fragmentation into account if the metaslab group has a valid * fragmentation metric (i.e. a value between 0 and 100). */ mg->mg_allocatable = (mg->mg_activation_count > 0 && mg->mg_free_capacity > zfs_mg_noalloc_threshold && (mg->mg_fragmentation == ZFS_FRAG_INVALID || mg->mg_fragmentation <= zfs_mg_fragmentation_threshold)); /* * The mc_alloc_groups maintains a count of the number of * groups in this metaslab class that are still above the * zfs_mg_noalloc_threshold. This is used by the allocating * threads to determine if they should avoid allocations to * a given group. The allocator will avoid allocations to a group * if that group has reached or is below the zfs_mg_noalloc_threshold * and there are still other groups that are above the threshold. * When a group transitions from allocatable to non-allocatable or * vice versa we update the metaslab class to reflect that change. * When the mc_alloc_groups value drops to 0 that means that all * groups have reached the zfs_mg_noalloc_threshold making all groups * eligible for allocations. This effectively means that all devices * are balanced again. */ if (was_allocatable && !mg->mg_allocatable) mc->mc_alloc_groups--; else if (!was_allocatable && mg->mg_allocatable) mc->mc_alloc_groups++; mutex_exit(&mc->mc_lock); mutex_exit(&mg->mg_lock); } int metaslab_sort_by_flushed(const void *va, const void *vb) { const metaslab_t *a = va; const metaslab_t *b = vb; int cmp = TREE_CMP(a->ms_unflushed_txg, b->ms_unflushed_txg); if (likely(cmp)) return (cmp); uint64_t a_vdev_id = a->ms_group->mg_vd->vdev_id; uint64_t b_vdev_id = b->ms_group->mg_vd->vdev_id; cmp = TREE_CMP(a_vdev_id, b_vdev_id); if (cmp) return (cmp); return (TREE_CMP(a->ms_id, b->ms_id)); } metaslab_group_t * metaslab_group_create(metaslab_class_t *mc, vdev_t *vd, int allocators) { metaslab_group_t *mg; mg = kmem_zalloc(sizeof (metaslab_group_t), KM_SLEEP); mutex_init(&mg->mg_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&mg->mg_ms_disabled_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&mg->mg_ms_disabled_cv, NULL, CV_DEFAULT, NULL); avl_create(&mg->mg_metaslab_tree, metaslab_compare, sizeof (metaslab_t), offsetof(metaslab_t, ms_group_node)); mg->mg_vd = vd; mg->mg_class = mc; mg->mg_activation_count = 0; mg->mg_initialized = B_FALSE; mg->mg_no_free_space = B_TRUE; mg->mg_allocators = allocators; mg->mg_allocator = kmem_zalloc(allocators * sizeof (metaslab_group_allocator_t), KM_SLEEP); for (int i = 0; i < allocators; i++) { metaslab_group_allocator_t *mga = &mg->mg_allocator[i]; zfs_refcount_create_tracked(&mga->mga_alloc_queue_depth); } mg->mg_taskq = taskq_create("metaslab_group_taskq", metaslab_load_pct, maxclsyspri, 10, INT_MAX, TASKQ_THREADS_CPU_PCT | TASKQ_DYNAMIC); return (mg); } void metaslab_group_destroy(metaslab_group_t *mg) { ASSERT(mg->mg_prev == NULL); ASSERT(mg->mg_next == NULL); /* * We may have gone below zero with the activation count * either because we never activated in the first place or * because we're done, and possibly removing the vdev. */ ASSERT(mg->mg_activation_count <= 0); taskq_destroy(mg->mg_taskq); avl_destroy(&mg->mg_metaslab_tree); mutex_destroy(&mg->mg_lock); mutex_destroy(&mg->mg_ms_disabled_lock); cv_destroy(&mg->mg_ms_disabled_cv); for (int i = 0; i < mg->mg_allocators; i++) { metaslab_group_allocator_t *mga = &mg->mg_allocator[i]; zfs_refcount_destroy(&mga->mga_alloc_queue_depth); } kmem_free(mg->mg_allocator, mg->mg_allocators * sizeof (metaslab_group_allocator_t)); kmem_free(mg, sizeof (metaslab_group_t)); } void metaslab_group_activate(metaslab_group_t *mg) { metaslab_class_t *mc = mg->mg_class; metaslab_group_t *mgprev, *mgnext; ASSERT3U(spa_config_held(mc->mc_spa, SCL_ALLOC, RW_WRITER), !=, 0); ASSERT(mc->mc_rotor != mg); ASSERT(mg->mg_prev == NULL); ASSERT(mg->mg_next == NULL); ASSERT(mg->mg_activation_count <= 0); if (++mg->mg_activation_count <= 0) return; mg->mg_aliquot = metaslab_aliquot * MAX(1, mg->mg_vd->vdev_children); metaslab_group_alloc_update(mg); if ((mgprev = mc->mc_rotor) == NULL) { mg->mg_prev = mg; mg->mg_next = mg; } else { mgnext = mgprev->mg_next; mg->mg_prev = mgprev; mg->mg_next = mgnext; mgprev->mg_next = mg; mgnext->mg_prev = mg; } mc->mc_rotor = mg; } /* * Passivate a metaslab group and remove it from the allocation rotor. * Callers must hold both the SCL_ALLOC and SCL_ZIO lock prior to passivating * a metaslab group. This function will momentarily drop spa_config_locks * that are lower than the SCL_ALLOC lock (see comment below). */ void metaslab_group_passivate(metaslab_group_t *mg) { metaslab_class_t *mc = mg->mg_class; spa_t *spa = mc->mc_spa; metaslab_group_t *mgprev, *mgnext; int locks = spa_config_held(spa, SCL_ALL, RW_WRITER); ASSERT3U(spa_config_held(spa, SCL_ALLOC | SCL_ZIO, RW_WRITER), ==, (SCL_ALLOC | SCL_ZIO)); if (--mg->mg_activation_count != 0) { ASSERT(mc->mc_rotor != mg); ASSERT(mg->mg_prev == NULL); ASSERT(mg->mg_next == NULL); ASSERT(mg->mg_activation_count < 0); return; } /* * The spa_config_lock is an array of rwlocks, ordered as * follows (from highest to lowest): * SCL_CONFIG > SCL_STATE > SCL_L2ARC > SCL_ALLOC > * SCL_ZIO > SCL_FREE > SCL_VDEV * (For more information about the spa_config_lock see spa_misc.c) * The higher the lock, the broader its coverage. When we passivate * a metaslab group, we must hold both the SCL_ALLOC and the SCL_ZIO * config locks. However, the metaslab group's taskq might be trying * to preload metaslabs so we must drop the SCL_ZIO lock and any * lower locks to allow the I/O to complete. At a minimum, * we continue to hold the SCL_ALLOC lock, which prevents any future * allocations from taking place and any changes to the vdev tree. */ spa_config_exit(spa, locks & ~(SCL_ZIO - 1), spa); taskq_wait_outstanding(mg->mg_taskq, 0); spa_config_enter(spa, locks & ~(SCL_ZIO - 1), spa, RW_WRITER); metaslab_group_alloc_update(mg); for (int i = 0; i < mg->mg_allocators; i++) { metaslab_group_allocator_t *mga = &mg->mg_allocator[i]; metaslab_t *msp = mga->mga_primary; if (msp != NULL) { mutex_enter(&msp->ms_lock); metaslab_passivate(msp, metaslab_weight_from_range_tree(msp)); mutex_exit(&msp->ms_lock); } msp = mga->mga_secondary; if (msp != NULL) { mutex_enter(&msp->ms_lock); metaslab_passivate(msp, metaslab_weight_from_range_tree(msp)); mutex_exit(&msp->ms_lock); } } mgprev = mg->mg_prev; mgnext = mg->mg_next; if (mg == mgnext) { mc->mc_rotor = NULL; } else { mc->mc_rotor = mgnext; mgprev->mg_next = mgnext; mgnext->mg_prev = mgprev; } mg->mg_prev = NULL; mg->mg_next = NULL; } boolean_t metaslab_group_initialized(metaslab_group_t *mg) { vdev_t *vd = mg->mg_vd; vdev_stat_t *vs = &vd->vdev_stat; return (vs->vs_space != 0 && mg->mg_activation_count > 0); } uint64_t metaslab_group_get_space(metaslab_group_t *mg) { return ((1ULL << mg->mg_vd->vdev_ms_shift) * mg->mg_vd->vdev_ms_count); } void metaslab_group_histogram_verify(metaslab_group_t *mg) { uint64_t *mg_hist; vdev_t *vd = mg->mg_vd; uint64_t ashift = vd->vdev_ashift; int i; if ((zfs_flags & ZFS_DEBUG_HISTOGRAM_VERIFY) == 0) return; mg_hist = kmem_zalloc(sizeof (uint64_t) * RANGE_TREE_HISTOGRAM_SIZE, KM_SLEEP); ASSERT3U(RANGE_TREE_HISTOGRAM_SIZE, >=, SPACE_MAP_HISTOGRAM_SIZE + ashift); for (int m = 0; m < vd->vdev_ms_count; m++) { metaslab_t *msp = vd->vdev_ms[m]; /* skip if not active or not a member */ if (msp->ms_sm == NULL || msp->ms_group != mg) continue; for (i = 0; i < SPACE_MAP_HISTOGRAM_SIZE; i++) mg_hist[i + ashift] += msp->ms_sm->sm_phys->smp_histogram[i]; } for (i = 0; i < RANGE_TREE_HISTOGRAM_SIZE; i ++) VERIFY3U(mg_hist[i], ==, mg->mg_histogram[i]); kmem_free(mg_hist, sizeof (uint64_t) * RANGE_TREE_HISTOGRAM_SIZE); } static void metaslab_group_histogram_add(metaslab_group_t *mg, metaslab_t *msp) { metaslab_class_t *mc = mg->mg_class; uint64_t ashift = mg->mg_vd->vdev_ashift; ASSERT(MUTEX_HELD(&msp->ms_lock)); if (msp->ms_sm == NULL) return; mutex_enter(&mg->mg_lock); for (int i = 0; i < SPACE_MAP_HISTOGRAM_SIZE; i++) { mg->mg_histogram[i + ashift] += msp->ms_sm->sm_phys->smp_histogram[i]; mc->mc_histogram[i + ashift] += msp->ms_sm->sm_phys->smp_histogram[i]; } mutex_exit(&mg->mg_lock); } void metaslab_group_histogram_remove(metaslab_group_t *mg, metaslab_t *msp) { metaslab_class_t *mc = mg->mg_class; uint64_t ashift = mg->mg_vd->vdev_ashift; ASSERT(MUTEX_HELD(&msp->ms_lock)); if (msp->ms_sm == NULL) return; mutex_enter(&mg->mg_lock); for (int i = 0; i < SPACE_MAP_HISTOGRAM_SIZE; i++) { ASSERT3U(mg->mg_histogram[i + ashift], >=, msp->ms_sm->sm_phys->smp_histogram[i]); ASSERT3U(mc->mc_histogram[i + ashift], >=, msp->ms_sm->sm_phys->smp_histogram[i]); mg->mg_histogram[i + ashift] -= msp->ms_sm->sm_phys->smp_histogram[i]; mc->mc_histogram[i + ashift] -= msp->ms_sm->sm_phys->smp_histogram[i]; } mutex_exit(&mg->mg_lock); } static void metaslab_group_add(metaslab_group_t *mg, metaslab_t *msp) { ASSERT(msp->ms_group == NULL); mutex_enter(&mg->mg_lock); msp->ms_group = mg; msp->ms_weight = 0; avl_add(&mg->mg_metaslab_tree, msp); mutex_exit(&mg->mg_lock); mutex_enter(&msp->ms_lock); metaslab_group_histogram_add(mg, msp); mutex_exit(&msp->ms_lock); } static void metaslab_group_remove(metaslab_group_t *mg, metaslab_t *msp) { mutex_enter(&msp->ms_lock); metaslab_group_histogram_remove(mg, msp); mutex_exit(&msp->ms_lock); mutex_enter(&mg->mg_lock); ASSERT(msp->ms_group == mg); avl_remove(&mg->mg_metaslab_tree, msp); metaslab_class_t *mc = msp->ms_group->mg_class; multilist_sublist_t *mls = multilist_sublist_lock_obj(mc->mc_metaslab_txg_list, msp); if (multilist_link_active(&msp->ms_class_txg_node)) multilist_sublist_remove(mls, msp); multilist_sublist_unlock(mls); msp->ms_group = NULL; mutex_exit(&mg->mg_lock); } static void metaslab_group_sort_impl(metaslab_group_t *mg, metaslab_t *msp, uint64_t weight) { ASSERT(MUTEX_HELD(&msp->ms_lock)); ASSERT(MUTEX_HELD(&mg->mg_lock)); ASSERT(msp->ms_group == mg); avl_remove(&mg->mg_metaslab_tree, msp); msp->ms_weight = weight; avl_add(&mg->mg_metaslab_tree, msp); } static void metaslab_group_sort(metaslab_group_t *mg, metaslab_t *msp, uint64_t weight) { /* * Although in principle the weight can be any value, in * practice we do not use values in the range [1, 511]. */ ASSERT(weight >= SPA_MINBLOCKSIZE || weight == 0); ASSERT(MUTEX_HELD(&msp->ms_lock)); mutex_enter(&mg->mg_lock); metaslab_group_sort_impl(mg, msp, weight); mutex_exit(&mg->mg_lock); } /* * Calculate the fragmentation for a given metaslab group. We can use * a simple average here since all metaslabs within the group must have * the same size. The return value will be a value between 0 and 100 * (inclusive), or ZFS_FRAG_INVALID if less than half of the metaslab in this * group have a fragmentation metric. */ uint64_t metaslab_group_fragmentation(metaslab_group_t *mg) { vdev_t *vd = mg->mg_vd; uint64_t fragmentation = 0; uint64_t valid_ms = 0; for (int m = 0; m < vd->vdev_ms_count; m++) { metaslab_t *msp = vd->vdev_ms[m]; if (msp->ms_fragmentation == ZFS_FRAG_INVALID) continue; if (msp->ms_group != mg) continue; valid_ms++; fragmentation += msp->ms_fragmentation; } if (valid_ms <= mg->mg_vd->vdev_ms_count / 2) return (ZFS_FRAG_INVALID); fragmentation /= valid_ms; ASSERT3U(fragmentation, <=, 100); return (fragmentation); } /* * Determine if a given metaslab group should skip allocations. A metaslab * group should avoid allocations if its free capacity is less than the * zfs_mg_noalloc_threshold or its fragmentation metric is greater than * zfs_mg_fragmentation_threshold and there is at least one metaslab group * that can still handle allocations. If the allocation throttle is enabled * then we skip allocations to devices that have reached their maximum * allocation queue depth unless the selected metaslab group is the only * eligible group remaining. */ static boolean_t metaslab_group_allocatable(metaslab_group_t *mg, metaslab_group_t *rotor, uint64_t psize, int allocator, int d) { spa_t *spa = mg->mg_vd->vdev_spa; metaslab_class_t *mc = mg->mg_class; /* * We can only consider skipping this metaslab group if it's * in the normal metaslab class and there are other metaslab * groups to select from. Otherwise, we always consider it eligible * for allocations. */ if ((mc != spa_normal_class(spa) && mc != spa_special_class(spa) && mc != spa_dedup_class(spa)) || mc->mc_groups <= 1) return (B_TRUE); /* * If the metaslab group's mg_allocatable flag is set (see comments * in metaslab_group_alloc_update() for more information) and * the allocation throttle is disabled then allow allocations to this * device. However, if the allocation throttle is enabled then * check if we have reached our allocation limit (mg_alloc_queue_depth) * to determine if we should allow allocations to this metaslab group. * If all metaslab groups are no longer considered allocatable * (mc_alloc_groups == 0) or we're trying to allocate the smallest * gang block size then we allow allocations on this metaslab group * regardless of the mg_allocatable or throttle settings. */ if (mg->mg_allocatable) { metaslab_group_allocator_t *mga = &mg->mg_allocator[allocator]; int64_t qdepth; uint64_t qmax = mga->mga_cur_max_alloc_queue_depth; if (!mc->mc_alloc_throttle_enabled) return (B_TRUE); /* * If this metaslab group does not have any free space, then * there is no point in looking further. */ if (mg->mg_no_free_space) return (B_FALSE); /* * Relax allocation throttling for ditto blocks. Due to * random imbalances in allocation it tends to push copies * to one vdev, that looks a bit better at the moment. */ qmax = qmax * (4 + d) / 4; qdepth = zfs_refcount_count(&mga->mga_alloc_queue_depth); /* * If this metaslab group is below its qmax or it's * the only allocatable metasable group, then attempt * to allocate from it. */ if (qdepth < qmax || mc->mc_alloc_groups == 1) return (B_TRUE); ASSERT3U(mc->mc_alloc_groups, >, 1); /* * Since this metaslab group is at or over its qmax, we * need to determine if there are metaslab groups after this * one that might be able to handle this allocation. This is * racy since we can't hold the locks for all metaslab * groups at the same time when we make this check. */ for (metaslab_group_t *mgp = mg->mg_next; mgp != rotor; mgp = mgp->mg_next) { metaslab_group_allocator_t *mgap = &mgp->mg_allocator[allocator]; qmax = mgap->mga_cur_max_alloc_queue_depth; qmax = qmax * (4 + d) / 4; qdepth = zfs_refcount_count(&mgap->mga_alloc_queue_depth); /* * If there is another metaslab group that * might be able to handle the allocation, then * we return false so that we skip this group. */ if (qdepth < qmax && !mgp->mg_no_free_space) return (B_FALSE); } /* * We didn't find another group to handle the allocation * so we can't skip this metaslab group even though * we are at or over our qmax. */ return (B_TRUE); } else if (mc->mc_alloc_groups == 0 || psize == SPA_MINBLOCKSIZE) { return (B_TRUE); } return (B_FALSE); } /* * ========================================================================== * Range tree callbacks * ========================================================================== */ /* * Comparison function for the private size-ordered tree using 32-bit * ranges. Tree is sorted by size, larger sizes at the end of the tree. */ static int metaslab_rangesize32_compare(const void *x1, const void *x2) { const range_seg32_t *r1 = x1; const range_seg32_t *r2 = x2; uint64_t rs_size1 = r1->rs_end - r1->rs_start; uint64_t rs_size2 = r2->rs_end - r2->rs_start; int cmp = TREE_CMP(rs_size1, rs_size2); if (likely(cmp)) return (cmp); return (TREE_CMP(r1->rs_start, r2->rs_start)); } /* * Comparison function for the private size-ordered tree using 64-bit * ranges. Tree is sorted by size, larger sizes at the end of the tree. */ static int metaslab_rangesize64_compare(const void *x1, const void *x2) { const range_seg64_t *r1 = x1; const range_seg64_t *r2 = x2; uint64_t rs_size1 = r1->rs_end - r1->rs_start; uint64_t rs_size2 = r2->rs_end - r2->rs_start; int cmp = TREE_CMP(rs_size1, rs_size2); if (likely(cmp)) return (cmp); return (TREE_CMP(r1->rs_start, r2->rs_start)); } typedef struct metaslab_rt_arg { zfs_btree_t *mra_bt; uint32_t mra_floor_shift; } metaslab_rt_arg_t; struct mssa_arg { range_tree_t *rt; metaslab_rt_arg_t *mra; }; static void metaslab_size_sorted_add(void *arg, uint64_t start, uint64_t size) { struct mssa_arg *mssap = arg; range_tree_t *rt = mssap->rt; metaslab_rt_arg_t *mrap = mssap->mra; range_seg_max_t seg = {0}; rs_set_start(&seg, rt, start); rs_set_end(&seg, rt, start + size); metaslab_rt_add(rt, &seg, mrap); } static void metaslab_size_tree_full_load(range_tree_t *rt) { metaslab_rt_arg_t *mrap = rt->rt_arg; #ifdef _METASLAB_TRACING METASLABSTAT_BUMP(metaslabstat_reload_tree); #endif ASSERT0(zfs_btree_numnodes(mrap->mra_bt)); mrap->mra_floor_shift = 0; struct mssa_arg arg = {0}; arg.rt = rt; arg.mra = mrap; range_tree_walk(rt, metaslab_size_sorted_add, &arg); } /* * Create any block allocator specific components. The current allocators * rely on using both a size-ordered range_tree_t and an array of uint64_t's. */ /* ARGSUSED */ static void metaslab_rt_create(range_tree_t *rt, void *arg) { metaslab_rt_arg_t *mrap = arg; zfs_btree_t *size_tree = mrap->mra_bt; size_t size; int (*compare) (const void *, const void *); switch (rt->rt_type) { case RANGE_SEG32: size = sizeof (range_seg32_t); compare = metaslab_rangesize32_compare; break; case RANGE_SEG64: size = sizeof (range_seg64_t); compare = metaslab_rangesize64_compare; break; default: panic("Invalid range seg type %d", rt->rt_type); } zfs_btree_create(size_tree, compare, size); mrap->mra_floor_shift = metaslab_by_size_min_shift; } /* ARGSUSED */ static void metaslab_rt_destroy(range_tree_t *rt, void *arg) { metaslab_rt_arg_t *mrap = arg; zfs_btree_t *size_tree = mrap->mra_bt; zfs_btree_destroy(size_tree); kmem_free(mrap, sizeof (*mrap)); } /* ARGSUSED */ static void metaslab_rt_add(range_tree_t *rt, range_seg_t *rs, void *arg) { metaslab_rt_arg_t *mrap = arg; zfs_btree_t *size_tree = mrap->mra_bt; if (rs_get_end(rs, rt) - rs_get_start(rs, rt) < (1 << mrap->mra_floor_shift)) return; zfs_btree_add(size_tree, rs); } /* ARGSUSED */ static void metaslab_rt_remove(range_tree_t *rt, range_seg_t *rs, void *arg) { metaslab_rt_arg_t *mrap = arg; zfs_btree_t *size_tree = mrap->mra_bt; if (rs_get_end(rs, rt) - rs_get_start(rs, rt) < (1 << mrap->mra_floor_shift)) return; zfs_btree_remove(size_tree, rs); } /* ARGSUSED */ static void metaslab_rt_vacate(range_tree_t *rt, void *arg) { metaslab_rt_arg_t *mrap = arg; zfs_btree_t *size_tree = mrap->mra_bt; zfs_btree_clear(size_tree); zfs_btree_destroy(size_tree); metaslab_rt_create(rt, arg); } static range_tree_ops_t metaslab_rt_ops = { .rtop_create = metaslab_rt_create, .rtop_destroy = metaslab_rt_destroy, .rtop_add = metaslab_rt_add, .rtop_remove = metaslab_rt_remove, .rtop_vacate = metaslab_rt_vacate }; /* * ========================================================================== * Common allocator routines * ========================================================================== */ /* * Return the maximum contiguous segment within the metaslab. */ uint64_t metaslab_largest_allocatable(metaslab_t *msp) { zfs_btree_t *t = &msp->ms_allocatable_by_size; range_seg_t *rs; if (t == NULL) return (0); if (zfs_btree_numnodes(t) == 0) metaslab_size_tree_full_load(msp->ms_allocatable); rs = zfs_btree_last(t, NULL); if (rs == NULL) return (0); return (rs_get_end(rs, msp->ms_allocatable) - rs_get_start(rs, msp->ms_allocatable)); } /* * Return the maximum contiguous segment within the unflushed frees of this * metaslab. */ static uint64_t metaslab_largest_unflushed_free(metaslab_t *msp) { ASSERT(MUTEX_HELD(&msp->ms_lock)); if (msp->ms_unflushed_frees == NULL) return (0); if (zfs_btree_numnodes(&msp->ms_unflushed_frees_by_size) == 0) metaslab_size_tree_full_load(msp->ms_unflushed_frees); range_seg_t *rs = zfs_btree_last(&msp->ms_unflushed_frees_by_size, NULL); if (rs == NULL) return (0); /* * When a range is freed from the metaslab, that range is added to * both the unflushed frees and the deferred frees. While the block * will eventually be usable, if the metaslab were loaded the range * would not be added to the ms_allocatable tree until TXG_DEFER_SIZE * txgs had passed. As a result, when attempting to estimate an upper * bound for the largest currently-usable free segment in the * metaslab, we need to not consider any ranges currently in the defer * trees. This algorithm approximates the largest available chunk in * the largest range in the unflushed_frees tree by taking the first * chunk. While this may be a poor estimate, it should only remain so * briefly and should eventually self-correct as frees are no longer * deferred. Similar logic applies to the ms_freed tree. See * metaslab_load() for more details. * * There are two primary sources of inaccuracy in this estimate. Both * are tolerated for performance reasons. The first source is that we * only check the largest segment for overlaps. Smaller segments may * have more favorable overlaps with the other trees, resulting in * larger usable chunks. Second, we only look at the first chunk in * the largest segment; there may be other usable chunks in the * largest segment, but we ignore them. */ uint64_t rstart = rs_get_start(rs, msp->ms_unflushed_frees); uint64_t rsize = rs_get_end(rs, msp->ms_unflushed_frees) - rstart; for (int t = 0; t < TXG_DEFER_SIZE; t++) { uint64_t start = 0; uint64_t size = 0; boolean_t found = range_tree_find_in(msp->ms_defer[t], rstart, rsize, &start, &size); if (found) { if (rstart == start) return (0); rsize = start - rstart; } } uint64_t start = 0; uint64_t size = 0; boolean_t found = range_tree_find_in(msp->ms_freed, rstart, rsize, &start, &size); if (found) rsize = start - rstart; return (rsize); } static range_seg_t * metaslab_block_find(zfs_btree_t *t, range_tree_t *rt, uint64_t start, uint64_t size, zfs_btree_index_t *where) { range_seg_t *rs; range_seg_max_t rsearch; rs_set_start(&rsearch, rt, start); rs_set_end(&rsearch, rt, start + size); rs = zfs_btree_find(t, &rsearch, where); if (rs == NULL) { rs = zfs_btree_next(t, where, where); } return (rs); } #if defined(WITH_DF_BLOCK_ALLOCATOR) || \ defined(WITH_CF_BLOCK_ALLOCATOR) /* * This is a helper function that can be used by the allocator to find a * suitable block to allocate. This will search the specified B-tree looking * for a block that matches the specified criteria. */ static uint64_t metaslab_block_picker(range_tree_t *rt, uint64_t *cursor, uint64_t size, uint64_t max_search) { if (*cursor == 0) *cursor = rt->rt_start; zfs_btree_t *bt = &rt->rt_root; zfs_btree_index_t where; range_seg_t *rs = metaslab_block_find(bt, rt, *cursor, size, &where); uint64_t first_found; int count_searched = 0; if (rs != NULL) first_found = rs_get_start(rs, rt); while (rs != NULL && (rs_get_start(rs, rt) - first_found <= max_search || count_searched < metaslab_min_search_count)) { uint64_t offset = rs_get_start(rs, rt); if (offset + size <= rs_get_end(rs, rt)) { *cursor = offset + size; return (offset); } rs = zfs_btree_next(bt, &where, &where); count_searched++; } *cursor = 0; return (-1ULL); } #endif /* WITH_DF/CF_BLOCK_ALLOCATOR */ #if defined(WITH_DF_BLOCK_ALLOCATOR) /* * ========================================================================== * Dynamic Fit (df) block allocator * * Search for a free chunk of at least this size, starting from the last * offset (for this alignment of block) looking for up to * metaslab_df_max_search bytes (16MB). If a large enough free chunk is not * found within 16MB, then return a free chunk of exactly the requested size (or * larger). * * If it seems like searching from the last offset will be unproductive, skip * that and just return a free chunk of exactly the requested size (or larger). * This is based on metaslab_df_alloc_threshold and metaslab_df_free_pct. This * mechanism is probably not very useful and may be removed in the future. * * The behavior when not searching can be changed to return the largest free * chunk, instead of a free chunk of exactly the requested size, by setting * metaslab_df_use_largest_segment. * ========================================================================== */ static uint64_t metaslab_df_alloc(metaslab_t *msp, uint64_t size) { /* * Find the largest power of 2 block size that evenly divides the * requested size. This is used to try to allocate blocks with similar * alignment from the same area of the metaslab (i.e. same cursor * bucket) but it does not guarantee that other allocations sizes * may exist in the same region. */ uint64_t align = size & -size; uint64_t *cursor = &msp->ms_lbas[highbit64(align) - 1]; range_tree_t *rt = msp->ms_allocatable; int free_pct = range_tree_space(rt) * 100 / msp->ms_size; uint64_t offset; ASSERT(MUTEX_HELD(&msp->ms_lock)); /* * If we're running low on space, find a segment based on size, * rather than iterating based on offset. */ if (metaslab_largest_allocatable(msp) < metaslab_df_alloc_threshold || free_pct < metaslab_df_free_pct) { offset = -1; } else { offset = metaslab_block_picker(rt, cursor, size, metaslab_df_max_search); } if (offset == -1) { range_seg_t *rs; if (zfs_btree_numnodes(&msp->ms_allocatable_by_size) == 0) metaslab_size_tree_full_load(msp->ms_allocatable); if (metaslab_df_use_largest_segment) { /* use largest free segment */ rs = zfs_btree_last(&msp->ms_allocatable_by_size, NULL); } else { zfs_btree_index_t where; /* use segment of this size, or next largest */ #ifdef _METASLAB_TRACING metaslab_rt_arg_t *mrap = msp->ms_allocatable->rt_arg; if (size < (1 << mrap->mra_floor_shift)) { METASLABSTAT_BUMP( metaslabstat_df_find_under_floor); } #endif rs = metaslab_block_find(&msp->ms_allocatable_by_size, rt, msp->ms_start, size, &where); } if (rs != NULL && rs_get_start(rs, rt) + size <= rs_get_end(rs, rt)) { offset = rs_get_start(rs, rt); *cursor = offset + size; } } return (offset); } static metaslab_ops_t metaslab_df_ops = { metaslab_df_alloc }; metaslab_ops_t *zfs_metaslab_ops = &metaslab_df_ops; #endif /* WITH_DF_BLOCK_ALLOCATOR */ #if defined(WITH_CF_BLOCK_ALLOCATOR) /* * ========================================================================== * Cursor fit block allocator - * Select the largest region in the metaslab, set the cursor to the beginning * of the range and the cursor_end to the end of the range. As allocations * are made advance the cursor. Continue allocating from the cursor until * the range is exhausted and then find a new range. * ========================================================================== */ static uint64_t metaslab_cf_alloc(metaslab_t *msp, uint64_t size) { range_tree_t *rt = msp->ms_allocatable; zfs_btree_t *t = &msp->ms_allocatable_by_size; uint64_t *cursor = &msp->ms_lbas[0]; uint64_t *cursor_end = &msp->ms_lbas[1]; uint64_t offset = 0; ASSERT(MUTEX_HELD(&msp->ms_lock)); ASSERT3U(*cursor_end, >=, *cursor); if ((*cursor + size) > *cursor_end) { range_seg_t *rs; if (zfs_btree_numnodes(t) == 0) metaslab_size_tree_full_load(msp->ms_allocatable); rs = zfs_btree_last(t, NULL); if (rs == NULL || (rs_get_end(rs, rt) - rs_get_start(rs, rt)) < size) return (-1ULL); *cursor = rs_get_start(rs, rt); *cursor_end = rs_get_end(rs, rt); } offset = *cursor; *cursor += size; return (offset); } static metaslab_ops_t metaslab_cf_ops = { metaslab_cf_alloc }; metaslab_ops_t *zfs_metaslab_ops = &metaslab_cf_ops; #endif /* WITH_CF_BLOCK_ALLOCATOR */ #if defined(WITH_NDF_BLOCK_ALLOCATOR) /* * ========================================================================== * New dynamic fit allocator - * Select a region that is large enough to allocate 2^metaslab_ndf_clump_shift * contiguous blocks. If no region is found then just use the largest segment * that remains. * ========================================================================== */ /* * Determines desired number of contiguous blocks (2^metaslab_ndf_clump_shift) * to request from the allocator. */ uint64_t metaslab_ndf_clump_shift = 4; static uint64_t metaslab_ndf_alloc(metaslab_t *msp, uint64_t size) { zfs_btree_t *t = &msp->ms_allocatable->rt_root; range_tree_t *rt = msp->ms_allocatable; zfs_btree_index_t where; range_seg_t *rs; range_seg_max_t rsearch; uint64_t hbit = highbit64(size); uint64_t *cursor = &msp->ms_lbas[hbit - 1]; uint64_t max_size = metaslab_largest_allocatable(msp); ASSERT(MUTEX_HELD(&msp->ms_lock)); if (max_size < size) return (-1ULL); rs_set_start(&rsearch, rt, *cursor); rs_set_end(&rsearch, rt, *cursor + size); rs = zfs_btree_find(t, &rsearch, &where); if (rs == NULL || (rs_get_end(rs, rt) - rs_get_start(rs, rt)) < size) { t = &msp->ms_allocatable_by_size; rs_set_start(&rsearch, rt, 0); rs_set_end(&rsearch, rt, MIN(max_size, 1ULL << (hbit + metaslab_ndf_clump_shift))); rs = zfs_btree_find(t, &rsearch, &where); if (rs == NULL) rs = zfs_btree_next(t, &where, &where); ASSERT(rs != NULL); } if ((rs_get_end(rs, rt) - rs_get_start(rs, rt)) >= size) { *cursor = rs_get_start(rs, rt) + size; return (rs_get_start(rs, rt)); } return (-1ULL); } static metaslab_ops_t metaslab_ndf_ops = { metaslab_ndf_alloc }; metaslab_ops_t *zfs_metaslab_ops = &metaslab_ndf_ops; #endif /* WITH_NDF_BLOCK_ALLOCATOR */ /* * ========================================================================== * Metaslabs * ========================================================================== */ /* * Wait for any in-progress metaslab loads to complete. */ static void metaslab_load_wait(metaslab_t *msp) { ASSERT(MUTEX_HELD(&msp->ms_lock)); while (msp->ms_loading) { ASSERT(!msp->ms_loaded); cv_wait(&msp->ms_load_cv, &msp->ms_lock); } } /* * Wait for any in-progress flushing to complete. */ static void metaslab_flush_wait(metaslab_t *msp) { ASSERT(MUTEX_HELD(&msp->ms_lock)); while (msp->ms_flushing) cv_wait(&msp->ms_flush_cv, &msp->ms_lock); } static unsigned int metaslab_idx_func(multilist_t *ml, void *arg) { metaslab_t *msp = arg; return (msp->ms_id % multilist_get_num_sublists(ml)); } uint64_t metaslab_allocated_space(metaslab_t *msp) { return (msp->ms_allocated_space); } /* * Verify that the space accounting on disk matches the in-core range_trees. */ static void metaslab_verify_space(metaslab_t *msp, uint64_t txg) { spa_t *spa = msp->ms_group->mg_vd->vdev_spa; uint64_t allocating = 0; uint64_t sm_free_space, msp_free_space; ASSERT(MUTEX_HELD(&msp->ms_lock)); ASSERT(!msp->ms_condensing); if ((zfs_flags & ZFS_DEBUG_METASLAB_VERIFY) == 0) return; /* * We can only verify the metaslab space when we're called * from syncing context with a loaded metaslab that has an * allocated space map. Calling this in non-syncing context * does not provide a consistent view of the metaslab since * we're performing allocations in the future. */ if (txg != spa_syncing_txg(spa) || msp->ms_sm == NULL || !msp->ms_loaded) return; /* * Even though the smp_alloc field can get negative, * when it comes to a metaslab's space map, that should * never be the case. */ ASSERT3S(space_map_allocated(msp->ms_sm), >=, 0); ASSERT3U(space_map_allocated(msp->ms_sm), >=, range_tree_space(msp->ms_unflushed_frees)); ASSERT3U(metaslab_allocated_space(msp), ==, space_map_allocated(msp->ms_sm) + range_tree_space(msp->ms_unflushed_allocs) - range_tree_space(msp->ms_unflushed_frees)); sm_free_space = msp->ms_size - metaslab_allocated_space(msp); /* * Account for future allocations since we would have * already deducted that space from the ms_allocatable. */ for (int t = 0; t < TXG_CONCURRENT_STATES; t++) { allocating += range_tree_space(msp->ms_allocating[(txg + t) & TXG_MASK]); } ASSERT3U(allocating + msp->ms_allocated_this_txg, ==, msp->ms_allocating_total); ASSERT3U(msp->ms_deferspace, ==, range_tree_space(msp->ms_defer[0]) + range_tree_space(msp->ms_defer[1])); msp_free_space = range_tree_space(msp->ms_allocatable) + allocating + msp->ms_deferspace + range_tree_space(msp->ms_freed); VERIFY3U(sm_free_space, ==, msp_free_space); } static void metaslab_aux_histograms_clear(metaslab_t *msp) { /* * Auxiliary histograms are only cleared when resetting them, * which can only happen while the metaslab is loaded. */ ASSERT(msp->ms_loaded); bzero(msp->ms_synchist, sizeof (msp->ms_synchist)); for (int t = 0; t < TXG_DEFER_SIZE; t++) bzero(msp->ms_deferhist[t], sizeof (msp->ms_deferhist[t])); } static void metaslab_aux_histogram_add(uint64_t *histogram, uint64_t shift, range_tree_t *rt) { /* * This is modeled after space_map_histogram_add(), so refer to that * function for implementation details. We want this to work like * the space map histogram, and not the range tree histogram, as we * are essentially constructing a delta that will be later subtracted * from the space map histogram. */ int idx = 0; for (int i = shift; i < RANGE_TREE_HISTOGRAM_SIZE; i++) { ASSERT3U(i, >=, idx + shift); histogram[idx] += rt->rt_histogram[i] << (i - idx - shift); if (idx < SPACE_MAP_HISTOGRAM_SIZE - 1) { ASSERT3U(idx + shift, ==, i); idx++; ASSERT3U(idx, <, SPACE_MAP_HISTOGRAM_SIZE); } } } /* * Called at every sync pass that the metaslab gets synced. * * The reason is that we want our auxiliary histograms to be updated * wherever the metaslab's space map histogram is updated. This way * we stay consistent on which parts of the metaslab space map's * histogram are currently not available for allocations (e.g because * they are in the defer, freed, and freeing trees). */ static void metaslab_aux_histograms_update(metaslab_t *msp) { space_map_t *sm = msp->ms_sm; ASSERT(sm != NULL); /* * This is similar to the metaslab's space map histogram updates * that take place in metaslab_sync(). The only difference is that * we only care about segments that haven't made it into the * ms_allocatable tree yet. */ if (msp->ms_loaded) { metaslab_aux_histograms_clear(msp); metaslab_aux_histogram_add(msp->ms_synchist, sm->sm_shift, msp->ms_freed); for (int t = 0; t < TXG_DEFER_SIZE; t++) { metaslab_aux_histogram_add(msp->ms_deferhist[t], sm->sm_shift, msp->ms_defer[t]); } } metaslab_aux_histogram_add(msp->ms_synchist, sm->sm_shift, msp->ms_freeing); } /* * Called every time we are done syncing (writing to) the metaslab, * i.e. at the end of each sync pass. * [see the comment in metaslab_impl.h for ms_synchist, ms_deferhist] */ static void metaslab_aux_histograms_update_done(metaslab_t *msp, boolean_t defer_allowed) { spa_t *spa = msp->ms_group->mg_vd->vdev_spa; space_map_t *sm = msp->ms_sm; if (sm == NULL) { /* * We came here from metaslab_init() when creating/opening a * pool, looking at a metaslab that hasn't had any allocations * yet. */ return; } /* * This is similar to the actions that we take for the ms_freed * and ms_defer trees in metaslab_sync_done(). */ uint64_t hist_index = spa_syncing_txg(spa) % TXG_DEFER_SIZE; if (defer_allowed) { bcopy(msp->ms_synchist, msp->ms_deferhist[hist_index], sizeof (msp->ms_synchist)); } else { bzero(msp->ms_deferhist[hist_index], sizeof (msp->ms_deferhist[hist_index])); } bzero(msp->ms_synchist, sizeof (msp->ms_synchist)); } /* * Ensure that the metaslab's weight and fragmentation are consistent * with the contents of the histogram (either the range tree's histogram * or the space map's depending whether the metaslab is loaded). */ static void metaslab_verify_weight_and_frag(metaslab_t *msp) { ASSERT(MUTEX_HELD(&msp->ms_lock)); if ((zfs_flags & ZFS_DEBUG_METASLAB_VERIFY) == 0) return; /* * We can end up here from vdev_remove_complete(), in which case we * cannot do these assertions because we hold spa config locks and * thus we are not allowed to read from the DMU. * * We check if the metaslab group has been removed and if that's * the case we return immediately as that would mean that we are * here from the aforementioned code path. */ if (msp->ms_group == NULL) return; /* * Devices being removed always return a weight of 0 and leave * fragmentation and ms_max_size as is - there is nothing for * us to verify here. */ vdev_t *vd = msp->ms_group->mg_vd; if (vd->vdev_removing) return; /* * If the metaslab is dirty it probably means that we've done * some allocations or frees that have changed our histograms * and thus the weight. */ for (int t = 0; t < TXG_SIZE; t++) { if (txg_list_member(&vd->vdev_ms_list, msp, t)) return; } /* * This verification checks that our in-memory state is consistent * with what's on disk. If the pool is read-only then there aren't * any changes and we just have the initially-loaded state. */ if (!spa_writeable(msp->ms_group->mg_vd->vdev_spa)) return; /* some extra verification for in-core tree if you can */ if (msp->ms_loaded) { range_tree_stat_verify(msp->ms_allocatable); VERIFY(space_map_histogram_verify(msp->ms_sm, msp->ms_allocatable)); } uint64_t weight = msp->ms_weight; uint64_t was_active = msp->ms_weight & METASLAB_ACTIVE_MASK; boolean_t space_based = WEIGHT_IS_SPACEBASED(msp->ms_weight); uint64_t frag = msp->ms_fragmentation; uint64_t max_segsize = msp->ms_max_size; msp->ms_weight = 0; msp->ms_fragmentation = 0; /* * This function is used for verification purposes and thus should * not introduce any side-effects/mutations on the system's state. * * Regardless of whether metaslab_weight() thinks this metaslab * should be active or not, we want to ensure that the actual weight * (and therefore the value of ms_weight) would be the same if it * was to be recalculated at this point. * * In addition we set the nodirty flag so metaslab_weight() does * not dirty the metaslab for future TXGs (e.g. when trying to * force condensing to upgrade the metaslab spacemaps). */ msp->ms_weight = metaslab_weight(msp, B_TRUE) | was_active; VERIFY3U(max_segsize, ==, msp->ms_max_size); /* * If the weight type changed then there is no point in doing * verification. Revert fields to their original values. */ if ((space_based && !WEIGHT_IS_SPACEBASED(msp->ms_weight)) || (!space_based && WEIGHT_IS_SPACEBASED(msp->ms_weight))) { msp->ms_fragmentation = frag; msp->ms_weight = weight; return; } VERIFY3U(msp->ms_fragmentation, ==, frag); VERIFY3U(msp->ms_weight, ==, weight); } /* * If we're over the zfs_metaslab_mem_limit, select the loaded metaslab from * this class that was used longest ago, and attempt to unload it. We don't * want to spend too much time in this loop to prevent performance * degradation, and we expect that most of the time this operation will * succeed. Between that and the normal unloading processing during txg sync, * we expect this to keep the metaslab memory usage under control. */ static void metaslab_potentially_evict(metaslab_class_t *mc) { #ifdef _KERNEL uint64_t allmem = arc_all_memory(); uint64_t inuse = spl_kmem_cache_inuse(zfs_btree_leaf_cache); uint64_t size = spl_kmem_cache_entry_size(zfs_btree_leaf_cache); int tries = 0; for (; allmem * zfs_metaslab_mem_limit / 100 < inuse * size && tries < multilist_get_num_sublists(mc->mc_metaslab_txg_list) * 2; tries++) { unsigned int idx = multilist_get_random_index( mc->mc_metaslab_txg_list); multilist_sublist_t *mls = multilist_sublist_lock(mc->mc_metaslab_txg_list, idx); metaslab_t *msp = multilist_sublist_head(mls); multilist_sublist_unlock(mls); while (msp != NULL && allmem * zfs_metaslab_mem_limit / 100 < inuse * size) { VERIFY3P(mls, ==, multilist_sublist_lock( mc->mc_metaslab_txg_list, idx)); ASSERT3U(idx, ==, metaslab_idx_func(mc->mc_metaslab_txg_list, msp)); if (!multilist_link_active(&msp->ms_class_txg_node)) { multilist_sublist_unlock(mls); break; } metaslab_t *next_msp = multilist_sublist_next(mls, msp); multilist_sublist_unlock(mls); /* * If the metaslab is currently loading there are two * cases. If it's the metaslab we're evicting, we * can't continue on or we'll panic when we attempt to * recursively lock the mutex. If it's another * metaslab that's loading, it can be safely skipped, * since we know it's very new and therefore not a * good eviction candidate. We check later once the * lock is held that the metaslab is fully loaded * before actually unloading it. */ if (msp->ms_loading) { msp = next_msp; inuse = spl_kmem_cache_inuse(zfs_btree_leaf_cache); continue; } /* * We can't unload metaslabs with no spacemap because * they're not ready to be unloaded yet. We can't * unload metaslabs with outstanding allocations * because doing so could cause the metaslab's weight * to decrease while it's unloaded, which violates an * invariant that we use to prevent unnecessary * loading. We also don't unload metaslabs that are * currently active because they are high-weight * metaslabs that are likely to be used in the near * future. */ mutex_enter(&msp->ms_lock); if (msp->ms_allocator == -1 && msp->ms_sm != NULL && msp->ms_allocating_total == 0) { metaslab_unload(msp); } mutex_exit(&msp->ms_lock); msp = next_msp; inuse = spl_kmem_cache_inuse(zfs_btree_leaf_cache); } } #endif } static int metaslab_load_impl(metaslab_t *msp) { int error = 0; ASSERT(MUTEX_HELD(&msp->ms_lock)); ASSERT(msp->ms_loading); ASSERT(!msp->ms_condensing); /* * We temporarily drop the lock to unblock other operations while we * are reading the space map. Therefore, metaslab_sync() and * metaslab_sync_done() can run at the same time as we do. * * If we are using the log space maps, metaslab_sync() can't write to * the metaslab's space map while we are loading as we only write to * it when we are flushing the metaslab, and that can't happen while * we are loading it. * * If we are not using log space maps though, metaslab_sync() can * append to the space map while we are loading. Therefore we load * only entries that existed when we started the load. Additionally, * metaslab_sync_done() has to wait for the load to complete because * there are potential races like metaslab_load() loading parts of the * space map that are currently being appended by metaslab_sync(). If * we didn't, the ms_allocatable would have entries that * metaslab_sync_done() would try to re-add later. * * That's why before dropping the lock we remember the synced length * of the metaslab and read up to that point of the space map, * ignoring entries appended by metaslab_sync() that happen after we * drop the lock. */ uint64_t length = msp->ms_synced_length; mutex_exit(&msp->ms_lock); hrtime_t load_start = gethrtime(); metaslab_rt_arg_t *mrap; if (msp->ms_allocatable->rt_arg == NULL) { mrap = kmem_zalloc(sizeof (*mrap), KM_SLEEP); } else { mrap = msp->ms_allocatable->rt_arg; msp->ms_allocatable->rt_ops = NULL; msp->ms_allocatable->rt_arg = NULL; } mrap->mra_bt = &msp->ms_allocatable_by_size; mrap->mra_floor_shift = metaslab_by_size_min_shift; if (msp->ms_sm != NULL) { error = space_map_load_length(msp->ms_sm, msp->ms_allocatable, SM_FREE, length); /* Now, populate the size-sorted tree. */ metaslab_rt_create(msp->ms_allocatable, mrap); msp->ms_allocatable->rt_ops = &metaslab_rt_ops; msp->ms_allocatable->rt_arg = mrap; struct mssa_arg arg = {0}; arg.rt = msp->ms_allocatable; arg.mra = mrap; range_tree_walk(msp->ms_allocatable, metaslab_size_sorted_add, &arg); } else { /* * Add the size-sorted tree first, since we don't need to load * the metaslab from the spacemap. */ metaslab_rt_create(msp->ms_allocatable, mrap); msp->ms_allocatable->rt_ops = &metaslab_rt_ops; msp->ms_allocatable->rt_arg = mrap; /* * The space map has not been allocated yet, so treat * all the space in the metaslab as free and add it to the * ms_allocatable tree. */ range_tree_add(msp->ms_allocatable, msp->ms_start, msp->ms_size); if (msp->ms_freed != NULL) { /* * If the ms_sm doesn't exist, this means that this * metaslab hasn't gone through metaslab_sync() and * thus has never been dirtied. So we shouldn't * expect any unflushed allocs or frees from previous * TXGs. * * Note: ms_freed and all the other trees except for * the ms_allocatable, can be NULL at this point only * if this is a new metaslab of a vdev that just got * expanded. */ ASSERT(range_tree_is_empty(msp->ms_unflushed_allocs)); ASSERT(range_tree_is_empty(msp->ms_unflushed_frees)); } } /* * We need to grab the ms_sync_lock to prevent metaslab_sync() from * changing the ms_sm (or log_sm) and the metaslab's range trees * while we are about to use them and populate the ms_allocatable. * The ms_lock is insufficient for this because metaslab_sync() doesn't * hold the ms_lock while writing the ms_checkpointing tree to disk. */ mutex_enter(&msp->ms_sync_lock); mutex_enter(&msp->ms_lock); ASSERT(!msp->ms_condensing); ASSERT(!msp->ms_flushing); if (error != 0) { mutex_exit(&msp->ms_sync_lock); return (error); } ASSERT3P(msp->ms_group, !=, NULL); msp->ms_loaded = B_TRUE; /* * Apply all the unflushed changes to ms_allocatable right * away so any manipulations we do below have a clear view * of what is allocated and what is free. */ range_tree_walk(msp->ms_unflushed_allocs, range_tree_remove, msp->ms_allocatable); range_tree_walk(msp->ms_unflushed_frees, range_tree_add, msp->ms_allocatable); msp->ms_loaded = B_TRUE; ASSERT3P(msp->ms_group, !=, NULL); spa_t *spa = msp->ms_group->mg_vd->vdev_spa; if (spa_syncing_log_sm(spa) != NULL) { ASSERT(spa_feature_is_enabled(spa, SPA_FEATURE_LOG_SPACEMAP)); /* * If we use a log space map we add all the segments * that are in ms_unflushed_frees so they are available * for allocation. * * ms_allocatable needs to contain all free segments * that are ready for allocations (thus not segments * from ms_freeing, ms_freed, and the ms_defer trees). * But if we grab the lock in this code path at a sync * pass later that 1, then it also contains the * segments of ms_freed (they were added to it earlier * in this path through ms_unflushed_frees). So we * need to remove all the segments that exist in * ms_freed from ms_allocatable as they will be added * later in metaslab_sync_done(). * * When there's no log space map, the ms_allocatable * correctly doesn't contain any segments that exist * in ms_freed [see ms_synced_length]. */ range_tree_walk(msp->ms_freed, range_tree_remove, msp->ms_allocatable); } /* * If we are not using the log space map, ms_allocatable * contains the segments that exist in the ms_defer trees * [see ms_synced_length]. Thus we need to remove them * from ms_allocatable as they will be added again in * metaslab_sync_done(). * * If we are using the log space map, ms_allocatable still * contains the segments that exist in the ms_defer trees. * Not because it read them through the ms_sm though. But * because these segments are part of ms_unflushed_frees * whose segments we add to ms_allocatable earlier in this * code path. */ for (int t = 0; t < TXG_DEFER_SIZE; t++) { range_tree_walk(msp->ms_defer[t], range_tree_remove, msp->ms_allocatable); } /* * Call metaslab_recalculate_weight_and_sort() now that the * metaslab is loaded so we get the metaslab's real weight. * * Unless this metaslab was created with older software and * has not yet been converted to use segment-based weight, we * expect the new weight to be better or equal to the weight * that the metaslab had while it was not loaded. This is * because the old weight does not take into account the * consolidation of adjacent segments between TXGs. [see * comment for ms_synchist and ms_deferhist[] for more info] */ uint64_t weight = msp->ms_weight; uint64_t max_size = msp->ms_max_size; metaslab_recalculate_weight_and_sort(msp); if (!WEIGHT_IS_SPACEBASED(weight)) ASSERT3U(weight, <=, msp->ms_weight); msp->ms_max_size = metaslab_largest_allocatable(msp); ASSERT3U(max_size, <=, msp->ms_max_size); hrtime_t load_end = gethrtime(); msp->ms_load_time = load_end; zfs_dbgmsg("metaslab_load: txg %llu, spa %s, vdev_id %llu, " "ms_id %llu, smp_length %llu, " "unflushed_allocs %llu, unflushed_frees %llu, " "freed %llu, defer %llu + %llu, unloaded time %llu ms, " "loading_time %lld ms, ms_max_size %llu, " "max size error %lld, " "old_weight %llx, new_weight %llx", spa_syncing_txg(spa), spa_name(spa), msp->ms_group->mg_vd->vdev_id, msp->ms_id, space_map_length(msp->ms_sm), range_tree_space(msp->ms_unflushed_allocs), range_tree_space(msp->ms_unflushed_frees), range_tree_space(msp->ms_freed), range_tree_space(msp->ms_defer[0]), range_tree_space(msp->ms_defer[1]), (longlong_t)((load_start - msp->ms_unload_time) / 1000000), (longlong_t)((load_end - load_start) / 1000000), msp->ms_max_size, msp->ms_max_size - max_size, weight, msp->ms_weight); metaslab_verify_space(msp, spa_syncing_txg(spa)); mutex_exit(&msp->ms_sync_lock); return (0); } int metaslab_load(metaslab_t *msp) { ASSERT(MUTEX_HELD(&msp->ms_lock)); /* * There may be another thread loading the same metaslab, if that's * the case just wait until the other thread is done and return. */ metaslab_load_wait(msp); if (msp->ms_loaded) return (0); VERIFY(!msp->ms_loading); ASSERT(!msp->ms_condensing); /* * We set the loading flag BEFORE potentially dropping the lock to * wait for an ongoing flush (see ms_flushing below). This way other * threads know that there is already a thread that is loading this * metaslab. */ msp->ms_loading = B_TRUE; /* * Wait for any in-progress flushing to finish as we drop the ms_lock * both here (during space_map_load()) and in metaslab_flush() (when * we flush our changes to the ms_sm). */ if (msp->ms_flushing) metaslab_flush_wait(msp); /* * In the possibility that we were waiting for the metaslab to be * flushed (where we temporarily dropped the ms_lock), ensure that * no one else loaded the metaslab somehow. */ ASSERT(!msp->ms_loaded); /* * If we're loading a metaslab in the normal class, consider evicting * another one to keep our memory usage under the limit defined by the * zfs_metaslab_mem_limit tunable. */ if (spa_normal_class(msp->ms_group->mg_class->mc_spa) == msp->ms_group->mg_class) { metaslab_potentially_evict(msp->ms_group->mg_class); } int error = metaslab_load_impl(msp); ASSERT(MUTEX_HELD(&msp->ms_lock)); msp->ms_loading = B_FALSE; cv_broadcast(&msp->ms_load_cv); return (error); } void metaslab_unload(metaslab_t *msp) { ASSERT(MUTEX_HELD(&msp->ms_lock)); /* * This can happen if a metaslab is selected for eviction (in * metaslab_potentially_evict) and then unloaded during spa_sync (via * metaslab_class_evict_old). */ if (!msp->ms_loaded) return; range_tree_vacate(msp->ms_allocatable, NULL, NULL); msp->ms_loaded = B_FALSE; msp->ms_unload_time = gethrtime(); msp->ms_activation_weight = 0; msp->ms_weight &= ~METASLAB_ACTIVE_MASK; if (msp->ms_group != NULL) { metaslab_class_t *mc = msp->ms_group->mg_class; multilist_sublist_t *mls = multilist_sublist_lock_obj(mc->mc_metaslab_txg_list, msp); if (multilist_link_active(&msp->ms_class_txg_node)) multilist_sublist_remove(mls, msp); multilist_sublist_unlock(mls); spa_t *spa = msp->ms_group->mg_vd->vdev_spa; zfs_dbgmsg("metaslab_unload: txg %llu, spa %s, vdev_id %llu, " "ms_id %llu, weight %llx, " "selected txg %llu (%llu ms ago), alloc_txg %llu, " "loaded %llu ms ago, max_size %llu", spa_syncing_txg(spa), spa_name(spa), msp->ms_group->mg_vd->vdev_id, msp->ms_id, msp->ms_weight, msp->ms_selected_txg, (msp->ms_unload_time - msp->ms_selected_time) / 1000 / 1000, msp->ms_alloc_txg, (msp->ms_unload_time - msp->ms_load_time) / 1000 / 1000, msp->ms_max_size); } /* * We explicitly recalculate the metaslab's weight based on its space * map (as it is now not loaded). We want unload metaslabs to always * have their weights calculated from the space map histograms, while * loaded ones have it calculated from their in-core range tree * [see metaslab_load()]. This way, the weight reflects the information * available in-core, whether it is loaded or not. * * If ms_group == NULL means that we came here from metaslab_fini(), * at which point it doesn't make sense for us to do the recalculation * and the sorting. */ if (msp->ms_group != NULL) metaslab_recalculate_weight_and_sort(msp); } /* * We want to optimize the memory use of the per-metaslab range * trees. To do this, we store the segments in the range trees in * units of sectors, zero-indexing from the start of the metaslab. If * the vdev_ms_shift - the vdev_ashift is less than 32, we can store * the ranges using two uint32_ts, rather than two uint64_ts. */ range_seg_type_t metaslab_calculate_range_tree_type(vdev_t *vdev, metaslab_t *msp, uint64_t *start, uint64_t *shift) { if (vdev->vdev_ms_shift - vdev->vdev_ashift < 32 && !zfs_metaslab_force_large_segs) { *shift = vdev->vdev_ashift; *start = msp->ms_start; return (RANGE_SEG32); } else { *shift = 0; *start = 0; return (RANGE_SEG64); } } void metaslab_set_selected_txg(metaslab_t *msp, uint64_t txg) { ASSERT(MUTEX_HELD(&msp->ms_lock)); metaslab_class_t *mc = msp->ms_group->mg_class; multilist_sublist_t *mls = multilist_sublist_lock_obj(mc->mc_metaslab_txg_list, msp); if (multilist_link_active(&msp->ms_class_txg_node)) multilist_sublist_remove(mls, msp); msp->ms_selected_txg = txg; msp->ms_selected_time = gethrtime(); multilist_sublist_insert_tail(mls, msp); multilist_sublist_unlock(mls); } void metaslab_space_update(vdev_t *vd, metaslab_class_t *mc, int64_t alloc_delta, int64_t defer_delta, int64_t space_delta) { vdev_space_update(vd, alloc_delta, defer_delta, space_delta); ASSERT3P(vd->vdev_spa->spa_root_vdev, ==, vd->vdev_parent); ASSERT(vd->vdev_ms_count != 0); metaslab_class_space_update(mc, alloc_delta, defer_delta, space_delta, vdev_deflated_space(vd, space_delta)); } int metaslab_init(metaslab_group_t *mg, uint64_t id, uint64_t object, uint64_t txg, metaslab_t **msp) { vdev_t *vd = mg->mg_vd; spa_t *spa = vd->vdev_spa; objset_t *mos = spa->spa_meta_objset; metaslab_t *ms; int error; ms = kmem_zalloc(sizeof (metaslab_t), KM_SLEEP); mutex_init(&ms->ms_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&ms->ms_sync_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&ms->ms_load_cv, NULL, CV_DEFAULT, NULL); cv_init(&ms->ms_flush_cv, NULL, CV_DEFAULT, NULL); multilist_link_init(&ms->ms_class_txg_node); ms->ms_id = id; ms->ms_start = id << vd->vdev_ms_shift; ms->ms_size = 1ULL << vd->vdev_ms_shift; ms->ms_allocator = -1; ms->ms_new = B_TRUE; /* * We only open space map objects that already exist. All others * will be opened when we finally allocate an object for it. * * Note: * When called from vdev_expand(), we can't call into the DMU as * we are holding the spa_config_lock as a writer and we would * deadlock [see relevant comment in vdev_metaslab_init()]. in * that case, the object parameter is zero though, so we won't * call into the DMU. */ if (object != 0) { error = space_map_open(&ms->ms_sm, mos, object, ms->ms_start, ms->ms_size, vd->vdev_ashift); if (error != 0) { kmem_free(ms, sizeof (metaslab_t)); return (error); } ASSERT(ms->ms_sm != NULL); ms->ms_allocated_space = space_map_allocated(ms->ms_sm); } range_seg_type_t type; uint64_t shift, start; type = metaslab_calculate_range_tree_type(vd, ms, &start, &shift); /* * We create the ms_allocatable here, but we don't create the * other range trees until metaslab_sync_done(). This serves * two purposes: it allows metaslab_sync_done() to detect the * addition of new space; and for debugging, it ensures that * we'd data fault on any attempt to use this metaslab before * it's ready. */ ms->ms_allocatable = range_tree_create(NULL, type, NULL, start, shift); ms->ms_trim = range_tree_create(NULL, type, NULL, start, shift); metaslab_group_add(mg, ms); metaslab_set_fragmentation(ms, B_FALSE); /* * If we're opening an existing pool (txg == 0) or creating * a new one (txg == TXG_INITIAL), all space is available now. * If we're adding space to an existing pool, the new space * does not become available until after this txg has synced. * The metaslab's weight will also be initialized when we sync * out this txg. This ensures that we don't attempt to allocate * from it before we have initialized it completely. */ if (txg <= TXG_INITIAL) { metaslab_sync_done(ms, 0); metaslab_space_update(vd, mg->mg_class, metaslab_allocated_space(ms), 0, 0); } if (txg != 0) { vdev_dirty(vd, 0, NULL, txg); vdev_dirty(vd, VDD_METASLAB, ms, txg); } *msp = ms; return (0); } static void metaslab_fini_flush_data(metaslab_t *msp) { spa_t *spa = msp->ms_group->mg_vd->vdev_spa; if (metaslab_unflushed_txg(msp) == 0) { ASSERT3P(avl_find(&spa->spa_metaslabs_by_flushed, msp, NULL), ==, NULL); return; } ASSERT(spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP)); mutex_enter(&spa->spa_flushed_ms_lock); avl_remove(&spa->spa_metaslabs_by_flushed, msp); mutex_exit(&spa->spa_flushed_ms_lock); spa_log_sm_decrement_mscount(spa, metaslab_unflushed_txg(msp)); spa_log_summary_decrement_mscount(spa, metaslab_unflushed_txg(msp)); } uint64_t metaslab_unflushed_changes_memused(metaslab_t *ms) { return ((range_tree_numsegs(ms->ms_unflushed_allocs) + range_tree_numsegs(ms->ms_unflushed_frees)) * ms->ms_unflushed_allocs->rt_root.bt_elem_size); } void metaslab_fini(metaslab_t *msp) { metaslab_group_t *mg = msp->ms_group; vdev_t *vd = mg->mg_vd; spa_t *spa = vd->vdev_spa; metaslab_fini_flush_data(msp); metaslab_group_remove(mg, msp); mutex_enter(&msp->ms_lock); VERIFY(msp->ms_group == NULL); metaslab_space_update(vd, mg->mg_class, -metaslab_allocated_space(msp), 0, -msp->ms_size); space_map_close(msp->ms_sm); msp->ms_sm = NULL; metaslab_unload(msp); range_tree_destroy(msp->ms_allocatable); range_tree_destroy(msp->ms_freeing); range_tree_destroy(msp->ms_freed); ASSERT3U(spa->spa_unflushed_stats.sus_memused, >=, metaslab_unflushed_changes_memused(msp)); spa->spa_unflushed_stats.sus_memused -= metaslab_unflushed_changes_memused(msp); range_tree_vacate(msp->ms_unflushed_allocs, NULL, NULL); range_tree_destroy(msp->ms_unflushed_allocs); range_tree_vacate(msp->ms_unflushed_frees, NULL, NULL); range_tree_destroy(msp->ms_unflushed_frees); for (int t = 0; t < TXG_SIZE; t++) { range_tree_destroy(msp->ms_allocating[t]); } for (int t = 0; t < TXG_DEFER_SIZE; t++) { range_tree_destroy(msp->ms_defer[t]); } ASSERT0(msp->ms_deferspace); range_tree_destroy(msp->ms_checkpointing); for (int t = 0; t < TXG_SIZE; t++) ASSERT(!txg_list_member(&vd->vdev_ms_list, msp, t)); range_tree_vacate(msp->ms_trim, NULL, NULL); range_tree_destroy(msp->ms_trim); mutex_exit(&msp->ms_lock); cv_destroy(&msp->ms_load_cv); cv_destroy(&msp->ms_flush_cv); mutex_destroy(&msp->ms_lock); mutex_destroy(&msp->ms_sync_lock); ASSERT3U(msp->ms_allocator, ==, -1); kmem_free(msp, sizeof (metaslab_t)); } #define FRAGMENTATION_TABLE_SIZE 17 /* * This table defines a segment size based fragmentation metric that will * allow each metaslab to derive its own fragmentation value. This is done * by calculating the space in each bucket of the spacemap histogram and * multiplying that by the fragmentation metric in this table. Doing * this for all buckets and dividing it by the total amount of free * space in this metaslab (i.e. the total free space in all buckets) gives * us the fragmentation metric. This means that a high fragmentation metric * equates to most of the free space being comprised of small segments. * Conversely, if the metric is low, then most of the free space is in * large segments. A 10% change in fragmentation equates to approximately * double the number of segments. * * This table defines 0% fragmented space using 16MB segments. Testing has * shown that segments that are greater than or equal to 16MB do not suffer * from drastic performance problems. Using this value, we derive the rest * of the table. Since the fragmentation value is never stored on disk, it * is possible to change these calculations in the future. */ int zfs_frag_table[FRAGMENTATION_TABLE_SIZE] = { 100, /* 512B */ 100, /* 1K */ 98, /* 2K */ 95, /* 4K */ 90, /* 8K */ 80, /* 16K */ 70, /* 32K */ 60, /* 64K */ 50, /* 128K */ 40, /* 256K */ 30, /* 512K */ 20, /* 1M */ 15, /* 2M */ 10, /* 4M */ 5, /* 8M */ 0 /* 16M */ }; /* * Calculate the metaslab's fragmentation metric and set ms_fragmentation. * Setting this value to ZFS_FRAG_INVALID means that the metaslab has not * been upgraded and does not support this metric. Otherwise, the return * value should be in the range [0, 100]. */ static void metaslab_set_fragmentation(metaslab_t *msp, boolean_t nodirty) { spa_t *spa = msp->ms_group->mg_vd->vdev_spa; uint64_t fragmentation = 0; uint64_t total = 0; boolean_t feature_enabled = spa_feature_is_enabled(spa, SPA_FEATURE_SPACEMAP_HISTOGRAM); if (!feature_enabled) { msp->ms_fragmentation = ZFS_FRAG_INVALID; return; } /* * A null space map means that the entire metaslab is free * and thus is not fragmented. */ if (msp->ms_sm == NULL) { msp->ms_fragmentation = 0; return; } /* * If this metaslab's space map has not been upgraded, flag it * so that we upgrade next time we encounter it. */ if (msp->ms_sm->sm_dbuf->db_size != sizeof (space_map_phys_t)) { uint64_t txg = spa_syncing_txg(spa); vdev_t *vd = msp->ms_group->mg_vd; /* * If we've reached the final dirty txg, then we must * be shutting down the pool. We don't want to dirty * any data past this point so skip setting the condense * flag. We can retry this action the next time the pool * is imported. We also skip marking this metaslab for * condensing if the caller has explicitly set nodirty. */ if (!nodirty && spa_writeable(spa) && txg < spa_final_dirty_txg(spa)) { msp->ms_condense_wanted = B_TRUE; vdev_dirty(vd, VDD_METASLAB, msp, txg + 1); zfs_dbgmsg("txg %llu, requesting force condense: " "ms_id %llu, vdev_id %llu", txg, msp->ms_id, vd->vdev_id); } msp->ms_fragmentation = ZFS_FRAG_INVALID; return; } for (int i = 0; i < SPACE_MAP_HISTOGRAM_SIZE; i++) { uint64_t space = 0; uint8_t shift = msp->ms_sm->sm_shift; int idx = MIN(shift - SPA_MINBLOCKSHIFT + i, FRAGMENTATION_TABLE_SIZE - 1); if (msp->ms_sm->sm_phys->smp_histogram[i] == 0) continue; space = msp->ms_sm->sm_phys->smp_histogram[i] << (i + shift); total += space; ASSERT3U(idx, <, FRAGMENTATION_TABLE_SIZE); fragmentation += space * zfs_frag_table[idx]; } if (total > 0) fragmentation /= total; ASSERT3U(fragmentation, <=, 100); msp->ms_fragmentation = fragmentation; } /* * Compute a weight -- a selection preference value -- for the given metaslab. * This is based on the amount of free space, the level of fragmentation, * the LBA range, and whether the metaslab is loaded. */ static uint64_t metaslab_space_weight(metaslab_t *msp) { metaslab_group_t *mg = msp->ms_group; vdev_t *vd = mg->mg_vd; uint64_t weight, space; ASSERT(MUTEX_HELD(&msp->ms_lock)); /* * The baseline weight is the metaslab's free space. */ space = msp->ms_size - metaslab_allocated_space(msp); if (metaslab_fragmentation_factor_enabled && msp->ms_fragmentation != ZFS_FRAG_INVALID) { /* * Use the fragmentation information to inversely scale * down the baseline weight. We need to ensure that we * don't exclude this metaslab completely when it's 100% * fragmented. To avoid this we reduce the fragmented value * by 1. */ space = (space * (100 - (msp->ms_fragmentation - 1))) / 100; /* * If space < SPA_MINBLOCKSIZE, then we will not allocate from * this metaslab again. The fragmentation metric may have * decreased the space to something smaller than * SPA_MINBLOCKSIZE, so reset the space to SPA_MINBLOCKSIZE * so that we can consume any remaining space. */ if (space > 0 && space < SPA_MINBLOCKSIZE) space = SPA_MINBLOCKSIZE; } weight = space; /* * Modern disks have uniform bit density and constant angular velocity. * Therefore, the outer recording zones are faster (higher bandwidth) * than the inner zones by the ratio of outer to inner track diameter, * which is typically around 2:1. We account for this by assigning * higher weight to lower metaslabs (multiplier ranging from 2x to 1x). * In effect, this means that we'll select the metaslab with the most * free bandwidth rather than simply the one with the most free space. */ if (!vd->vdev_nonrot && metaslab_lba_weighting_enabled) { weight = 2 * weight - (msp->ms_id * weight) / vd->vdev_ms_count; ASSERT(weight >= space && weight <= 2 * space); } /* * If this metaslab is one we're actively using, adjust its * weight to make it preferable to any inactive metaslab so * we'll polish it off. If the fragmentation on this metaslab * has exceed our threshold, then don't mark it active. */ if (msp->ms_loaded && msp->ms_fragmentation != ZFS_FRAG_INVALID && msp->ms_fragmentation <= zfs_metaslab_fragmentation_threshold) { weight |= (msp->ms_weight & METASLAB_ACTIVE_MASK); } WEIGHT_SET_SPACEBASED(weight); return (weight); } /* * Return the weight of the specified metaslab, according to the segment-based * weighting algorithm. The metaslab must be loaded. This function can * be called within a sync pass since it relies only on the metaslab's * range tree which is always accurate when the metaslab is loaded. */ static uint64_t metaslab_weight_from_range_tree(metaslab_t *msp) { uint64_t weight = 0; uint32_t segments = 0; ASSERT(msp->ms_loaded); for (int i = RANGE_TREE_HISTOGRAM_SIZE - 1; i >= SPA_MINBLOCKSHIFT; i--) { uint8_t shift = msp->ms_group->mg_vd->vdev_ashift; int max_idx = SPACE_MAP_HISTOGRAM_SIZE + shift - 1; segments <<= 1; segments += msp->ms_allocatable->rt_histogram[i]; /* * The range tree provides more precision than the space map * and must be downgraded so that all values fit within the * space map's histogram. This allows us to compare loaded * vs. unloaded metaslabs to determine which metaslab is * considered "best". */ if (i > max_idx) continue; if (segments != 0) { WEIGHT_SET_COUNT(weight, segments); WEIGHT_SET_INDEX(weight, i); WEIGHT_SET_ACTIVE(weight, 0); break; } } return (weight); } /* * Calculate the weight based on the on-disk histogram. Should be applied * only to unloaded metaslabs (i.e no incoming allocations) in-order to * give results consistent with the on-disk state */ static uint64_t metaslab_weight_from_spacemap(metaslab_t *msp) { space_map_t *sm = msp->ms_sm; ASSERT(!msp->ms_loaded); ASSERT(sm != NULL); ASSERT3U(space_map_object(sm), !=, 0); ASSERT3U(sm->sm_dbuf->db_size, ==, sizeof (space_map_phys_t)); /* * Create a joint histogram from all the segments that have made * it to the metaslab's space map histogram, that are not yet * available for allocation because they are still in the freeing * pipeline (e.g. freeing, freed, and defer trees). Then subtract * these segments from the space map's histogram to get a more * accurate weight. */ uint64_t deferspace_histogram[SPACE_MAP_HISTOGRAM_SIZE] = {0}; for (int i = 0; i < SPACE_MAP_HISTOGRAM_SIZE; i++) deferspace_histogram[i] += msp->ms_synchist[i]; for (int t = 0; t < TXG_DEFER_SIZE; t++) { for (int i = 0; i < SPACE_MAP_HISTOGRAM_SIZE; i++) { deferspace_histogram[i] += msp->ms_deferhist[t][i]; } } uint64_t weight = 0; for (int i = SPACE_MAP_HISTOGRAM_SIZE - 1; i >= 0; i--) { ASSERT3U(sm->sm_phys->smp_histogram[i], >=, deferspace_histogram[i]); uint64_t count = sm->sm_phys->smp_histogram[i] - deferspace_histogram[i]; if (count != 0) { WEIGHT_SET_COUNT(weight, count); WEIGHT_SET_INDEX(weight, i + sm->sm_shift); WEIGHT_SET_ACTIVE(weight, 0); break; } } return (weight); } /* * Compute a segment-based weight for the specified metaslab. The weight * is determined by highest bucket in the histogram. The information * for the highest bucket is encoded into the weight value. */ static uint64_t metaslab_segment_weight(metaslab_t *msp) { metaslab_group_t *mg = msp->ms_group; uint64_t weight = 0; uint8_t shift = mg->mg_vd->vdev_ashift; ASSERT(MUTEX_HELD(&msp->ms_lock)); /* * The metaslab is completely free. */ if (metaslab_allocated_space(msp) == 0) { int idx = highbit64(msp->ms_size) - 1; int max_idx = SPACE_MAP_HISTOGRAM_SIZE + shift - 1; if (idx < max_idx) { WEIGHT_SET_COUNT(weight, 1ULL); WEIGHT_SET_INDEX(weight, idx); } else { WEIGHT_SET_COUNT(weight, 1ULL << (idx - max_idx)); WEIGHT_SET_INDEX(weight, max_idx); } WEIGHT_SET_ACTIVE(weight, 0); ASSERT(!WEIGHT_IS_SPACEBASED(weight)); return (weight); } ASSERT3U(msp->ms_sm->sm_dbuf->db_size, ==, sizeof (space_map_phys_t)); /* * If the metaslab is fully allocated then just make the weight 0. */ if (metaslab_allocated_space(msp) == msp->ms_size) return (0); /* * If the metaslab is already loaded, then use the range tree to * determine the weight. Otherwise, we rely on the space map information * to generate the weight. */ if (msp->ms_loaded) { weight = metaslab_weight_from_range_tree(msp); } else { weight = metaslab_weight_from_spacemap(msp); } /* * If the metaslab was active the last time we calculated its weight * then keep it active. We want to consume the entire region that * is associated with this weight. */ if (msp->ms_activation_weight != 0 && weight != 0) WEIGHT_SET_ACTIVE(weight, WEIGHT_GET_ACTIVE(msp->ms_weight)); return (weight); } /* * Determine if we should attempt to allocate from this metaslab. If the * metaslab is loaded, then we can determine if the desired allocation * can be satisfied by looking at the size of the maximum free segment * on that metaslab. Otherwise, we make our decision based on the metaslab's * weight. For segment-based weighting we can determine the maximum * allocation based on the index encoded in its value. For space-based * weights we rely on the entire weight (excluding the weight-type bit). */ static boolean_t metaslab_should_allocate(metaslab_t *msp, uint64_t asize, boolean_t try_hard) { /* * If the metaslab is loaded, ms_max_size is definitive and we can use * the fast check. If it's not, the ms_max_size is a lower bound (once * set), and we should use the fast check as long as we're not in * try_hard and it's been less than zfs_metaslab_max_size_cache_sec * seconds since the metaslab was unloaded. */ if (msp->ms_loaded || (msp->ms_max_size != 0 && !try_hard && gethrtime() < msp->ms_unload_time + SEC2NSEC(zfs_metaslab_max_size_cache_sec))) return (msp->ms_max_size >= asize); boolean_t should_allocate; if (!WEIGHT_IS_SPACEBASED(msp->ms_weight)) { /* * The metaslab segment weight indicates segments in the * range [2^i, 2^(i+1)), where i is the index in the weight. * Since the asize might be in the middle of the range, we * should attempt the allocation if asize < 2^(i+1). */ should_allocate = (asize < 1ULL << (WEIGHT_GET_INDEX(msp->ms_weight) + 1)); } else { should_allocate = (asize <= (msp->ms_weight & ~METASLAB_WEIGHT_TYPE)); } return (should_allocate); } static uint64_t metaslab_weight(metaslab_t *msp, boolean_t nodirty) { vdev_t *vd = msp->ms_group->mg_vd; spa_t *spa = vd->vdev_spa; uint64_t weight; ASSERT(MUTEX_HELD(&msp->ms_lock)); metaslab_set_fragmentation(msp, nodirty); /* * Update the maximum size. If the metaslab is loaded, this will * ensure that we get an accurate maximum size if newly freed space * has been added back into the free tree. If the metaslab is * unloaded, we check if there's a larger free segment in the * unflushed frees. This is a lower bound on the largest allocatable * segment size. Coalescing of adjacent entries may reveal larger * allocatable segments, but we aren't aware of those until loading * the space map into a range tree. */ if (msp->ms_loaded) { msp->ms_max_size = metaslab_largest_allocatable(msp); } else { msp->ms_max_size = MAX(msp->ms_max_size, metaslab_largest_unflushed_free(msp)); } /* * Segment-based weighting requires space map histogram support. */ if (zfs_metaslab_segment_weight_enabled && spa_feature_is_enabled(spa, SPA_FEATURE_SPACEMAP_HISTOGRAM) && (msp->ms_sm == NULL || msp->ms_sm->sm_dbuf->db_size == sizeof (space_map_phys_t))) { weight = metaslab_segment_weight(msp); } else { weight = metaslab_space_weight(msp); } return (weight); } void metaslab_recalculate_weight_and_sort(metaslab_t *msp) { ASSERT(MUTEX_HELD(&msp->ms_lock)); /* note: we preserve the mask (e.g. indication of primary, etc..) */ uint64_t was_active = msp->ms_weight & METASLAB_ACTIVE_MASK; metaslab_group_sort(msp->ms_group, msp, metaslab_weight(msp, B_FALSE) | was_active); } static int metaslab_activate_allocator(metaslab_group_t *mg, metaslab_t *msp, int allocator, uint64_t activation_weight) { metaslab_group_allocator_t *mga = &mg->mg_allocator[allocator]; ASSERT(MUTEX_HELD(&msp->ms_lock)); /* * If we're activating for the claim code, we don't want to actually * set the metaslab up for a specific allocator. */ if (activation_weight == METASLAB_WEIGHT_CLAIM) { ASSERT0(msp->ms_activation_weight); msp->ms_activation_weight = msp->ms_weight; metaslab_group_sort(mg, msp, msp->ms_weight | activation_weight); return (0); } metaslab_t **mspp = (activation_weight == METASLAB_WEIGHT_PRIMARY ? &mga->mga_primary : &mga->mga_secondary); mutex_enter(&mg->mg_lock); if (*mspp != NULL) { mutex_exit(&mg->mg_lock); return (EEXIST); } *mspp = msp; ASSERT3S(msp->ms_allocator, ==, -1); msp->ms_allocator = allocator; msp->ms_primary = (activation_weight == METASLAB_WEIGHT_PRIMARY); ASSERT0(msp->ms_activation_weight); msp->ms_activation_weight = msp->ms_weight; metaslab_group_sort_impl(mg, msp, msp->ms_weight | activation_weight); mutex_exit(&mg->mg_lock); return (0); } static int metaslab_activate(metaslab_t *msp, int allocator, uint64_t activation_weight) { ASSERT(MUTEX_HELD(&msp->ms_lock)); /* * The current metaslab is already activated for us so there * is nothing to do. Already activated though, doesn't mean * that this metaslab is activated for our allocator nor our * requested activation weight. The metaslab could have started * as an active one for our allocator but changed allocators * while we were waiting to grab its ms_lock or we stole it * [see find_valid_metaslab()]. This means that there is a * possibility of passivating a metaslab of another allocator * or from a different activation mask, from this thread. */ if ((msp->ms_weight & METASLAB_ACTIVE_MASK) != 0) { ASSERT(msp->ms_loaded); return (0); } int error = metaslab_load(msp); if (error != 0) { metaslab_group_sort(msp->ms_group, msp, 0); return (error); } /* * When entering metaslab_load() we may have dropped the * ms_lock because we were loading this metaslab, or we * were waiting for another thread to load it for us. In * that scenario, we recheck the weight of the metaslab * to see if it was activated by another thread. * * If the metaslab was activated for another allocator or * it was activated with a different activation weight (e.g. * we wanted to make it a primary but it was activated as * secondary) we return error (EBUSY). * * If the metaslab was activated for the same allocator * and requested activation mask, skip activating it. */ if ((msp->ms_weight & METASLAB_ACTIVE_MASK) != 0) { if (msp->ms_allocator != allocator) return (EBUSY); if ((msp->ms_weight & activation_weight) == 0) return (SET_ERROR(EBUSY)); EQUIV((activation_weight == METASLAB_WEIGHT_PRIMARY), msp->ms_primary); return (0); } /* * If the metaslab has literally 0 space, it will have weight 0. In * that case, don't bother activating it. This can happen if the * metaslab had space during find_valid_metaslab, but another thread * loaded it and used all that space while we were waiting to grab the * lock. */ if (msp->ms_weight == 0) { ASSERT0(range_tree_space(msp->ms_allocatable)); return (SET_ERROR(ENOSPC)); } if ((error = metaslab_activate_allocator(msp->ms_group, msp, allocator, activation_weight)) != 0) { return (error); } ASSERT(msp->ms_loaded); ASSERT(msp->ms_weight & METASLAB_ACTIVE_MASK); return (0); } static void metaslab_passivate_allocator(metaslab_group_t *mg, metaslab_t *msp, uint64_t weight) { ASSERT(MUTEX_HELD(&msp->ms_lock)); ASSERT(msp->ms_loaded); if (msp->ms_weight & METASLAB_WEIGHT_CLAIM) { metaslab_group_sort(mg, msp, weight); return; } mutex_enter(&mg->mg_lock); ASSERT3P(msp->ms_group, ==, mg); ASSERT3S(0, <=, msp->ms_allocator); ASSERT3U(msp->ms_allocator, <, mg->mg_allocators); metaslab_group_allocator_t *mga = &mg->mg_allocator[msp->ms_allocator]; if (msp->ms_primary) { ASSERT3P(mga->mga_primary, ==, msp); ASSERT(msp->ms_weight & METASLAB_WEIGHT_PRIMARY); mga->mga_primary = NULL; } else { ASSERT3P(mga->mga_secondary, ==, msp); ASSERT(msp->ms_weight & METASLAB_WEIGHT_SECONDARY); mga->mga_secondary = NULL; } msp->ms_allocator = -1; metaslab_group_sort_impl(mg, msp, weight); mutex_exit(&mg->mg_lock); } static void metaslab_passivate(metaslab_t *msp, uint64_t weight) { uint64_t size __maybe_unused = weight & ~METASLAB_WEIGHT_TYPE; /* * If size < SPA_MINBLOCKSIZE, then we will not allocate from * this metaslab again. In that case, it had better be empty, * or we would be leaving space on the table. */ ASSERT(!WEIGHT_IS_SPACEBASED(msp->ms_weight) || size >= SPA_MINBLOCKSIZE || range_tree_space(msp->ms_allocatable) == 0); ASSERT0(weight & METASLAB_ACTIVE_MASK); ASSERT(msp->ms_activation_weight != 0); msp->ms_activation_weight = 0; metaslab_passivate_allocator(msp->ms_group, msp, weight); ASSERT0(msp->ms_weight & METASLAB_ACTIVE_MASK); } /* * Segment-based metaslabs are activated once and remain active until * we either fail an allocation attempt (similar to space-based metaslabs) * or have exhausted the free space in zfs_metaslab_switch_threshold * buckets since the metaslab was activated. This function checks to see * if we've exhausted the zfs_metaslab_switch_threshold buckets in the * metaslab and passivates it proactively. This will allow us to select a * metaslab with a larger contiguous region, if any, remaining within this * metaslab group. If we're in sync pass > 1, then we continue using this * metaslab so that we don't dirty more block and cause more sync passes. */ static void metaslab_segment_may_passivate(metaslab_t *msp) { spa_t *spa = msp->ms_group->mg_vd->vdev_spa; if (WEIGHT_IS_SPACEBASED(msp->ms_weight) || spa_sync_pass(spa) > 1) return; /* * Since we are in the middle of a sync pass, the most accurate * information that is accessible to us is the in-core range tree * histogram; calculate the new weight based on that information. */ uint64_t weight = metaslab_weight_from_range_tree(msp); int activation_idx = WEIGHT_GET_INDEX(msp->ms_activation_weight); int current_idx = WEIGHT_GET_INDEX(weight); if (current_idx <= activation_idx - zfs_metaslab_switch_threshold) metaslab_passivate(msp, weight); } static void metaslab_preload(void *arg) { metaslab_t *msp = arg; metaslab_class_t *mc = msp->ms_group->mg_class; spa_t *spa = mc->mc_spa; fstrans_cookie_t cookie = spl_fstrans_mark(); ASSERT(!MUTEX_HELD(&msp->ms_group->mg_lock)); mutex_enter(&msp->ms_lock); (void) metaslab_load(msp); metaslab_set_selected_txg(msp, spa_syncing_txg(spa)); mutex_exit(&msp->ms_lock); spl_fstrans_unmark(cookie); } static void metaslab_group_preload(metaslab_group_t *mg) { spa_t *spa = mg->mg_vd->vdev_spa; metaslab_t *msp; avl_tree_t *t = &mg->mg_metaslab_tree; int m = 0; if (spa_shutting_down(spa) || !metaslab_preload_enabled) { taskq_wait_outstanding(mg->mg_taskq, 0); return; } mutex_enter(&mg->mg_lock); /* * Load the next potential metaslabs */ for (msp = avl_first(t); msp != NULL; msp = AVL_NEXT(t, msp)) { ASSERT3P(msp->ms_group, ==, mg); /* * We preload only the maximum number of metaslabs specified * by metaslab_preload_limit. If a metaslab is being forced * to condense then we preload it too. This will ensure * that force condensing happens in the next txg. */ if (++m > metaslab_preload_limit && !msp->ms_condense_wanted) { continue; } VERIFY(taskq_dispatch(mg->mg_taskq, metaslab_preload, msp, TQ_SLEEP) != TASKQID_INVALID); } mutex_exit(&mg->mg_lock); } /* * Determine if the space map's on-disk footprint is past our tolerance for * inefficiency. We would like to use the following criteria to make our * decision: * * 1. Do not condense if the size of the space map object would dramatically * increase as a result of writing out the free space range tree. * * 2. Condense if the on on-disk space map representation is at least * zfs_condense_pct/100 times the size of the optimal representation * (i.e. zfs_condense_pct = 110 and in-core = 1MB, optimal = 1.1MB). * * 3. Do not condense if the on-disk size of the space map does not actually * decrease. * * Unfortunately, we cannot compute the on-disk size of the space map in this * context because we cannot accurately compute the effects of compression, etc. * Instead, we apply the heuristic described in the block comment for * zfs_metaslab_condense_block_threshold - we only condense if the space used * is greater than a threshold number of blocks. */ static boolean_t metaslab_should_condense(metaslab_t *msp) { space_map_t *sm = msp->ms_sm; vdev_t *vd = msp->ms_group->mg_vd; uint64_t vdev_blocksize = 1 << vd->vdev_ashift; ASSERT(MUTEX_HELD(&msp->ms_lock)); ASSERT(msp->ms_loaded); ASSERT(sm != NULL); ASSERT3U(spa_sync_pass(vd->vdev_spa), ==, 1); /* * We always condense metaslabs that are empty and metaslabs for * which a condense request has been made. */ if (range_tree_numsegs(msp->ms_allocatable) == 0 || msp->ms_condense_wanted) return (B_TRUE); uint64_t record_size = MAX(sm->sm_blksz, vdev_blocksize); uint64_t object_size = space_map_length(sm); uint64_t optimal_size = space_map_estimate_optimal_size(sm, msp->ms_allocatable, SM_NO_VDEVID); return (object_size >= (optimal_size * zfs_condense_pct / 100) && object_size > zfs_metaslab_condense_block_threshold * record_size); } /* * Condense the on-disk space map representation to its minimized form. * The minimized form consists of a small number of allocations followed * by the entries of the free range tree (ms_allocatable). The condensed * spacemap contains all the entries of previous TXGs (including those in * the pool-wide log spacemaps; thus this is effectively a superset of * metaslab_flush()), but this TXG's entries still need to be written. */ static void metaslab_condense(metaslab_t *msp, dmu_tx_t *tx) { range_tree_t *condense_tree; space_map_t *sm = msp->ms_sm; uint64_t txg = dmu_tx_get_txg(tx); spa_t *spa = msp->ms_group->mg_vd->vdev_spa; ASSERT(MUTEX_HELD(&msp->ms_lock)); ASSERT(msp->ms_loaded); ASSERT(msp->ms_sm != NULL); /* * In order to condense the space map, we need to change it so it * only describes which segments are currently allocated and free. * * All the current free space resides in the ms_allocatable, all * the ms_defer trees, and all the ms_allocating trees. We ignore * ms_freed because it is empty because we're in sync pass 1. We * ignore ms_freeing because these changes are not yet reflected * in the spacemap (they will be written later this txg). * * So to truncate the space map to represent all the entries of * previous TXGs we do the following: * * 1] We create a range tree (condense tree) that is 100% empty. * 2] We add to it all segments found in the ms_defer trees * as those segments are marked as free in the original space * map. We do the same with the ms_allocating trees for the same * reason. Adding these segments should be a relatively * inexpensive operation since we expect these trees to have a * small number of nodes. * 3] We vacate any unflushed allocs, since they are not frees we * need to add to the condense tree. Then we vacate any * unflushed frees as they should already be part of ms_allocatable. * 4] At this point, we would ideally like to add all segments * in the ms_allocatable tree from the condense tree. This way * we would write all the entries of the condense tree as the * condensed space map, which would only contain freed * segments with everything else assumed to be allocated. * * Doing so can be prohibitively expensive as ms_allocatable can * be large, and therefore computationally expensive to add to * the condense_tree. Instead we first sync out an entry marking * everything as allocated, then the condense_tree and then the * ms_allocatable, in the condensed space map. While this is not * optimal, it is typically close to optimal and more importantly * much cheaper to compute. * * 5] Finally, as both of the unflushed trees were written to our * new and condensed metaslab space map, we basically flushed * all the unflushed changes to disk, thus we call * metaslab_flush_update(). */ ASSERT3U(spa_sync_pass(spa), ==, 1); ASSERT(range_tree_is_empty(msp->ms_freed)); /* since it is pass 1 */ zfs_dbgmsg("condensing: txg %llu, msp[%llu] %px, vdev id %llu, " "spa %s, smp size %llu, segments %lu, forcing condense=%s", txg, msp->ms_id, msp, msp->ms_group->mg_vd->vdev_id, spa->spa_name, space_map_length(msp->ms_sm), range_tree_numsegs(msp->ms_allocatable), msp->ms_condense_wanted ? "TRUE" : "FALSE"); msp->ms_condense_wanted = B_FALSE; range_seg_type_t type; uint64_t shift, start; type = metaslab_calculate_range_tree_type(msp->ms_group->mg_vd, msp, &start, &shift); condense_tree = range_tree_create(NULL, type, NULL, start, shift); for (int t = 0; t < TXG_DEFER_SIZE; t++) { range_tree_walk(msp->ms_defer[t], range_tree_add, condense_tree); } for (int t = 0; t < TXG_CONCURRENT_STATES; t++) { range_tree_walk(msp->ms_allocating[(txg + t) & TXG_MASK], range_tree_add, condense_tree); } ASSERT3U(spa->spa_unflushed_stats.sus_memused, >=, metaslab_unflushed_changes_memused(msp)); spa->spa_unflushed_stats.sus_memused -= metaslab_unflushed_changes_memused(msp); range_tree_vacate(msp->ms_unflushed_allocs, NULL, NULL); range_tree_vacate(msp->ms_unflushed_frees, NULL, NULL); /* * We're about to drop the metaslab's lock thus allowing other * consumers to change it's content. Set the metaslab's ms_condensing * flag to ensure that allocations on this metaslab do not occur * while we're in the middle of committing it to disk. This is only * critical for ms_allocatable as all other range trees use per TXG * views of their content. */ msp->ms_condensing = B_TRUE; mutex_exit(&msp->ms_lock); uint64_t object = space_map_object(msp->ms_sm); space_map_truncate(sm, spa_feature_is_enabled(spa, SPA_FEATURE_LOG_SPACEMAP) ? zfs_metaslab_sm_blksz_with_log : zfs_metaslab_sm_blksz_no_log, tx); /* * space_map_truncate() may have reallocated the spacemap object. * If so, update the vdev_ms_array. */ if (space_map_object(msp->ms_sm) != object) { object = space_map_object(msp->ms_sm); dmu_write(spa->spa_meta_objset, msp->ms_group->mg_vd->vdev_ms_array, sizeof (uint64_t) * msp->ms_id, sizeof (uint64_t), &object, tx); } /* * Note: * When the log space map feature is enabled, each space map will * always have ALLOCS followed by FREES for each sync pass. This is * typically true even when the log space map feature is disabled, * except from the case where a metaslab goes through metaslab_sync() * and gets condensed. In that case the metaslab's space map will have * ALLOCS followed by FREES (due to condensing) followed by ALLOCS * followed by FREES (due to space_map_write() in metaslab_sync()) for * sync pass 1. */ range_tree_t *tmp_tree = range_tree_create(NULL, type, NULL, start, shift); range_tree_add(tmp_tree, msp->ms_start, msp->ms_size); space_map_write(sm, tmp_tree, SM_ALLOC, SM_NO_VDEVID, tx); space_map_write(sm, msp->ms_allocatable, SM_FREE, SM_NO_VDEVID, tx); space_map_write(sm, condense_tree, SM_FREE, SM_NO_VDEVID, tx); range_tree_vacate(condense_tree, NULL, NULL); range_tree_destroy(condense_tree); range_tree_vacate(tmp_tree, NULL, NULL); range_tree_destroy(tmp_tree); mutex_enter(&msp->ms_lock); msp->ms_condensing = B_FALSE; metaslab_flush_update(msp, tx); } /* * Called when the metaslab has been flushed (its own spacemap now reflects * all the contents of the pool-wide spacemap log). Updates the metaslab's * metadata and any pool-wide related log space map data (e.g. summary, * obsolete logs, etc..) to reflect that. */ static void metaslab_flush_update(metaslab_t *msp, dmu_tx_t *tx) { metaslab_group_t *mg = msp->ms_group; spa_t *spa = mg->mg_vd->vdev_spa; ASSERT(MUTEX_HELD(&msp->ms_lock)); ASSERT3U(spa_sync_pass(spa), ==, 1); ASSERT(range_tree_is_empty(msp->ms_unflushed_allocs)); ASSERT(range_tree_is_empty(msp->ms_unflushed_frees)); /* * Just because a metaslab got flushed, that doesn't mean that * it will pass through metaslab_sync_done(). Thus, make sure to * update ms_synced_length here in case it doesn't. */ msp->ms_synced_length = space_map_length(msp->ms_sm); /* * We may end up here from metaslab_condense() without the * feature being active. In that case this is a no-op. */ if (!spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP)) return; ASSERT(spa_syncing_log_sm(spa) != NULL); ASSERT(msp->ms_sm != NULL); ASSERT(metaslab_unflushed_txg(msp) != 0); ASSERT3P(avl_find(&spa->spa_metaslabs_by_flushed, msp, NULL), ==, msp); VERIFY3U(tx->tx_txg, <=, spa_final_dirty_txg(spa)); /* update metaslab's position in our flushing tree */ uint64_t ms_prev_flushed_txg = metaslab_unflushed_txg(msp); mutex_enter(&spa->spa_flushed_ms_lock); avl_remove(&spa->spa_metaslabs_by_flushed, msp); metaslab_set_unflushed_txg(msp, spa_syncing_txg(spa), tx); avl_add(&spa->spa_metaslabs_by_flushed, msp); mutex_exit(&spa->spa_flushed_ms_lock); /* update metaslab counts of spa_log_sm_t nodes */ spa_log_sm_decrement_mscount(spa, ms_prev_flushed_txg); spa_log_sm_increment_current_mscount(spa); /* cleanup obsolete logs if any */ uint64_t log_blocks_before = spa_log_sm_nblocks(spa); spa_cleanup_old_sm_logs(spa, tx); uint64_t log_blocks_after = spa_log_sm_nblocks(spa); VERIFY3U(log_blocks_after, <=, log_blocks_before); /* update log space map summary */ uint64_t blocks_gone = log_blocks_before - log_blocks_after; spa_log_summary_add_flushed_metaslab(spa); spa_log_summary_decrement_mscount(spa, ms_prev_flushed_txg); spa_log_summary_decrement_blkcount(spa, blocks_gone); } boolean_t metaslab_flush(metaslab_t *msp, dmu_tx_t *tx) { spa_t *spa = msp->ms_group->mg_vd->vdev_spa; ASSERT(MUTEX_HELD(&msp->ms_lock)); ASSERT3U(spa_sync_pass(spa), ==, 1); ASSERT(spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP)); ASSERT(msp->ms_sm != NULL); ASSERT(metaslab_unflushed_txg(msp) != 0); ASSERT(avl_find(&spa->spa_metaslabs_by_flushed, msp, NULL) != NULL); /* * There is nothing wrong with flushing the same metaslab twice, as * this codepath should work on that case. However, the current * flushing scheme makes sure to avoid this situation as we would be * making all these calls without having anything meaningful to write * to disk. We assert this behavior here. */ ASSERT3U(metaslab_unflushed_txg(msp), <, dmu_tx_get_txg(tx)); /* * We can not flush while loading, because then we would * not load the ms_unflushed_{allocs,frees}. */ if (msp->ms_loading) return (B_FALSE); metaslab_verify_space(msp, dmu_tx_get_txg(tx)); metaslab_verify_weight_and_frag(msp); /* * Metaslab condensing is effectively flushing. Therefore if the * metaslab can be condensed we can just condense it instead of * flushing it. * * Note that metaslab_condense() does call metaslab_flush_update() * so we can just return immediately after condensing. We also * don't need to care about setting ms_flushing or broadcasting * ms_flush_cv, even if we temporarily drop the ms_lock in * metaslab_condense(), as the metaslab is already loaded. */ if (msp->ms_loaded && metaslab_should_condense(msp)) { metaslab_group_t *mg = msp->ms_group; /* * For all histogram operations below refer to the * comments of metaslab_sync() where we follow a * similar procedure. */ metaslab_group_histogram_verify(mg); metaslab_class_histogram_verify(mg->mg_class); metaslab_group_histogram_remove(mg, msp); metaslab_condense(msp, tx); space_map_histogram_clear(msp->ms_sm); space_map_histogram_add(msp->ms_sm, msp->ms_allocatable, tx); ASSERT(range_tree_is_empty(msp->ms_freed)); for (int t = 0; t < TXG_DEFER_SIZE; t++) { space_map_histogram_add(msp->ms_sm, msp->ms_defer[t], tx); } metaslab_aux_histograms_update(msp); metaslab_group_histogram_add(mg, msp); metaslab_group_histogram_verify(mg); metaslab_class_histogram_verify(mg->mg_class); metaslab_verify_space(msp, dmu_tx_get_txg(tx)); /* * Since we recreated the histogram (and potentially * the ms_sm too while condensing) ensure that the * weight is updated too because we are not guaranteed * that this metaslab is dirty and will go through * metaslab_sync_done(). */ metaslab_recalculate_weight_and_sort(msp); return (B_TRUE); } msp->ms_flushing = B_TRUE; uint64_t sm_len_before = space_map_length(msp->ms_sm); mutex_exit(&msp->ms_lock); space_map_write(msp->ms_sm, msp->ms_unflushed_allocs, SM_ALLOC, SM_NO_VDEVID, tx); space_map_write(msp->ms_sm, msp->ms_unflushed_frees, SM_FREE, SM_NO_VDEVID, tx); mutex_enter(&msp->ms_lock); uint64_t sm_len_after = space_map_length(msp->ms_sm); if (zfs_flags & ZFS_DEBUG_LOG_SPACEMAP) { zfs_dbgmsg("flushing: txg %llu, spa %s, vdev_id %llu, " "ms_id %llu, unflushed_allocs %llu, unflushed_frees %llu, " "appended %llu bytes", dmu_tx_get_txg(tx), spa_name(spa), msp->ms_group->mg_vd->vdev_id, msp->ms_id, range_tree_space(msp->ms_unflushed_allocs), range_tree_space(msp->ms_unflushed_frees), (sm_len_after - sm_len_before)); } ASSERT3U(spa->spa_unflushed_stats.sus_memused, >=, metaslab_unflushed_changes_memused(msp)); spa->spa_unflushed_stats.sus_memused -= metaslab_unflushed_changes_memused(msp); range_tree_vacate(msp->ms_unflushed_allocs, NULL, NULL); range_tree_vacate(msp->ms_unflushed_frees, NULL, NULL); metaslab_verify_space(msp, dmu_tx_get_txg(tx)); metaslab_verify_weight_and_frag(msp); metaslab_flush_update(msp, tx); metaslab_verify_space(msp, dmu_tx_get_txg(tx)); metaslab_verify_weight_and_frag(msp); msp->ms_flushing = B_FALSE; cv_broadcast(&msp->ms_flush_cv); return (B_TRUE); } /* * Write a metaslab to disk in the context of the specified transaction group. */ void metaslab_sync(metaslab_t *msp, uint64_t txg) { metaslab_group_t *mg = msp->ms_group; vdev_t *vd = mg->mg_vd; spa_t *spa = vd->vdev_spa; objset_t *mos = spa_meta_objset(spa); range_tree_t *alloctree = msp->ms_allocating[txg & TXG_MASK]; dmu_tx_t *tx; ASSERT(!vd->vdev_ishole); /* * This metaslab has just been added so there's no work to do now. */ if (msp->ms_freeing == NULL) { ASSERT3P(alloctree, ==, NULL); return; } ASSERT3P(alloctree, !=, NULL); ASSERT3P(msp->ms_freeing, !=, NULL); ASSERT3P(msp->ms_freed, !=, NULL); ASSERT3P(msp->ms_checkpointing, !=, NULL); ASSERT3P(msp->ms_trim, !=, NULL); /* * Normally, we don't want to process a metaslab if there are no * allocations or frees to perform. However, if the metaslab is being * forced to condense, it's loaded and we're not beyond the final * dirty txg, we need to let it through. Not condensing beyond the * final dirty txg prevents an issue where metaslabs that need to be * condensed but were loaded for other reasons could cause a panic * here. By only checking the txg in that branch of the conditional, * we preserve the utility of the VERIFY statements in all other * cases. */ if (range_tree_is_empty(alloctree) && range_tree_is_empty(msp->ms_freeing) && range_tree_is_empty(msp->ms_checkpointing) && !(msp->ms_loaded && msp->ms_condense_wanted && txg <= spa_final_dirty_txg(spa))) return; VERIFY3U(txg, <=, spa_final_dirty_txg(spa)); /* * The only state that can actually be changing concurrently * with metaslab_sync() is the metaslab's ms_allocatable. No * other thread can be modifying this txg's alloc, freeing, * freed, or space_map_phys_t. We drop ms_lock whenever we * could call into the DMU, because the DMU can call down to * us (e.g. via zio_free()) at any time. * * The spa_vdev_remove_thread() can be reading metaslab state * concurrently, and it is locked out by the ms_sync_lock. * Note that the ms_lock is insufficient for this, because it * is dropped by space_map_write(). */ tx = dmu_tx_create_assigned(spa_get_dsl(spa), txg); /* * Generate a log space map if one doesn't exist already. */ spa_generate_syncing_log_sm(spa, tx); if (msp->ms_sm == NULL) { uint64_t new_object = space_map_alloc(mos, spa_feature_is_enabled(spa, SPA_FEATURE_LOG_SPACEMAP) ? zfs_metaslab_sm_blksz_with_log : zfs_metaslab_sm_blksz_no_log, tx); VERIFY3U(new_object, !=, 0); dmu_write(mos, vd->vdev_ms_array, sizeof (uint64_t) * msp->ms_id, sizeof (uint64_t), &new_object, tx); VERIFY0(space_map_open(&msp->ms_sm, mos, new_object, msp->ms_start, msp->ms_size, vd->vdev_ashift)); ASSERT(msp->ms_sm != NULL); ASSERT(range_tree_is_empty(msp->ms_unflushed_allocs)); ASSERT(range_tree_is_empty(msp->ms_unflushed_frees)); ASSERT0(metaslab_allocated_space(msp)); } if (metaslab_unflushed_txg(msp) == 0 && spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP)) { ASSERT(spa_syncing_log_sm(spa) != NULL); metaslab_set_unflushed_txg(msp, spa_syncing_txg(spa), tx); spa_log_sm_increment_current_mscount(spa); spa_log_summary_add_flushed_metaslab(spa); ASSERT(msp->ms_sm != NULL); mutex_enter(&spa->spa_flushed_ms_lock); avl_add(&spa->spa_metaslabs_by_flushed, msp); mutex_exit(&spa->spa_flushed_ms_lock); ASSERT(range_tree_is_empty(msp->ms_unflushed_allocs)); ASSERT(range_tree_is_empty(msp->ms_unflushed_frees)); } if (!range_tree_is_empty(msp->ms_checkpointing) && vd->vdev_checkpoint_sm == NULL) { ASSERT(spa_has_checkpoint(spa)); uint64_t new_object = space_map_alloc(mos, zfs_vdev_standard_sm_blksz, tx); VERIFY3U(new_object, !=, 0); VERIFY0(space_map_open(&vd->vdev_checkpoint_sm, mos, new_object, 0, vd->vdev_asize, vd->vdev_ashift)); ASSERT3P(vd->vdev_checkpoint_sm, !=, NULL); /* * We save the space map object as an entry in vdev_top_zap * so it can be retrieved when the pool is reopened after an * export or through zdb. */ VERIFY0(zap_add(vd->vdev_spa->spa_meta_objset, vd->vdev_top_zap, VDEV_TOP_ZAP_POOL_CHECKPOINT_SM, sizeof (new_object), 1, &new_object, tx)); } mutex_enter(&msp->ms_sync_lock); mutex_enter(&msp->ms_lock); /* * Note: metaslab_condense() clears the space map's histogram. * Therefore we must verify and remove this histogram before * condensing. */ metaslab_group_histogram_verify(mg); metaslab_class_histogram_verify(mg->mg_class); metaslab_group_histogram_remove(mg, msp); if (spa->spa_sync_pass == 1 && msp->ms_loaded && metaslab_should_condense(msp)) metaslab_condense(msp, tx); /* * We'll be going to disk to sync our space accounting, thus we * drop the ms_lock during that time so allocations coming from * open-context (ZIL) for future TXGs do not block. */ mutex_exit(&msp->ms_lock); space_map_t *log_sm = spa_syncing_log_sm(spa); if (log_sm != NULL) { ASSERT(spa_feature_is_enabled(spa, SPA_FEATURE_LOG_SPACEMAP)); space_map_write(log_sm, alloctree, SM_ALLOC, vd->vdev_id, tx); space_map_write(log_sm, msp->ms_freeing, SM_FREE, vd->vdev_id, tx); mutex_enter(&msp->ms_lock); ASSERT3U(spa->spa_unflushed_stats.sus_memused, >=, metaslab_unflushed_changes_memused(msp)); spa->spa_unflushed_stats.sus_memused -= metaslab_unflushed_changes_memused(msp); range_tree_remove_xor_add(alloctree, msp->ms_unflushed_frees, msp->ms_unflushed_allocs); range_tree_remove_xor_add(msp->ms_freeing, msp->ms_unflushed_allocs, msp->ms_unflushed_frees); spa->spa_unflushed_stats.sus_memused += metaslab_unflushed_changes_memused(msp); } else { ASSERT(!spa_feature_is_enabled(spa, SPA_FEATURE_LOG_SPACEMAP)); space_map_write(msp->ms_sm, alloctree, SM_ALLOC, SM_NO_VDEVID, tx); space_map_write(msp->ms_sm, msp->ms_freeing, SM_FREE, SM_NO_VDEVID, tx); mutex_enter(&msp->ms_lock); } msp->ms_allocated_space += range_tree_space(alloctree); ASSERT3U(msp->ms_allocated_space, >=, range_tree_space(msp->ms_freeing)); msp->ms_allocated_space -= range_tree_space(msp->ms_freeing); if (!range_tree_is_empty(msp->ms_checkpointing)) { ASSERT(spa_has_checkpoint(spa)); ASSERT3P(vd->vdev_checkpoint_sm, !=, NULL); /* * Since we are doing writes to disk and the ms_checkpointing * tree won't be changing during that time, we drop the * ms_lock while writing to the checkpoint space map, for the * same reason mentioned above. */ mutex_exit(&msp->ms_lock); space_map_write(vd->vdev_checkpoint_sm, msp->ms_checkpointing, SM_FREE, SM_NO_VDEVID, tx); mutex_enter(&msp->ms_lock); spa->spa_checkpoint_info.sci_dspace += range_tree_space(msp->ms_checkpointing); vd->vdev_stat.vs_checkpoint_space += range_tree_space(msp->ms_checkpointing); ASSERT3U(vd->vdev_stat.vs_checkpoint_space, ==, -space_map_allocated(vd->vdev_checkpoint_sm)); range_tree_vacate(msp->ms_checkpointing, NULL, NULL); } if (msp->ms_loaded) { /* * When the space map is loaded, we have an accurate * histogram in the range tree. This gives us an opportunity * to bring the space map's histogram up-to-date so we clear * it first before updating it. */ space_map_histogram_clear(msp->ms_sm); space_map_histogram_add(msp->ms_sm, msp->ms_allocatable, tx); /* * Since we've cleared the histogram we need to add back * any free space that has already been processed, plus * any deferred space. This allows the on-disk histogram * to accurately reflect all free space even if some space * is not yet available for allocation (i.e. deferred). */ space_map_histogram_add(msp->ms_sm, msp->ms_freed, tx); /* * Add back any deferred free space that has not been * added back into the in-core free tree yet. This will * ensure that we don't end up with a space map histogram * that is completely empty unless the metaslab is fully * allocated. */ for (int t = 0; t < TXG_DEFER_SIZE; t++) { space_map_histogram_add(msp->ms_sm, msp->ms_defer[t], tx); } } /* * Always add the free space from this sync pass to the space * map histogram. We want to make sure that the on-disk histogram * accounts for all free space. If the space map is not loaded, * then we will lose some accuracy but will correct it the next * time we load the space map. */ space_map_histogram_add(msp->ms_sm, msp->ms_freeing, tx); metaslab_aux_histograms_update(msp); metaslab_group_histogram_add(mg, msp); metaslab_group_histogram_verify(mg); metaslab_class_histogram_verify(mg->mg_class); /* * For sync pass 1, we avoid traversing this txg's free range tree * and instead will just swap the pointers for freeing and freed. * We can safely do this since the freed_tree is guaranteed to be * empty on the initial pass. * * Keep in mind that even if we are currently using a log spacemap * we want current frees to end up in the ms_allocatable (but not * get appended to the ms_sm) so their ranges can be reused as usual. */ if (spa_sync_pass(spa) == 1) { range_tree_swap(&msp->ms_freeing, &msp->ms_freed); ASSERT0(msp->ms_allocated_this_txg); } else { range_tree_vacate(msp->ms_freeing, range_tree_add, msp->ms_freed); } msp->ms_allocated_this_txg += range_tree_space(alloctree); range_tree_vacate(alloctree, NULL, NULL); ASSERT0(range_tree_space(msp->ms_allocating[txg & TXG_MASK])); ASSERT0(range_tree_space(msp->ms_allocating[TXG_CLEAN(txg) & TXG_MASK])); ASSERT0(range_tree_space(msp->ms_freeing)); ASSERT0(range_tree_space(msp->ms_checkpointing)); mutex_exit(&msp->ms_lock); /* * Verify that the space map object ID has been recorded in the * vdev_ms_array. */ uint64_t object; VERIFY0(dmu_read(mos, vd->vdev_ms_array, msp->ms_id * sizeof (uint64_t), sizeof (uint64_t), &object, 0)); VERIFY3U(object, ==, space_map_object(msp->ms_sm)); mutex_exit(&msp->ms_sync_lock); dmu_tx_commit(tx); } static void metaslab_evict(metaslab_t *msp, uint64_t txg) { if (!msp->ms_loaded || msp->ms_disabled != 0) return; for (int t = 1; t < TXG_CONCURRENT_STATES; t++) { VERIFY0(range_tree_space( msp->ms_allocating[(txg + t) & TXG_MASK])); } if (msp->ms_allocator != -1) metaslab_passivate(msp, msp->ms_weight & ~METASLAB_ACTIVE_MASK); if (!metaslab_debug_unload) metaslab_unload(msp); } /* * Called after a transaction group has completely synced to mark * all of the metaslab's free space as usable. */ void metaslab_sync_done(metaslab_t *msp, uint64_t txg) { metaslab_group_t *mg = msp->ms_group; vdev_t *vd = mg->mg_vd; spa_t *spa = vd->vdev_spa; range_tree_t **defer_tree; int64_t alloc_delta, defer_delta; boolean_t defer_allowed = B_TRUE; ASSERT(!vd->vdev_ishole); mutex_enter(&msp->ms_lock); /* * If this metaslab is just becoming available, initialize its * range trees and add its capacity to the vdev. */ if (msp->ms_freed == NULL) { range_seg_type_t type; uint64_t shift, start; type = metaslab_calculate_range_tree_type(vd, msp, &start, &shift); for (int t = 0; t < TXG_SIZE; t++) { ASSERT(msp->ms_allocating[t] == NULL); msp->ms_allocating[t] = range_tree_create(NULL, type, NULL, start, shift); } ASSERT3P(msp->ms_freeing, ==, NULL); msp->ms_freeing = range_tree_create(NULL, type, NULL, start, shift); ASSERT3P(msp->ms_freed, ==, NULL); msp->ms_freed = range_tree_create(NULL, type, NULL, start, shift); for (int t = 0; t < TXG_DEFER_SIZE; t++) { ASSERT3P(msp->ms_defer[t], ==, NULL); msp->ms_defer[t] = range_tree_create(NULL, type, NULL, start, shift); } ASSERT3P(msp->ms_checkpointing, ==, NULL); msp->ms_checkpointing = range_tree_create(NULL, type, NULL, start, shift); ASSERT3P(msp->ms_unflushed_allocs, ==, NULL); msp->ms_unflushed_allocs = range_tree_create(NULL, type, NULL, start, shift); metaslab_rt_arg_t *mrap = kmem_zalloc(sizeof (*mrap), KM_SLEEP); mrap->mra_bt = &msp->ms_unflushed_frees_by_size; mrap->mra_floor_shift = metaslab_by_size_min_shift; ASSERT3P(msp->ms_unflushed_frees, ==, NULL); msp->ms_unflushed_frees = range_tree_create(&metaslab_rt_ops, type, mrap, start, shift); metaslab_space_update(vd, mg->mg_class, 0, 0, msp->ms_size); } ASSERT0(range_tree_space(msp->ms_freeing)); ASSERT0(range_tree_space(msp->ms_checkpointing)); defer_tree = &msp->ms_defer[txg % TXG_DEFER_SIZE]; uint64_t free_space = metaslab_class_get_space(spa_normal_class(spa)) - metaslab_class_get_alloc(spa_normal_class(spa)); if (free_space <= spa_get_slop_space(spa) || vd->vdev_removing) { defer_allowed = B_FALSE; } defer_delta = 0; alloc_delta = msp->ms_allocated_this_txg - range_tree_space(msp->ms_freed); if (defer_allowed) { defer_delta = range_tree_space(msp->ms_freed) - range_tree_space(*defer_tree); } else { defer_delta -= range_tree_space(*defer_tree); } metaslab_space_update(vd, mg->mg_class, alloc_delta + defer_delta, defer_delta, 0); if (spa_syncing_log_sm(spa) == NULL) { /* * If there's a metaslab_load() in progress and we don't have * a log space map, it means that we probably wrote to the * metaslab's space map. If this is the case, we need to * make sure that we wait for the load to complete so that we * have a consistent view at the in-core side of the metaslab. */ metaslab_load_wait(msp); } else { ASSERT(spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP)); } /* * When auto-trimming is enabled, free ranges which are added to * ms_allocatable are also be added to ms_trim. The ms_trim tree is * periodically consumed by the vdev_autotrim_thread() which issues * trims for all ranges and then vacates the tree. The ms_trim tree * can be discarded at any time with the sole consequence of recent * frees not being trimmed. */ if (spa_get_autotrim(spa) == SPA_AUTOTRIM_ON) { range_tree_walk(*defer_tree, range_tree_add, msp->ms_trim); if (!defer_allowed) { range_tree_walk(msp->ms_freed, range_tree_add, msp->ms_trim); } } else { range_tree_vacate(msp->ms_trim, NULL, NULL); } /* * Move the frees from the defer_tree back to the free * range tree (if it's loaded). Swap the freed_tree and * the defer_tree -- this is safe to do because we've * just emptied out the defer_tree. */ range_tree_vacate(*defer_tree, msp->ms_loaded ? range_tree_add : NULL, msp->ms_allocatable); if (defer_allowed) { range_tree_swap(&msp->ms_freed, defer_tree); } else { range_tree_vacate(msp->ms_freed, msp->ms_loaded ? range_tree_add : NULL, msp->ms_allocatable); } msp->ms_synced_length = space_map_length(msp->ms_sm); msp->ms_deferspace += defer_delta; ASSERT3S(msp->ms_deferspace, >=, 0); ASSERT3S(msp->ms_deferspace, <=, msp->ms_size); if (msp->ms_deferspace != 0) { /* * Keep syncing this metaslab until all deferred frees * are back in circulation. */ vdev_dirty(vd, VDD_METASLAB, msp, txg + 1); } metaslab_aux_histograms_update_done(msp, defer_allowed); if (msp->ms_new) { msp->ms_new = B_FALSE; mutex_enter(&mg->mg_lock); mg->mg_ms_ready++; mutex_exit(&mg->mg_lock); } /* * Re-sort metaslab within its group now that we've adjusted * its allocatable space. */ metaslab_recalculate_weight_and_sort(msp); ASSERT0(range_tree_space(msp->ms_allocating[txg & TXG_MASK])); ASSERT0(range_tree_space(msp->ms_freeing)); ASSERT0(range_tree_space(msp->ms_freed)); ASSERT0(range_tree_space(msp->ms_checkpointing)); msp->ms_allocating_total -= msp->ms_allocated_this_txg; msp->ms_allocated_this_txg = 0; mutex_exit(&msp->ms_lock); } void metaslab_sync_reassess(metaslab_group_t *mg) { spa_t *spa = mg->mg_class->mc_spa; spa_config_enter(spa, SCL_ALLOC, FTAG, RW_READER); metaslab_group_alloc_update(mg); mg->mg_fragmentation = metaslab_group_fragmentation(mg); /* * Preload the next potential metaslabs but only on active * metaslab groups. We can get into a state where the metaslab * is no longer active since we dirty metaslabs as we remove a * a device, thus potentially making the metaslab group eligible * for preloading. */ if (mg->mg_activation_count > 0) { metaslab_group_preload(mg); } spa_config_exit(spa, SCL_ALLOC, FTAG); } /* * When writing a ditto block (i.e. more than one DVA for a given BP) on * the same vdev as an existing DVA of this BP, then try to allocate it * on a different metaslab than existing DVAs (i.e. a unique metaslab). */ static boolean_t metaslab_is_unique(metaslab_t *msp, dva_t *dva) { uint64_t dva_ms_id; if (DVA_GET_ASIZE(dva) == 0) return (B_TRUE); if (msp->ms_group->mg_vd->vdev_id != DVA_GET_VDEV(dva)) return (B_TRUE); dva_ms_id = DVA_GET_OFFSET(dva) >> msp->ms_group->mg_vd->vdev_ms_shift; return (msp->ms_id != dva_ms_id); } /* * ========================================================================== * Metaslab allocation tracing facility * ========================================================================== */ #ifdef _METASLAB_TRACING /* * Add an allocation trace element to the allocation tracing list. */ static void metaslab_trace_add(zio_alloc_list_t *zal, metaslab_group_t *mg, metaslab_t *msp, uint64_t psize, uint32_t dva_id, uint64_t offset, int allocator) { metaslab_alloc_trace_t *mat; if (!metaslab_trace_enabled) return; /* * When the tracing list reaches its maximum we remove * the second element in the list before adding a new one. * By removing the second element we preserve the original * entry as a clue to what allocations steps have already been * performed. */ if (zal->zal_size == metaslab_trace_max_entries) { metaslab_alloc_trace_t *mat_next; #ifdef ZFS_DEBUG panic("too many entries in allocation list"); #endif METASLABSTAT_BUMP(metaslabstat_trace_over_limit); zal->zal_size--; mat_next = list_next(&zal->zal_list, list_head(&zal->zal_list)); list_remove(&zal->zal_list, mat_next); kmem_cache_free(metaslab_alloc_trace_cache, mat_next); } mat = kmem_cache_alloc(metaslab_alloc_trace_cache, KM_SLEEP); list_link_init(&mat->mat_list_node); mat->mat_mg = mg; mat->mat_msp = msp; mat->mat_size = psize; mat->mat_dva_id = dva_id; mat->mat_offset = offset; mat->mat_weight = 0; mat->mat_allocator = allocator; if (msp != NULL) mat->mat_weight = msp->ms_weight; /* * The list is part of the zio so locking is not required. Only * a single thread will perform allocations for a given zio. */ list_insert_tail(&zal->zal_list, mat); zal->zal_size++; ASSERT3U(zal->zal_size, <=, metaslab_trace_max_entries); } void metaslab_trace_init(zio_alloc_list_t *zal) { list_create(&zal->zal_list, sizeof (metaslab_alloc_trace_t), offsetof(metaslab_alloc_trace_t, mat_list_node)); zal->zal_size = 0; } void metaslab_trace_fini(zio_alloc_list_t *zal) { metaslab_alloc_trace_t *mat; while ((mat = list_remove_head(&zal->zal_list)) != NULL) kmem_cache_free(metaslab_alloc_trace_cache, mat); list_destroy(&zal->zal_list); zal->zal_size = 0; } #else #define metaslab_trace_add(zal, mg, msp, psize, id, off, alloc) void metaslab_trace_init(zio_alloc_list_t *zal) { } void metaslab_trace_fini(zio_alloc_list_t *zal) { } #endif /* _METASLAB_TRACING */ /* * ========================================================================== * Metaslab block operations * ========================================================================== */ static void metaslab_group_alloc_increment(spa_t *spa, uint64_t vdev, void *tag, int flags, int allocator) { if (!(flags & METASLAB_ASYNC_ALLOC) || (flags & METASLAB_DONT_THROTTLE)) return; metaslab_group_t *mg = vdev_lookup_top(spa, vdev)->vdev_mg; if (!mg->mg_class->mc_alloc_throttle_enabled) return; metaslab_group_allocator_t *mga = &mg->mg_allocator[allocator]; (void) zfs_refcount_add(&mga->mga_alloc_queue_depth, tag); } static void metaslab_group_increment_qdepth(metaslab_group_t *mg, int allocator) { metaslab_group_allocator_t *mga = &mg->mg_allocator[allocator]; uint64_t max = mg->mg_max_alloc_queue_depth; uint64_t cur = mga->mga_cur_max_alloc_queue_depth; while (cur < max) { if (atomic_cas_64(&mga->mga_cur_max_alloc_queue_depth, cur, cur + 1) == cur) { atomic_inc_64( &mg->mg_class->mc_alloc_max_slots[allocator]); return; } cur = mga->mga_cur_max_alloc_queue_depth; } } void metaslab_group_alloc_decrement(spa_t *spa, uint64_t vdev, void *tag, int flags, int allocator, boolean_t io_complete) { if (!(flags & METASLAB_ASYNC_ALLOC) || (flags & METASLAB_DONT_THROTTLE)) return; metaslab_group_t *mg = vdev_lookup_top(spa, vdev)->vdev_mg; if (!mg->mg_class->mc_alloc_throttle_enabled) return; metaslab_group_allocator_t *mga = &mg->mg_allocator[allocator]; (void) zfs_refcount_remove(&mga->mga_alloc_queue_depth, tag); if (io_complete) metaslab_group_increment_qdepth(mg, allocator); } void metaslab_group_alloc_verify(spa_t *spa, const blkptr_t *bp, void *tag, int allocator) { #ifdef ZFS_DEBUG const dva_t *dva = bp->blk_dva; int ndvas = BP_GET_NDVAS(bp); for (int d = 0; d < ndvas; d++) { uint64_t vdev = DVA_GET_VDEV(&dva[d]); metaslab_group_t *mg = vdev_lookup_top(spa, vdev)->vdev_mg; metaslab_group_allocator_t *mga = &mg->mg_allocator[allocator]; VERIFY(zfs_refcount_not_held(&mga->mga_alloc_queue_depth, tag)); } #endif } static uint64_t metaslab_block_alloc(metaslab_t *msp, uint64_t size, uint64_t txg) { uint64_t start; range_tree_t *rt = msp->ms_allocatable; metaslab_class_t *mc = msp->ms_group->mg_class; ASSERT(MUTEX_HELD(&msp->ms_lock)); VERIFY(!msp->ms_condensing); VERIFY0(msp->ms_disabled); start = mc->mc_ops->msop_alloc(msp, size); if (start != -1ULL) { metaslab_group_t *mg = msp->ms_group; vdev_t *vd = mg->mg_vd; VERIFY0(P2PHASE(start, 1ULL << vd->vdev_ashift)); VERIFY0(P2PHASE(size, 1ULL << vd->vdev_ashift)); VERIFY3U(range_tree_space(rt) - size, <=, msp->ms_size); range_tree_remove(rt, start, size); range_tree_clear(msp->ms_trim, start, size); if (range_tree_is_empty(msp->ms_allocating[txg & TXG_MASK])) vdev_dirty(mg->mg_vd, VDD_METASLAB, msp, txg); range_tree_add(msp->ms_allocating[txg & TXG_MASK], start, size); msp->ms_allocating_total += size; /* Track the last successful allocation */ msp->ms_alloc_txg = txg; metaslab_verify_space(msp, txg); } /* * Now that we've attempted the allocation we need to update the * metaslab's maximum block size since it may have changed. */ msp->ms_max_size = metaslab_largest_allocatable(msp); return (start); } /* * Find the metaslab with the highest weight that is less than what we've * already tried. In the common case, this means that we will examine each * metaslab at most once. Note that concurrent callers could reorder metaslabs * by activation/passivation once we have dropped the mg_lock. If a metaslab is * activated by another thread, and we fail to allocate from the metaslab we * have selected, we may not try the newly-activated metaslab, and instead * activate another metaslab. This is not optimal, but generally does not cause * any problems (a possible exception being if every metaslab is completely full * except for the newly-activated metaslab which we fail to examine). */ static metaslab_t * find_valid_metaslab(metaslab_group_t *mg, uint64_t activation_weight, dva_t *dva, int d, boolean_t want_unique, uint64_t asize, int allocator, boolean_t try_hard, zio_alloc_list_t *zal, metaslab_t *search, boolean_t *was_active) { avl_index_t idx; avl_tree_t *t = &mg->mg_metaslab_tree; metaslab_t *msp = avl_find(t, search, &idx); if (msp == NULL) msp = avl_nearest(t, idx, AVL_AFTER); for (; msp != NULL; msp = AVL_NEXT(t, msp)) { int i; if (!metaslab_should_allocate(msp, asize, try_hard)) { metaslab_trace_add(zal, mg, msp, asize, d, TRACE_TOO_SMALL, allocator); continue; } /* * If the selected metaslab is condensing or disabled, * skip it. */ if (msp->ms_condensing || msp->ms_disabled > 0) continue; *was_active = msp->ms_allocator != -1; /* * If we're activating as primary, this is our first allocation * from this disk, so we don't need to check how close we are. * If the metaslab under consideration was already active, * we're getting desperate enough to steal another allocator's * metaslab, so we still don't care about distances. */ if (activation_weight == METASLAB_WEIGHT_PRIMARY || *was_active) break; for (i = 0; i < d; i++) { if (want_unique && !metaslab_is_unique(msp, &dva[i])) break; /* try another metaslab */ } if (i == d) break; } if (msp != NULL) { search->ms_weight = msp->ms_weight; search->ms_start = msp->ms_start + 1; search->ms_allocator = msp->ms_allocator; search->ms_primary = msp->ms_primary; } return (msp); } static void metaslab_active_mask_verify(metaslab_t *msp) { ASSERT(MUTEX_HELD(&msp->ms_lock)); if ((zfs_flags & ZFS_DEBUG_METASLAB_VERIFY) == 0) return; if ((msp->ms_weight & METASLAB_ACTIVE_MASK) == 0) return; if (msp->ms_weight & METASLAB_WEIGHT_PRIMARY) { VERIFY0(msp->ms_weight & METASLAB_WEIGHT_SECONDARY); VERIFY0(msp->ms_weight & METASLAB_WEIGHT_CLAIM); VERIFY3S(msp->ms_allocator, !=, -1); VERIFY(msp->ms_primary); return; } if (msp->ms_weight & METASLAB_WEIGHT_SECONDARY) { VERIFY0(msp->ms_weight & METASLAB_WEIGHT_PRIMARY); VERIFY0(msp->ms_weight & METASLAB_WEIGHT_CLAIM); VERIFY3S(msp->ms_allocator, !=, -1); VERIFY(!msp->ms_primary); return; } if (msp->ms_weight & METASLAB_WEIGHT_CLAIM) { VERIFY0(msp->ms_weight & METASLAB_WEIGHT_PRIMARY); VERIFY0(msp->ms_weight & METASLAB_WEIGHT_SECONDARY); VERIFY3S(msp->ms_allocator, ==, -1); return; } } /* ARGSUSED */ static uint64_t metaslab_group_alloc_normal(metaslab_group_t *mg, zio_alloc_list_t *zal, uint64_t asize, uint64_t txg, boolean_t want_unique, dva_t *dva, int d, int allocator, boolean_t try_hard) { metaslab_t *msp = NULL; uint64_t offset = -1ULL; uint64_t activation_weight = METASLAB_WEIGHT_PRIMARY; for (int i = 0; i < d; i++) { if (activation_weight == METASLAB_WEIGHT_PRIMARY && DVA_GET_VDEV(&dva[i]) == mg->mg_vd->vdev_id) { activation_weight = METASLAB_WEIGHT_SECONDARY; } else if (activation_weight == METASLAB_WEIGHT_SECONDARY && DVA_GET_VDEV(&dva[i]) == mg->mg_vd->vdev_id) { activation_weight = METASLAB_WEIGHT_CLAIM; break; } } /* * If we don't have enough metaslabs active to fill the entire array, we * just use the 0th slot. */ if (mg->mg_ms_ready < mg->mg_allocators * 3) allocator = 0; metaslab_group_allocator_t *mga = &mg->mg_allocator[allocator]; ASSERT3U(mg->mg_vd->vdev_ms_count, >=, 2); metaslab_t *search = kmem_alloc(sizeof (*search), KM_SLEEP); search->ms_weight = UINT64_MAX; search->ms_start = 0; /* * At the end of the metaslab tree are the already-active metaslabs, * first the primaries, then the secondaries. When we resume searching * through the tree, we need to consider ms_allocator and ms_primary so * we start in the location right after where we left off, and don't * accidentally loop forever considering the same metaslabs. */ search->ms_allocator = -1; search->ms_primary = B_TRUE; for (;;) { boolean_t was_active = B_FALSE; mutex_enter(&mg->mg_lock); if (activation_weight == METASLAB_WEIGHT_PRIMARY && mga->mga_primary != NULL) { msp = mga->mga_primary; /* * Even though we don't hold the ms_lock for the * primary metaslab, those fields should not * change while we hold the mg_lock. Thus it is * safe to make assertions on them. */ ASSERT(msp->ms_primary); ASSERT3S(msp->ms_allocator, ==, allocator); ASSERT(msp->ms_loaded); was_active = B_TRUE; ASSERT(msp->ms_weight & METASLAB_ACTIVE_MASK); } else if (activation_weight == METASLAB_WEIGHT_SECONDARY && mga->mga_secondary != NULL) { msp = mga->mga_secondary; /* * See comment above about the similar assertions * for the primary metaslab. */ ASSERT(!msp->ms_primary); ASSERT3S(msp->ms_allocator, ==, allocator); ASSERT(msp->ms_loaded); was_active = B_TRUE; ASSERT(msp->ms_weight & METASLAB_ACTIVE_MASK); } else { msp = find_valid_metaslab(mg, activation_weight, dva, d, want_unique, asize, allocator, try_hard, zal, search, &was_active); } mutex_exit(&mg->mg_lock); if (msp == NULL) { kmem_free(search, sizeof (*search)); return (-1ULL); } mutex_enter(&msp->ms_lock); metaslab_active_mask_verify(msp); /* * This code is disabled out because of issues with * tracepoints in non-gpl kernel modules. */ #if 0 DTRACE_PROBE3(ms__activation__attempt, metaslab_t *, msp, uint64_t, activation_weight, boolean_t, was_active); #endif /* * Ensure that the metaslab we have selected is still * capable of handling our request. It's possible that * another thread may have changed the weight while we * were blocked on the metaslab lock. We check the * active status first to see if we need to set_selected_txg * a new metaslab. */ if (was_active && !(msp->ms_weight & METASLAB_ACTIVE_MASK)) { ASSERT3S(msp->ms_allocator, ==, -1); mutex_exit(&msp->ms_lock); continue; } /* * If the metaslab was activated for another allocator * while we were waiting in the ms_lock above, or it's * a primary and we're seeking a secondary (or vice versa), * we go back and select a new metaslab. */ if (!was_active && (msp->ms_weight & METASLAB_ACTIVE_MASK) && (msp->ms_allocator != -1) && (msp->ms_allocator != allocator || ((activation_weight == METASLAB_WEIGHT_PRIMARY) != msp->ms_primary))) { ASSERT(msp->ms_loaded); ASSERT((msp->ms_weight & METASLAB_WEIGHT_CLAIM) || msp->ms_allocator != -1); mutex_exit(&msp->ms_lock); continue; } /* * This metaslab was used for claiming regions allocated * by the ZIL during pool import. Once these regions are * claimed we don't need to keep the CLAIM bit set * anymore. Passivate this metaslab to zero its activation * mask. */ if (msp->ms_weight & METASLAB_WEIGHT_CLAIM && activation_weight != METASLAB_WEIGHT_CLAIM) { ASSERT(msp->ms_loaded); ASSERT3S(msp->ms_allocator, ==, -1); metaslab_passivate(msp, msp->ms_weight & ~METASLAB_WEIGHT_CLAIM); mutex_exit(&msp->ms_lock); continue; } metaslab_set_selected_txg(msp, txg); int activation_error = metaslab_activate(msp, allocator, activation_weight); metaslab_active_mask_verify(msp); /* * If the metaslab was activated by another thread for * another allocator or activation_weight (EBUSY), or it * failed because another metaslab was assigned as primary * for this allocator (EEXIST) we continue using this * metaslab for our allocation, rather than going on to a * worse metaslab (we waited for that metaslab to be loaded * after all). * * If the activation failed due to an I/O error or ENOSPC we * skip to the next metaslab. */ boolean_t activated; if (activation_error == 0) { activated = B_TRUE; } else if (activation_error == EBUSY || activation_error == EEXIST) { activated = B_FALSE; } else { mutex_exit(&msp->ms_lock); continue; } ASSERT(msp->ms_loaded); /* * Now that we have the lock, recheck to see if we should * continue to use this metaslab for this allocation. The * the metaslab is now loaded so metaslab_should_allocate() * can accurately determine if the allocation attempt should * proceed. */ if (!metaslab_should_allocate(msp, asize, try_hard)) { /* Passivate this metaslab and select a new one. */ metaslab_trace_add(zal, mg, msp, asize, d, TRACE_TOO_SMALL, allocator); goto next; } /* * If this metaslab is currently condensing then pick again * as we can't manipulate this metaslab until it's committed * to disk. If this metaslab is being initialized, we shouldn't * allocate from it since the allocated region might be * overwritten after allocation. */ if (msp->ms_condensing) { metaslab_trace_add(zal, mg, msp, asize, d, TRACE_CONDENSING, allocator); if (activated) { metaslab_passivate(msp, msp->ms_weight & ~METASLAB_ACTIVE_MASK); } mutex_exit(&msp->ms_lock); continue; } else if (msp->ms_disabled > 0) { metaslab_trace_add(zal, mg, msp, asize, d, TRACE_DISABLED, allocator); if (activated) { metaslab_passivate(msp, msp->ms_weight & ~METASLAB_ACTIVE_MASK); } mutex_exit(&msp->ms_lock); continue; } offset = metaslab_block_alloc(msp, asize, txg); metaslab_trace_add(zal, mg, msp, asize, d, offset, allocator); if (offset != -1ULL) { /* Proactively passivate the metaslab, if needed */ if (activated) metaslab_segment_may_passivate(msp); break; } next: ASSERT(msp->ms_loaded); /* * This code is disabled out because of issues with * tracepoints in non-gpl kernel modules. */ #if 0 DTRACE_PROBE2(ms__alloc__failure, metaslab_t *, msp, uint64_t, asize); #endif /* * We were unable to allocate from this metaslab so determine * a new weight for this metaslab. Now that we have loaded * the metaslab we can provide a better hint to the metaslab * selector. * * For space-based metaslabs, we use the maximum block size. * This information is only available when the metaslab * is loaded and is more accurate than the generic free * space weight that was calculated by metaslab_weight(). * This information allows us to quickly compare the maximum * available allocation in the metaslab to the allocation * size being requested. * * For segment-based metaslabs, determine the new weight * based on the highest bucket in the range tree. We * explicitly use the loaded segment weight (i.e. the range * tree histogram) since it contains the space that is * currently available for allocation and is accurate * even within a sync pass. */ uint64_t weight; if (WEIGHT_IS_SPACEBASED(msp->ms_weight)) { weight = metaslab_largest_allocatable(msp); WEIGHT_SET_SPACEBASED(weight); } else { weight = metaslab_weight_from_range_tree(msp); } if (activated) { metaslab_passivate(msp, weight); } else { /* * For the case where we use the metaslab that is * active for another allocator we want to make * sure that we retain the activation mask. * * Note that we could attempt to use something like * metaslab_recalculate_weight_and_sort() that * retains the activation mask here. That function * uses metaslab_weight() to set the weight though * which is not as accurate as the calculations * above. */ weight |= msp->ms_weight & METASLAB_ACTIVE_MASK; metaslab_group_sort(mg, msp, weight); } metaslab_active_mask_verify(msp); /* * We have just failed an allocation attempt, check * that metaslab_should_allocate() agrees. Otherwise, * we may end up in an infinite loop retrying the same * metaslab. */ ASSERT(!metaslab_should_allocate(msp, asize, try_hard)); mutex_exit(&msp->ms_lock); } mutex_exit(&msp->ms_lock); kmem_free(search, sizeof (*search)); return (offset); } static uint64_t metaslab_group_alloc(metaslab_group_t *mg, zio_alloc_list_t *zal, uint64_t asize, uint64_t txg, boolean_t want_unique, dva_t *dva, int d, int allocator, boolean_t try_hard) { uint64_t offset; ASSERT(mg->mg_initialized); offset = metaslab_group_alloc_normal(mg, zal, asize, txg, want_unique, dva, d, allocator, try_hard); mutex_enter(&mg->mg_lock); if (offset == -1ULL) { mg->mg_failed_allocations++; metaslab_trace_add(zal, mg, NULL, asize, d, TRACE_GROUP_FAILURE, allocator); if (asize == SPA_GANGBLOCKSIZE) { /* * This metaslab group was unable to allocate * the minimum gang block size so it must be out of * space. We must notify the allocation throttle * to start skipping allocation attempts to this * metaslab group until more space becomes available. * Note: this failure cannot be caused by the * allocation throttle since the allocation throttle * is only responsible for skipping devices and * not failing block allocations. */ mg->mg_no_free_space = B_TRUE; } } mg->mg_allocations++; mutex_exit(&mg->mg_lock); return (offset); } /* * Allocate a block for the specified i/o. */ int metaslab_alloc_dva(spa_t *spa, metaslab_class_t *mc, uint64_t psize, dva_t *dva, int d, dva_t *hintdva, uint64_t txg, int flags, zio_alloc_list_t *zal, int allocator) { metaslab_group_t *mg, *fast_mg, *rotor; vdev_t *vd; boolean_t try_hard = B_FALSE; ASSERT(!DVA_IS_VALID(&dva[d])); /* * For testing, make some blocks above a certain size be gang blocks. * This will result in more split blocks when using device removal, * and a large number of split blocks coupled with ztest-induced * damage can result in extremely long reconstruction times. This * will also test spilling from special to normal. */ if (psize >= metaslab_force_ganging && (spa_get_random(100) < 3)) { metaslab_trace_add(zal, NULL, NULL, psize, d, TRACE_FORCE_GANG, allocator); return (SET_ERROR(ENOSPC)); } /* * Start at the rotor and loop through all mgs until we find something. * Note that there's no locking on mc_rotor or mc_aliquot because * nothing actually breaks if we miss a few updates -- we just won't * allocate quite as evenly. It all balances out over time. * * If we are doing ditto or log blocks, try to spread them across * consecutive vdevs. If we're forced to reuse a vdev before we've * allocated all of our ditto blocks, then try and spread them out on * that vdev as much as possible. If it turns out to not be possible, * gradually lower our standards until anything becomes acceptable. * Also, allocating on consecutive vdevs (as opposed to random vdevs) * gives us hope of containing our fault domains to something we're * able to reason about. Otherwise, any two top-level vdev failures * will guarantee the loss of data. With consecutive allocation, * only two adjacent top-level vdev failures will result in data loss. * * If we are doing gang blocks (hintdva is non-NULL), try to keep * ourselves on the same vdev as our gang block header. That * way, we can hope for locality in vdev_cache, plus it makes our * fault domains something tractable. */ if (hintdva) { vd = vdev_lookup_top(spa, DVA_GET_VDEV(&hintdva[d])); /* * It's possible the vdev we're using as the hint no * longer exists or its mg has been closed (e.g. by * device removal). Consult the rotor when * all else fails. */ if (vd != NULL && vd->vdev_mg != NULL) { mg = vd->vdev_mg; if (flags & METASLAB_HINTBP_AVOID && mg->mg_next != NULL) mg = mg->mg_next; } else { mg = mc->mc_rotor; } } else if (d != 0) { vd = vdev_lookup_top(spa, DVA_GET_VDEV(&dva[d - 1])); mg = vd->vdev_mg->mg_next; } else if (flags & METASLAB_FASTWRITE) { mg = fast_mg = mc->mc_rotor; do { if (fast_mg->mg_vd->vdev_pending_fastwrite < mg->mg_vd->vdev_pending_fastwrite) mg = fast_mg; } while ((fast_mg = fast_mg->mg_next) != mc->mc_rotor); } else { ASSERT(mc->mc_rotor != NULL); mg = mc->mc_rotor; } /* * If the hint put us into the wrong metaslab class, or into a * metaslab group that has been passivated, just follow the rotor. */ if (mg->mg_class != mc || mg->mg_activation_count <= 0) mg = mc->mc_rotor; rotor = mg; top: do { boolean_t allocatable; ASSERT(mg->mg_activation_count == 1); vd = mg->mg_vd; /* * Don't allocate from faulted devices. */ if (try_hard) { spa_config_enter(spa, SCL_ZIO, FTAG, RW_READER); allocatable = vdev_allocatable(vd); spa_config_exit(spa, SCL_ZIO, FTAG); } else { allocatable = vdev_allocatable(vd); } /* * Determine if the selected metaslab group is eligible * for allocations. If we're ganging then don't allow * this metaslab group to skip allocations since that would * inadvertently return ENOSPC and suspend the pool * even though space is still available. */ if (allocatable && !GANG_ALLOCATION(flags) && !try_hard) { allocatable = metaslab_group_allocatable(mg, rotor, psize, allocator, d); } if (!allocatable) { metaslab_trace_add(zal, mg, NULL, psize, d, TRACE_NOT_ALLOCATABLE, allocator); goto next; } ASSERT(mg->mg_initialized); /* * Avoid writing single-copy data to a failing, * non-redundant vdev, unless we've already tried all * other vdevs. */ if ((vd->vdev_stat.vs_write_errors > 0 || vd->vdev_state < VDEV_STATE_HEALTHY) && d == 0 && !try_hard && vd->vdev_children == 0) { metaslab_trace_add(zal, mg, NULL, psize, d, TRACE_VDEV_ERROR, allocator); goto next; } ASSERT(mg->mg_class == mc); uint64_t asize = vdev_psize_to_asize(vd, psize); ASSERT(P2PHASE(asize, 1ULL << vd->vdev_ashift) == 0); /* * If we don't need to try hard, then require that the * block be on a different metaslab from any other DVAs * in this BP (unique=true). If we are trying hard, then * allow any metaslab to be used (unique=false). */ uint64_t offset = metaslab_group_alloc(mg, zal, asize, txg, !try_hard, dva, d, allocator, try_hard); if (offset != -1ULL) { /* * If we've just selected this metaslab group, * figure out whether the corresponding vdev is * over- or under-used relative to the pool, * and set an allocation bias to even it out. * * Bias is also used to compensate for unequally * sized vdevs so that space is allocated fairly. */ if (mc->mc_aliquot == 0 && metaslab_bias_enabled) { vdev_stat_t *vs = &vd->vdev_stat; int64_t vs_free = vs->vs_space - vs->vs_alloc; int64_t mc_free = mc->mc_space - mc->mc_alloc; int64_t ratio; /* * Calculate how much more or less we should * try to allocate from this device during * this iteration around the rotor. * * This basically introduces a zero-centered * bias towards the devices with the most * free space, while compensating for vdev * size differences. * * Examples: * vdev V1 = 16M/128M * vdev V2 = 16M/128M * ratio(V1) = 100% ratio(V2) = 100% * * vdev V1 = 16M/128M * vdev V2 = 64M/128M * ratio(V1) = 127% ratio(V2) = 72% * * vdev V1 = 16M/128M * vdev V2 = 64M/512M * ratio(V1) = 40% ratio(V2) = 160% */ ratio = (vs_free * mc->mc_alloc_groups * 100) / (mc_free + 1); mg->mg_bias = ((ratio - 100) * (int64_t)mg->mg_aliquot) / 100; } else if (!metaslab_bias_enabled) { mg->mg_bias = 0; } if ((flags & METASLAB_FASTWRITE) || atomic_add_64_nv(&mc->mc_aliquot, asize) >= mg->mg_aliquot + mg->mg_bias) { mc->mc_rotor = mg->mg_next; mc->mc_aliquot = 0; } DVA_SET_VDEV(&dva[d], vd->vdev_id); DVA_SET_OFFSET(&dva[d], offset); DVA_SET_GANG(&dva[d], ((flags & METASLAB_GANG_HEADER) ? 1 : 0)); DVA_SET_ASIZE(&dva[d], asize); if (flags & METASLAB_FASTWRITE) { atomic_add_64(&vd->vdev_pending_fastwrite, psize); } return (0); } next: mc->mc_rotor = mg->mg_next; mc->mc_aliquot = 0; } while ((mg = mg->mg_next) != rotor); /* * If we haven't tried hard, do so now. */ if (!try_hard) { try_hard = B_TRUE; goto top; } bzero(&dva[d], sizeof (dva_t)); metaslab_trace_add(zal, rotor, NULL, psize, d, TRACE_ENOSPC, allocator); return (SET_ERROR(ENOSPC)); } void metaslab_free_concrete(vdev_t *vd, uint64_t offset, uint64_t asize, boolean_t checkpoint) { metaslab_t *msp; spa_t *spa = vd->vdev_spa; ASSERT(vdev_is_concrete(vd)); ASSERT3U(spa_config_held(spa, SCL_ALL, RW_READER), !=, 0); ASSERT3U(offset >> vd->vdev_ms_shift, <, vd->vdev_ms_count); msp = vd->vdev_ms[offset >> vd->vdev_ms_shift]; VERIFY(!msp->ms_condensing); VERIFY3U(offset, >=, msp->ms_start); VERIFY3U(offset + asize, <=, msp->ms_start + msp->ms_size); VERIFY0(P2PHASE(offset, 1ULL << vd->vdev_ashift)); VERIFY0(P2PHASE(asize, 1ULL << vd->vdev_ashift)); metaslab_check_free_impl(vd, offset, asize); mutex_enter(&msp->ms_lock); if (range_tree_is_empty(msp->ms_freeing) && range_tree_is_empty(msp->ms_checkpointing)) { vdev_dirty(vd, VDD_METASLAB, msp, spa_syncing_txg(spa)); } if (checkpoint) { ASSERT(spa_has_checkpoint(spa)); range_tree_add(msp->ms_checkpointing, offset, asize); } else { range_tree_add(msp->ms_freeing, offset, asize); } mutex_exit(&msp->ms_lock); } /* ARGSUSED */ void metaslab_free_impl_cb(uint64_t inner_offset, vdev_t *vd, uint64_t offset, uint64_t size, void *arg) { boolean_t *checkpoint = arg; ASSERT3P(checkpoint, !=, NULL); if (vd->vdev_ops->vdev_op_remap != NULL) vdev_indirect_mark_obsolete(vd, offset, size); else metaslab_free_impl(vd, offset, size, *checkpoint); } static void metaslab_free_impl(vdev_t *vd, uint64_t offset, uint64_t size, boolean_t checkpoint) { spa_t *spa = vd->vdev_spa; ASSERT3U(spa_config_held(spa, SCL_ALL, RW_READER), !=, 0); if (spa_syncing_txg(spa) > spa_freeze_txg(spa)) return; if (spa->spa_vdev_removal != NULL && spa->spa_vdev_removal->svr_vdev_id == vd->vdev_id && vdev_is_concrete(vd)) { /* * Note: we check if the vdev is concrete because when * we complete the removal, we first change the vdev to be * an indirect vdev (in open context), and then (in syncing * context) clear spa_vdev_removal. */ free_from_removing_vdev(vd, offset, size); } else if (vd->vdev_ops->vdev_op_remap != NULL) { vdev_indirect_mark_obsolete(vd, offset, size); vd->vdev_ops->vdev_op_remap(vd, offset, size, metaslab_free_impl_cb, &checkpoint); } else { metaslab_free_concrete(vd, offset, size, checkpoint); } } typedef struct remap_blkptr_cb_arg { blkptr_t *rbca_bp; spa_remap_cb_t rbca_cb; vdev_t *rbca_remap_vd; uint64_t rbca_remap_offset; void *rbca_cb_arg; } remap_blkptr_cb_arg_t; static void remap_blkptr_cb(uint64_t inner_offset, vdev_t *vd, uint64_t offset, uint64_t size, void *arg) { remap_blkptr_cb_arg_t *rbca = arg; blkptr_t *bp = rbca->rbca_bp; /* We can not remap split blocks. */ if (size != DVA_GET_ASIZE(&bp->blk_dva[0])) return; ASSERT0(inner_offset); if (rbca->rbca_cb != NULL) { /* * At this point we know that we are not handling split * blocks and we invoke the callback on the previous * vdev which must be indirect. */ ASSERT3P(rbca->rbca_remap_vd->vdev_ops, ==, &vdev_indirect_ops); rbca->rbca_cb(rbca->rbca_remap_vd->vdev_id, rbca->rbca_remap_offset, size, rbca->rbca_cb_arg); /* set up remap_blkptr_cb_arg for the next call */ rbca->rbca_remap_vd = vd; rbca->rbca_remap_offset = offset; } /* * The phys birth time is that of dva[0]. This ensures that we know * when each dva was written, so that resilver can determine which * blocks need to be scrubbed (i.e. those written during the time * the vdev was offline). It also ensures that the key used in * the ARC hash table is unique (i.e. dva[0] + phys_birth). If * we didn't change the phys_birth, a lookup in the ARC for a * remapped BP could find the data that was previously stored at * this vdev + offset. */ vdev_t *oldvd = vdev_lookup_top(vd->vdev_spa, DVA_GET_VDEV(&bp->blk_dva[0])); vdev_indirect_births_t *vib = oldvd->vdev_indirect_births; bp->blk_phys_birth = vdev_indirect_births_physbirth(vib, DVA_GET_OFFSET(&bp->blk_dva[0]), DVA_GET_ASIZE(&bp->blk_dva[0])); DVA_SET_VDEV(&bp->blk_dva[0], vd->vdev_id); DVA_SET_OFFSET(&bp->blk_dva[0], offset); } /* * If the block pointer contains any indirect DVAs, modify them to refer to * concrete DVAs. Note that this will sometimes not be possible, leaving * the indirect DVA in place. This happens if the indirect DVA spans multiple * segments in the mapping (i.e. it is a "split block"). * * If the BP was remapped, calls the callback on the original dva (note the * callback can be called multiple times if the original indirect DVA refers * to another indirect DVA, etc). * * Returns TRUE if the BP was remapped. */ boolean_t spa_remap_blkptr(spa_t *spa, blkptr_t *bp, spa_remap_cb_t callback, void *arg) { remap_blkptr_cb_arg_t rbca; if (!zfs_remap_blkptr_enable) return (B_FALSE); if (!spa_feature_is_enabled(spa, SPA_FEATURE_OBSOLETE_COUNTS)) return (B_FALSE); /* * Dedup BP's can not be remapped, because ddt_phys_select() depends * on DVA[0] being the same in the BP as in the DDT (dedup table). */ if (BP_GET_DEDUP(bp)) return (B_FALSE); /* * Gang blocks can not be remapped, because * zio_checksum_gang_verifier() depends on the DVA[0] that's in * the BP used to read the gang block header (GBH) being the same * as the DVA[0] that we allocated for the GBH. */ if (BP_IS_GANG(bp)) return (B_FALSE); /* * Embedded BP's have no DVA to remap. */ if (BP_GET_NDVAS(bp) < 1) return (B_FALSE); /* * Note: we only remap dva[0]. If we remapped other dvas, we * would no longer know what their phys birth txg is. */ dva_t *dva = &bp->blk_dva[0]; uint64_t offset = DVA_GET_OFFSET(dva); uint64_t size = DVA_GET_ASIZE(dva); vdev_t *vd = vdev_lookup_top(spa, DVA_GET_VDEV(dva)); if (vd->vdev_ops->vdev_op_remap == NULL) return (B_FALSE); rbca.rbca_bp = bp; rbca.rbca_cb = callback; rbca.rbca_remap_vd = vd; rbca.rbca_remap_offset = offset; rbca.rbca_cb_arg = arg; /* * remap_blkptr_cb() will be called in order for each level of * indirection, until a concrete vdev is reached or a split block is * encountered. old_vd and old_offset are updated within the callback * as we go from the one indirect vdev to the next one (either concrete * or indirect again) in that order. */ vd->vdev_ops->vdev_op_remap(vd, offset, size, remap_blkptr_cb, &rbca); /* Check if the DVA wasn't remapped because it is a split block */ if (DVA_GET_VDEV(&rbca.rbca_bp->blk_dva[0]) == vd->vdev_id) return (B_FALSE); return (B_TRUE); } /* * Undo the allocation of a DVA which happened in the given transaction group. */ void metaslab_unalloc_dva(spa_t *spa, const dva_t *dva, uint64_t txg) { metaslab_t *msp; vdev_t *vd; uint64_t vdev = DVA_GET_VDEV(dva); uint64_t offset = DVA_GET_OFFSET(dva); uint64_t size = DVA_GET_ASIZE(dva); ASSERT(DVA_IS_VALID(dva)); ASSERT3U(spa_config_held(spa, SCL_ALL, RW_READER), !=, 0); if (txg > spa_freeze_txg(spa)) return; if ((vd = vdev_lookup_top(spa, vdev)) == NULL || !DVA_IS_VALID(dva) || (offset >> vd->vdev_ms_shift) >= vd->vdev_ms_count) { zfs_panic_recover("metaslab_free_dva(): bad DVA %llu:%llu:%llu", (u_longlong_t)vdev, (u_longlong_t)offset, (u_longlong_t)size); return; } ASSERT(!vd->vdev_removing); ASSERT(vdev_is_concrete(vd)); ASSERT0(vd->vdev_indirect_config.vic_mapping_object); ASSERT3P(vd->vdev_indirect_mapping, ==, NULL); if (DVA_GET_GANG(dva)) size = vdev_psize_to_asize(vd, SPA_GANGBLOCKSIZE); msp = vd->vdev_ms[offset >> vd->vdev_ms_shift]; mutex_enter(&msp->ms_lock); range_tree_remove(msp->ms_allocating[txg & TXG_MASK], offset, size); msp->ms_allocating_total -= size; VERIFY(!msp->ms_condensing); VERIFY3U(offset, >=, msp->ms_start); VERIFY3U(offset + size, <=, msp->ms_start + msp->ms_size); VERIFY3U(range_tree_space(msp->ms_allocatable) + size, <=, msp->ms_size); VERIFY0(P2PHASE(offset, 1ULL << vd->vdev_ashift)); VERIFY0(P2PHASE(size, 1ULL << vd->vdev_ashift)); range_tree_add(msp->ms_allocatable, offset, size); mutex_exit(&msp->ms_lock); } /* * Free the block represented by the given DVA. */ void metaslab_free_dva(spa_t *spa, const dva_t *dva, boolean_t checkpoint) { uint64_t vdev = DVA_GET_VDEV(dva); uint64_t offset = DVA_GET_OFFSET(dva); uint64_t size = DVA_GET_ASIZE(dva); vdev_t *vd = vdev_lookup_top(spa, vdev); ASSERT(DVA_IS_VALID(dva)); ASSERT3U(spa_config_held(spa, SCL_ALL, RW_READER), !=, 0); if (DVA_GET_GANG(dva)) { size = vdev_psize_to_asize(vd, SPA_GANGBLOCKSIZE); } metaslab_free_impl(vd, offset, size, checkpoint); } /* * Reserve some allocation slots. The reservation system must be called * before we call into the allocator. If there aren't any available slots * then the I/O will be throttled until an I/O completes and its slots are * freed up. The function returns true if it was successful in placing * the reservation. */ boolean_t metaslab_class_throttle_reserve(metaslab_class_t *mc, int slots, int allocator, zio_t *zio, int flags) { uint64_t available_slots = 0; boolean_t slot_reserved = B_FALSE; uint64_t max = mc->mc_alloc_max_slots[allocator]; ASSERT(mc->mc_alloc_throttle_enabled); mutex_enter(&mc->mc_lock); uint64_t reserved_slots = zfs_refcount_count(&mc->mc_alloc_slots[allocator]); if (reserved_slots < max) available_slots = max - reserved_slots; if (slots <= available_slots || GANG_ALLOCATION(flags) || flags & METASLAB_MUST_RESERVE) { /* * We reserve the slots individually so that we can unreserve * them individually when an I/O completes. */ for (int d = 0; d < slots; d++) { reserved_slots = zfs_refcount_add(&mc->mc_alloc_slots[allocator], zio); } zio->io_flags |= ZIO_FLAG_IO_ALLOCATING; slot_reserved = B_TRUE; } mutex_exit(&mc->mc_lock); return (slot_reserved); } void metaslab_class_throttle_unreserve(metaslab_class_t *mc, int slots, int allocator, zio_t *zio) { ASSERT(mc->mc_alloc_throttle_enabled); mutex_enter(&mc->mc_lock); for (int d = 0; d < slots; d++) { (void) zfs_refcount_remove(&mc->mc_alloc_slots[allocator], zio); } mutex_exit(&mc->mc_lock); } static int metaslab_claim_concrete(vdev_t *vd, uint64_t offset, uint64_t size, uint64_t txg) { metaslab_t *msp; spa_t *spa = vd->vdev_spa; int error = 0; if (offset >> vd->vdev_ms_shift >= vd->vdev_ms_count) return (SET_ERROR(ENXIO)); ASSERT3P(vd->vdev_ms, !=, NULL); msp = vd->vdev_ms[offset >> vd->vdev_ms_shift]; mutex_enter(&msp->ms_lock); if ((txg != 0 && spa_writeable(spa)) || !msp->ms_loaded) { error = metaslab_activate(msp, 0, METASLAB_WEIGHT_CLAIM); if (error == EBUSY) { ASSERT(msp->ms_loaded); ASSERT(msp->ms_weight & METASLAB_ACTIVE_MASK); error = 0; } } if (error == 0 && !range_tree_contains(msp->ms_allocatable, offset, size)) error = SET_ERROR(ENOENT); if (error || txg == 0) { /* txg == 0 indicates dry run */ mutex_exit(&msp->ms_lock); return (error); } VERIFY(!msp->ms_condensing); VERIFY0(P2PHASE(offset, 1ULL << vd->vdev_ashift)); VERIFY0(P2PHASE(size, 1ULL << vd->vdev_ashift)); VERIFY3U(range_tree_space(msp->ms_allocatable) - size, <=, msp->ms_size); range_tree_remove(msp->ms_allocatable, offset, size); range_tree_clear(msp->ms_trim, offset, size); - if (spa_writeable(spa)) { /* don't dirty if we're zdb(1M) */ + if (spa_writeable(spa)) { /* don't dirty if we're zdb(8) */ metaslab_class_t *mc = msp->ms_group->mg_class; multilist_sublist_t *mls = multilist_sublist_lock_obj(mc->mc_metaslab_txg_list, msp); if (!multilist_link_active(&msp->ms_class_txg_node)) { msp->ms_selected_txg = txg; multilist_sublist_insert_head(mls, msp); } multilist_sublist_unlock(mls); if (range_tree_is_empty(msp->ms_allocating[txg & TXG_MASK])) vdev_dirty(vd, VDD_METASLAB, msp, txg); range_tree_add(msp->ms_allocating[txg & TXG_MASK], offset, size); msp->ms_allocating_total += size; } mutex_exit(&msp->ms_lock); return (0); } typedef struct metaslab_claim_cb_arg_t { uint64_t mcca_txg; int mcca_error; } metaslab_claim_cb_arg_t; /* ARGSUSED */ static void metaslab_claim_impl_cb(uint64_t inner_offset, vdev_t *vd, uint64_t offset, uint64_t size, void *arg) { metaslab_claim_cb_arg_t *mcca_arg = arg; if (mcca_arg->mcca_error == 0) { mcca_arg->mcca_error = metaslab_claim_concrete(vd, offset, size, mcca_arg->mcca_txg); } } int metaslab_claim_impl(vdev_t *vd, uint64_t offset, uint64_t size, uint64_t txg) { if (vd->vdev_ops->vdev_op_remap != NULL) { metaslab_claim_cb_arg_t arg; /* - * Only zdb(1M) can claim on indirect vdevs. This is used + * Only zdb(8) can claim on indirect vdevs. This is used * to detect leaks of mapped space (that are not accounted * for in the obsolete counts, spacemap, or bpobj). */ ASSERT(!spa_writeable(vd->vdev_spa)); arg.mcca_error = 0; arg.mcca_txg = txg; vd->vdev_ops->vdev_op_remap(vd, offset, size, metaslab_claim_impl_cb, &arg); if (arg.mcca_error == 0) { arg.mcca_error = metaslab_claim_concrete(vd, offset, size, txg); } return (arg.mcca_error); } else { return (metaslab_claim_concrete(vd, offset, size, txg)); } } /* * Intent log support: upon opening the pool after a crash, notify the SPA * of blocks that the intent log has allocated for immediate write, but * which are still considered free by the SPA because the last transaction * group didn't commit yet. */ static int metaslab_claim_dva(spa_t *spa, const dva_t *dva, uint64_t txg) { uint64_t vdev = DVA_GET_VDEV(dva); uint64_t offset = DVA_GET_OFFSET(dva); uint64_t size = DVA_GET_ASIZE(dva); vdev_t *vd; if ((vd = vdev_lookup_top(spa, vdev)) == NULL) { return (SET_ERROR(ENXIO)); } ASSERT(DVA_IS_VALID(dva)); if (DVA_GET_GANG(dva)) size = vdev_psize_to_asize(vd, SPA_GANGBLOCKSIZE); return (metaslab_claim_impl(vd, offset, size, txg)); } int metaslab_alloc(spa_t *spa, metaslab_class_t *mc, uint64_t psize, blkptr_t *bp, int ndvas, uint64_t txg, blkptr_t *hintbp, int flags, zio_alloc_list_t *zal, zio_t *zio, int allocator) { dva_t *dva = bp->blk_dva; dva_t *hintdva = (hintbp != NULL) ? hintbp->blk_dva : NULL; int error = 0; ASSERT(bp->blk_birth == 0); ASSERT(BP_PHYSICAL_BIRTH(bp) == 0); spa_config_enter(spa, SCL_ALLOC, FTAG, RW_READER); if (mc->mc_rotor == NULL) { /* no vdevs in this class */ spa_config_exit(spa, SCL_ALLOC, FTAG); return (SET_ERROR(ENOSPC)); } ASSERT(ndvas > 0 && ndvas <= spa_max_replication(spa)); ASSERT(BP_GET_NDVAS(bp) == 0); ASSERT(hintbp == NULL || ndvas <= BP_GET_NDVAS(hintbp)); ASSERT3P(zal, !=, NULL); for (int d = 0; d < ndvas; d++) { error = metaslab_alloc_dva(spa, mc, psize, dva, d, hintdva, txg, flags, zal, allocator); if (error != 0) { for (d--; d >= 0; d--) { metaslab_unalloc_dva(spa, &dva[d], txg); metaslab_group_alloc_decrement(spa, DVA_GET_VDEV(&dva[d]), zio, flags, allocator, B_FALSE); bzero(&dva[d], sizeof (dva_t)); } spa_config_exit(spa, SCL_ALLOC, FTAG); return (error); } else { /* * Update the metaslab group's queue depth * based on the newly allocated dva. */ metaslab_group_alloc_increment(spa, DVA_GET_VDEV(&dva[d]), zio, flags, allocator); } } ASSERT(error == 0); ASSERT(BP_GET_NDVAS(bp) == ndvas); spa_config_exit(spa, SCL_ALLOC, FTAG); BP_SET_BIRTH(bp, txg, 0); return (0); } void metaslab_free(spa_t *spa, const blkptr_t *bp, uint64_t txg, boolean_t now) { const dva_t *dva = bp->blk_dva; int ndvas = BP_GET_NDVAS(bp); ASSERT(!BP_IS_HOLE(bp)); ASSERT(!now || bp->blk_birth >= spa_syncing_txg(spa)); /* * If we have a checkpoint for the pool we need to make sure that * the blocks that we free that are part of the checkpoint won't be * reused until the checkpoint is discarded or we revert to it. * * The checkpoint flag is passed down the metaslab_free code path * and is set whenever we want to add a block to the checkpoint's * accounting. That is, we "checkpoint" blocks that existed at the * time the checkpoint was created and are therefore referenced by * the checkpointed uberblock. * * Note that, we don't checkpoint any blocks if the current * syncing txg <= spa_checkpoint_txg. We want these frees to sync * normally as they will be referenced by the checkpointed uberblock. */ boolean_t checkpoint = B_FALSE; if (bp->blk_birth <= spa->spa_checkpoint_txg && spa_syncing_txg(spa) > spa->spa_checkpoint_txg) { /* * At this point, if the block is part of the checkpoint * there is no way it was created in the current txg. */ ASSERT(!now); ASSERT3U(spa_syncing_txg(spa), ==, txg); checkpoint = B_TRUE; } spa_config_enter(spa, SCL_FREE, FTAG, RW_READER); for (int d = 0; d < ndvas; d++) { if (now) { metaslab_unalloc_dva(spa, &dva[d], txg); } else { ASSERT3U(txg, ==, spa_syncing_txg(spa)); metaslab_free_dva(spa, &dva[d], checkpoint); } } spa_config_exit(spa, SCL_FREE, FTAG); } int metaslab_claim(spa_t *spa, const blkptr_t *bp, uint64_t txg) { const dva_t *dva = bp->blk_dva; int ndvas = BP_GET_NDVAS(bp); int error = 0; ASSERT(!BP_IS_HOLE(bp)); if (txg != 0) { /* * First do a dry run to make sure all DVAs are claimable, * so we don't have to unwind from partial failures below. */ if ((error = metaslab_claim(spa, bp, 0)) != 0) return (error); } spa_config_enter(spa, SCL_ALLOC, FTAG, RW_READER); for (int d = 0; d < ndvas; d++) { error = metaslab_claim_dva(spa, &dva[d], txg); if (error != 0) break; } spa_config_exit(spa, SCL_ALLOC, FTAG); ASSERT(error == 0 || txg == 0); return (error); } void metaslab_fastwrite_mark(spa_t *spa, const blkptr_t *bp) { const dva_t *dva = bp->blk_dva; int ndvas = BP_GET_NDVAS(bp); uint64_t psize = BP_GET_PSIZE(bp); int d; vdev_t *vd; ASSERT(!BP_IS_HOLE(bp)); ASSERT(!BP_IS_EMBEDDED(bp)); ASSERT(psize > 0); spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); for (d = 0; d < ndvas; d++) { if ((vd = vdev_lookup_top(spa, DVA_GET_VDEV(&dva[d]))) == NULL) continue; atomic_add_64(&vd->vdev_pending_fastwrite, psize); } spa_config_exit(spa, SCL_VDEV, FTAG); } void metaslab_fastwrite_unmark(spa_t *spa, const blkptr_t *bp) { const dva_t *dva = bp->blk_dva; int ndvas = BP_GET_NDVAS(bp); uint64_t psize = BP_GET_PSIZE(bp); int d; vdev_t *vd; ASSERT(!BP_IS_HOLE(bp)); ASSERT(!BP_IS_EMBEDDED(bp)); ASSERT(psize > 0); spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); for (d = 0; d < ndvas; d++) { if ((vd = vdev_lookup_top(spa, DVA_GET_VDEV(&dva[d]))) == NULL) continue; ASSERT3U(vd->vdev_pending_fastwrite, >=, psize); atomic_sub_64(&vd->vdev_pending_fastwrite, psize); } spa_config_exit(spa, SCL_VDEV, FTAG); } /* ARGSUSED */ static void metaslab_check_free_impl_cb(uint64_t inner, vdev_t *vd, uint64_t offset, uint64_t size, void *arg) { if (vd->vdev_ops == &vdev_indirect_ops) return; metaslab_check_free_impl(vd, offset, size); } static void metaslab_check_free_impl(vdev_t *vd, uint64_t offset, uint64_t size) { metaslab_t *msp; spa_t *spa __maybe_unused = vd->vdev_spa; if ((zfs_flags & ZFS_DEBUG_ZIO_FREE) == 0) return; if (vd->vdev_ops->vdev_op_remap != NULL) { vd->vdev_ops->vdev_op_remap(vd, offset, size, metaslab_check_free_impl_cb, NULL); return; } ASSERT(vdev_is_concrete(vd)); ASSERT3U(offset >> vd->vdev_ms_shift, <, vd->vdev_ms_count); ASSERT3U(spa_config_held(spa, SCL_ALL, RW_READER), !=, 0); msp = vd->vdev_ms[offset >> vd->vdev_ms_shift]; mutex_enter(&msp->ms_lock); if (msp->ms_loaded) { range_tree_verify_not_present(msp->ms_allocatable, offset, size); } /* * Check all segments that currently exist in the freeing pipeline. * * It would intuitively make sense to also check the current allocating * tree since metaslab_unalloc_dva() exists for extents that are * allocated and freed in the same sync pass within the same txg. * Unfortunately there are places (e.g. the ZIL) where we allocate a * segment but then we free part of it within the same txg * [see zil_sync()]. Thus, we don't call range_tree_verify() in the * current allocating tree. */ range_tree_verify_not_present(msp->ms_freeing, offset, size); range_tree_verify_not_present(msp->ms_checkpointing, offset, size); range_tree_verify_not_present(msp->ms_freed, offset, size); for (int j = 0; j < TXG_DEFER_SIZE; j++) range_tree_verify_not_present(msp->ms_defer[j], offset, size); range_tree_verify_not_present(msp->ms_trim, offset, size); mutex_exit(&msp->ms_lock); } void metaslab_check_free(spa_t *spa, const blkptr_t *bp) { if ((zfs_flags & ZFS_DEBUG_ZIO_FREE) == 0) return; spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); for (int i = 0; i < BP_GET_NDVAS(bp); i++) { uint64_t vdev = DVA_GET_VDEV(&bp->blk_dva[i]); vdev_t *vd = vdev_lookup_top(spa, vdev); uint64_t offset = DVA_GET_OFFSET(&bp->blk_dva[i]); uint64_t size = DVA_GET_ASIZE(&bp->blk_dva[i]); if (DVA_GET_GANG(&bp->blk_dva[i])) size = vdev_psize_to_asize(vd, SPA_GANGBLOCKSIZE); ASSERT3P(vd, !=, NULL); metaslab_check_free_impl(vd, offset, size); } spa_config_exit(spa, SCL_VDEV, FTAG); } static void metaslab_group_disable_wait(metaslab_group_t *mg) { ASSERT(MUTEX_HELD(&mg->mg_ms_disabled_lock)); while (mg->mg_disabled_updating) { cv_wait(&mg->mg_ms_disabled_cv, &mg->mg_ms_disabled_lock); } } static void metaslab_group_disabled_increment(metaslab_group_t *mg) { ASSERT(MUTEX_HELD(&mg->mg_ms_disabled_lock)); ASSERT(mg->mg_disabled_updating); while (mg->mg_ms_disabled >= max_disabled_ms) { cv_wait(&mg->mg_ms_disabled_cv, &mg->mg_ms_disabled_lock); } mg->mg_ms_disabled++; ASSERT3U(mg->mg_ms_disabled, <=, max_disabled_ms); } /* * Mark the metaslab as disabled to prevent any allocations on this metaslab. * We must also track how many metaslabs are currently disabled within a * metaslab group and limit them to prevent allocation failures from * occurring because all metaslabs are disabled. */ void metaslab_disable(metaslab_t *msp) { ASSERT(!MUTEX_HELD(&msp->ms_lock)); metaslab_group_t *mg = msp->ms_group; mutex_enter(&mg->mg_ms_disabled_lock); /* * To keep an accurate count of how many threads have disabled * a specific metaslab group, we only allow one thread to mark * the metaslab group at a time. This ensures that the value of * ms_disabled will be accurate when we decide to mark a metaslab * group as disabled. To do this we force all other threads * to wait till the metaslab's mg_disabled_updating flag is no * longer set. */ metaslab_group_disable_wait(mg); mg->mg_disabled_updating = B_TRUE; if (msp->ms_disabled == 0) { metaslab_group_disabled_increment(mg); } mutex_enter(&msp->ms_lock); msp->ms_disabled++; mutex_exit(&msp->ms_lock); mg->mg_disabled_updating = B_FALSE; cv_broadcast(&mg->mg_ms_disabled_cv); mutex_exit(&mg->mg_ms_disabled_lock); } void metaslab_enable(metaslab_t *msp, boolean_t sync, boolean_t unload) { metaslab_group_t *mg = msp->ms_group; spa_t *spa = mg->mg_vd->vdev_spa; /* * Wait for the outstanding IO to be synced to prevent newly * allocated blocks from being overwritten. This used by * initialize and TRIM which are modifying unallocated space. */ if (sync) txg_wait_synced(spa_get_dsl(spa), 0); mutex_enter(&mg->mg_ms_disabled_lock); mutex_enter(&msp->ms_lock); if (--msp->ms_disabled == 0) { mg->mg_ms_disabled--; cv_broadcast(&mg->mg_ms_disabled_cv); if (unload) metaslab_unload(msp); } mutex_exit(&msp->ms_lock); mutex_exit(&mg->mg_ms_disabled_lock); } static void metaslab_update_ondisk_flush_data(metaslab_t *ms, dmu_tx_t *tx) { vdev_t *vd = ms->ms_group->mg_vd; spa_t *spa = vd->vdev_spa; objset_t *mos = spa_meta_objset(spa); ASSERT(spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP)); metaslab_unflushed_phys_t entry = { .msp_unflushed_txg = metaslab_unflushed_txg(ms), }; uint64_t entry_size = sizeof (entry); uint64_t entry_offset = ms->ms_id * entry_size; uint64_t object = 0; int err = zap_lookup(mos, vd->vdev_top_zap, VDEV_TOP_ZAP_MS_UNFLUSHED_PHYS_TXGS, sizeof (uint64_t), 1, &object); if (err == ENOENT) { object = dmu_object_alloc(mos, DMU_OTN_UINT64_METADATA, SPA_OLD_MAXBLOCKSIZE, DMU_OT_NONE, 0, tx); VERIFY0(zap_add(mos, vd->vdev_top_zap, VDEV_TOP_ZAP_MS_UNFLUSHED_PHYS_TXGS, sizeof (uint64_t), 1, &object, tx)); } else { VERIFY0(err); } dmu_write(spa_meta_objset(spa), object, entry_offset, entry_size, &entry, tx); } void metaslab_set_unflushed_txg(metaslab_t *ms, uint64_t txg, dmu_tx_t *tx) { spa_t *spa = ms->ms_group->mg_vd->vdev_spa; if (!spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP)) return; ms->ms_unflushed_txg = txg; metaslab_update_ondisk_flush_data(ms, tx); } uint64_t metaslab_unflushed_txg(metaslab_t *ms) { return (ms->ms_unflushed_txg); } ZFS_MODULE_PARAM(zfs_metaslab, metaslab_, aliquot, ULONG, ZMOD_RW, "Allocation granularity (a.k.a. stripe size)"); ZFS_MODULE_PARAM(zfs_metaslab, metaslab_, debug_load, INT, ZMOD_RW, "Load all metaslabs when pool is first opened"); ZFS_MODULE_PARAM(zfs_metaslab, metaslab_, debug_unload, INT, ZMOD_RW, "Prevent metaslabs from being unloaded"); ZFS_MODULE_PARAM(zfs_metaslab, metaslab_, preload_enabled, INT, ZMOD_RW, "Preload potential metaslabs during reassessment"); ZFS_MODULE_PARAM(zfs_metaslab, metaslab_, unload_delay, INT, ZMOD_RW, "Delay in txgs after metaslab was last used before unloading"); ZFS_MODULE_PARAM(zfs_metaslab, metaslab_, unload_delay_ms, INT, ZMOD_RW, "Delay in milliseconds after metaslab was last used before unloading"); /* BEGIN CSTYLED */ ZFS_MODULE_PARAM(zfs_mg, zfs_mg_, noalloc_threshold, INT, ZMOD_RW, "Percentage of metaslab group size that should be free to make it " "eligible for allocation"); ZFS_MODULE_PARAM(zfs_mg, zfs_mg_, fragmentation_threshold, INT, ZMOD_RW, "Percentage of metaslab group size that should be considered eligible " "for allocations unless all metaslab groups within the metaslab class " "have also crossed this threshold"); ZFS_MODULE_PARAM(zfs_metaslab, zfs_metaslab_, fragmentation_threshold, INT, ZMOD_RW, "Fragmentation for metaslab to allow allocation"); ZFS_MODULE_PARAM(zfs_metaslab, metaslab_, fragmentation_factor_enabled, INT, ZMOD_RW, "Use the fragmentation metric to prefer less fragmented metaslabs"); /* END CSTYLED */ ZFS_MODULE_PARAM(zfs_metaslab, metaslab_, lba_weighting_enabled, INT, ZMOD_RW, "Prefer metaslabs with lower LBAs"); ZFS_MODULE_PARAM(zfs_metaslab, metaslab_, bias_enabled, INT, ZMOD_RW, "Enable metaslab group biasing"); ZFS_MODULE_PARAM(zfs_metaslab, zfs_metaslab_, segment_weight_enabled, INT, ZMOD_RW, "Enable segment-based metaslab selection"); ZFS_MODULE_PARAM(zfs_metaslab, zfs_metaslab_, switch_threshold, INT, ZMOD_RW, "Segment-based metaslab selection maximum buckets before switching"); ZFS_MODULE_PARAM(zfs_metaslab, metaslab_, force_ganging, ULONG, ZMOD_RW, "Blocks larger than this size are forced to be gang blocks"); ZFS_MODULE_PARAM(zfs_metaslab, metaslab_, df_max_search, INT, ZMOD_RW, "Max distance (bytes) to search forward before using size tree"); ZFS_MODULE_PARAM(zfs_metaslab, metaslab_, df_use_largest_segment, INT, ZMOD_RW, "When looking in size tree, use largest segment instead of exact fit"); ZFS_MODULE_PARAM(zfs_metaslab, zfs_metaslab_, max_size_cache_sec, ULONG, ZMOD_RW, "How long to trust the cached max chunk size of a metaslab"); ZFS_MODULE_PARAM(zfs_metaslab, zfs_metaslab_, mem_limit, INT, ZMOD_RW, "Percentage of memory that can be used to store metaslab range trees"); diff --git a/module/zfs/spa_history.c b/module/zfs/spa_history.c index 2ab58815400a..2939c0366504 100644 --- a/module/zfs/spa_history.c +++ b/module/zfs/spa_history.c @@ -1,628 +1,628 @@ /* * 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) 2011, 2018 by Delphix. All rights reserved. * Copyright (c) 2014 Integros [integros.com] * Copyright 2017 Joyent, Inc. */ #include #include #include #include #include #include #include #include #include #include #include #include "zfs_comutil.h" #include "zfs_gitrev.h" #ifdef _KERNEL #include #endif /* * Routines to manage the on-disk history log. * * The history log is stored as a dmu object containing * tuples. * * Where "record nvlist" is an nvlist containing uint64_ts and strings, and * "packed record length" is the packed length of the "record nvlist" stored * as a little endian uint64_t. * * The log is implemented as a ring buffer, though the original creation * of the pool ('zpool create') is never overwritten. * * The history log is tracked as object 'spa_t::spa_history'. The bonus buffer * of 'spa_history' stores the offsets for logging/retrieving history as * 'spa_history_phys_t'. 'sh_pool_create_len' is the ending offset in bytes of * where the 'zpool create' record is stored. This allows us to never * overwrite the original creation of the pool. 'sh_phys_max_off' is the * physical ending offset in bytes of the log. This tells you the length of * the buffer. 'sh_eof' is the logical EOF (in bytes). Whenever a record * is added, 'sh_eof' is incremented by the size of the record. * 'sh_eof' is never decremented. 'sh_bof' is the logical BOF (in bytes). * This is where the consumer should start reading from after reading in * the 'zpool create' portion of the log. * * 'sh_records_lost' keeps track of how many records have been overwritten * and permanently lost. */ /* convert a logical offset to physical */ static uint64_t spa_history_log_to_phys(uint64_t log_off, spa_history_phys_t *shpp) { uint64_t phys_len; phys_len = shpp->sh_phys_max_off - shpp->sh_pool_create_len; return ((log_off - shpp->sh_pool_create_len) % phys_len + shpp->sh_pool_create_len); } void spa_history_create_obj(spa_t *spa, dmu_tx_t *tx) { dmu_buf_t *dbp; spa_history_phys_t *shpp; objset_t *mos = spa->spa_meta_objset; ASSERT0(spa->spa_history); spa->spa_history = dmu_object_alloc(mos, DMU_OT_SPA_HISTORY, SPA_OLD_MAXBLOCKSIZE, DMU_OT_SPA_HISTORY_OFFSETS, sizeof (spa_history_phys_t), tx); VERIFY0(zap_add(mos, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_HISTORY, sizeof (uint64_t), 1, &spa->spa_history, tx)); VERIFY0(dmu_bonus_hold(mos, spa->spa_history, FTAG, &dbp)); ASSERT3U(dbp->db_size, >=, sizeof (spa_history_phys_t)); shpp = dbp->db_data; dmu_buf_will_dirty(dbp, tx); /* * Figure out maximum size of history log. We set it at * 0.1% of pool size, with a max of 1G and min of 128KB. */ shpp->sh_phys_max_off = metaslab_class_get_dspace(spa_normal_class(spa)) / 1000; shpp->sh_phys_max_off = MIN(shpp->sh_phys_max_off, 1<<30); shpp->sh_phys_max_off = MAX(shpp->sh_phys_max_off, 128<<10); dmu_buf_rele(dbp, FTAG); } /* * Change 'sh_bof' to the beginning of the next record. */ static int spa_history_advance_bof(spa_t *spa, spa_history_phys_t *shpp) { objset_t *mos = spa->spa_meta_objset; uint64_t firstread, reclen, phys_bof; char buf[sizeof (reclen)]; int err; phys_bof = spa_history_log_to_phys(shpp->sh_bof, shpp); firstread = MIN(sizeof (reclen), shpp->sh_phys_max_off - phys_bof); if ((err = dmu_read(mos, spa->spa_history, phys_bof, firstread, buf, DMU_READ_PREFETCH)) != 0) return (err); if (firstread != sizeof (reclen)) { if ((err = dmu_read(mos, spa->spa_history, shpp->sh_pool_create_len, sizeof (reclen) - firstread, buf + firstread, DMU_READ_PREFETCH)) != 0) return (err); } reclen = LE_64(*((uint64_t *)buf)); shpp->sh_bof += reclen + sizeof (reclen); shpp->sh_records_lost++; return (0); } static int spa_history_write(spa_t *spa, void *buf, uint64_t len, spa_history_phys_t *shpp, dmu_tx_t *tx) { uint64_t firstwrite, phys_eof; objset_t *mos = spa->spa_meta_objset; int err; ASSERT(MUTEX_HELD(&spa->spa_history_lock)); /* see if we need to reset logical BOF */ while (shpp->sh_phys_max_off - shpp->sh_pool_create_len - (shpp->sh_eof - shpp->sh_bof) <= len) { if ((err = spa_history_advance_bof(spa, shpp)) != 0) { return (err); } } phys_eof = spa_history_log_to_phys(shpp->sh_eof, shpp); firstwrite = MIN(len, shpp->sh_phys_max_off - phys_eof); shpp->sh_eof += len; dmu_write(mos, spa->spa_history, phys_eof, firstwrite, buf, tx); len -= firstwrite; if (len > 0) { /* write out the rest at the beginning of physical file */ dmu_write(mos, spa->spa_history, shpp->sh_pool_create_len, len, (char *)buf + firstwrite, tx); } return (0); } /* * Post a history sysevent. * * The nvlist_t* passed into this function will be transformed into a new * nvlist where: * * 1. Nested nvlists will be flattened to a single level * 2. Keys will have their names normalized (to remove any problematic * characters, such as whitespace) * * The nvlist_t passed into this function will duplicated and should be freed * by caller. * */ static void spa_history_log_notify(spa_t *spa, nvlist_t *nvl) { nvlist_t *hist_nvl = fnvlist_alloc(); uint64_t uint64; char *string; if (nvlist_lookup_string(nvl, ZPOOL_HIST_CMD, &string) == 0) fnvlist_add_string(hist_nvl, ZFS_EV_HIST_CMD, string); if (nvlist_lookup_string(nvl, ZPOOL_HIST_INT_NAME, &string) == 0) fnvlist_add_string(hist_nvl, ZFS_EV_HIST_INT_NAME, string); if (nvlist_lookup_string(nvl, ZPOOL_HIST_ZONE, &string) == 0) fnvlist_add_string(hist_nvl, ZFS_EV_HIST_ZONE, string); if (nvlist_lookup_string(nvl, ZPOOL_HIST_HOST, &string) == 0) fnvlist_add_string(hist_nvl, ZFS_EV_HIST_HOST, string); if (nvlist_lookup_string(nvl, ZPOOL_HIST_DSNAME, &string) == 0) fnvlist_add_string(hist_nvl, ZFS_EV_HIST_DSNAME, string); if (nvlist_lookup_string(nvl, ZPOOL_HIST_INT_STR, &string) == 0) fnvlist_add_string(hist_nvl, ZFS_EV_HIST_INT_STR, string); if (nvlist_lookup_string(nvl, ZPOOL_HIST_IOCTL, &string) == 0) fnvlist_add_string(hist_nvl, ZFS_EV_HIST_IOCTL, string); if (nvlist_lookup_string(nvl, ZPOOL_HIST_INT_NAME, &string) == 0) fnvlist_add_string(hist_nvl, ZFS_EV_HIST_INT_NAME, string); if (nvlist_lookup_uint64(nvl, ZPOOL_HIST_DSID, &uint64) == 0) fnvlist_add_uint64(hist_nvl, ZFS_EV_HIST_DSID, uint64); if (nvlist_lookup_uint64(nvl, ZPOOL_HIST_TXG, &uint64) == 0) fnvlist_add_uint64(hist_nvl, ZFS_EV_HIST_TXG, uint64); if (nvlist_lookup_uint64(nvl, ZPOOL_HIST_TIME, &uint64) == 0) fnvlist_add_uint64(hist_nvl, ZFS_EV_HIST_TIME, uint64); if (nvlist_lookup_uint64(nvl, ZPOOL_HIST_WHO, &uint64) == 0) fnvlist_add_uint64(hist_nvl, ZFS_EV_HIST_WHO, uint64); if (nvlist_lookup_uint64(nvl, ZPOOL_HIST_INT_EVENT, &uint64) == 0) fnvlist_add_uint64(hist_nvl, ZFS_EV_HIST_INT_EVENT, uint64); spa_event_notify(spa, NULL, hist_nvl, ESC_ZFS_HISTORY_EVENT); nvlist_free(hist_nvl); } /* * Write out a history event. */ /*ARGSUSED*/ static void spa_history_log_sync(void *arg, dmu_tx_t *tx) { nvlist_t *nvl = arg; spa_t *spa = dmu_tx_pool(tx)->dp_spa; objset_t *mos = spa->spa_meta_objset; dmu_buf_t *dbp; spa_history_phys_t *shpp; size_t reclen; uint64_t le_len; char *record_packed = NULL; int ret; /* * If we have an older pool that doesn't have a command * history object, create it now. */ mutex_enter(&spa->spa_history_lock); if (!spa->spa_history) spa_history_create_obj(spa, tx); mutex_exit(&spa->spa_history_lock); /* * Get the offset of where we need to write via the bonus buffer. * Update the offset when the write completes. */ VERIFY0(dmu_bonus_hold(mos, spa->spa_history, FTAG, &dbp)); shpp = dbp->db_data; dmu_buf_will_dirty(dbp, tx); #ifdef ZFS_DEBUG { dmu_object_info_t doi; dmu_object_info_from_db(dbp, &doi); ASSERT3U(doi.doi_bonus_type, ==, DMU_OT_SPA_HISTORY_OFFSETS); } #endif fnvlist_add_uint64(nvl, ZPOOL_HIST_TIME, gethrestime_sec()); fnvlist_add_string(nvl, ZPOOL_HIST_HOST, utsname()->nodename); 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)); } /* * The history sysevent is posted only for internal history * messages to show what has happened, not how it happened. For * example, the following command: * * # zfs destroy -r tank/foo * * will result in one sysevent posted per dataset that is * destroyed as a result of the command - which could be more * than one event in total. By contrast, if the sysevent was * posted as a result of the ZPOOL_HIST_CMD key being present * it would result in only one sysevent being posted with the * full command line arguments, requiring the consumer to know - * how to parse and understand zfs(1M) command invocations. + * how to parse and understand zfs(8) command invocations. */ spa_history_log_notify(spa, nvl); } else if (nvlist_exists(nvl, ZPOOL_HIST_IOCTL)) { zfs_dbgmsg("ioctl %s", fnvlist_lookup_string(nvl, ZPOOL_HIST_IOCTL)); } VERIFY3U(nvlist_pack(nvl, &record_packed, &reclen, NV_ENCODE_NATIVE, KM_SLEEP), ==, 0); mutex_enter(&spa->spa_history_lock); /* write out the packed length as little endian */ le_len = LE_64((uint64_t)reclen); ret = spa_history_write(spa, &le_len, sizeof (le_len), shpp, tx); if (!ret) ret = spa_history_write(spa, record_packed, reclen, shpp, tx); /* 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); fnvlist_pack_free(record_packed, reclen); dmu_buf_rele(dbp, FTAG); fnvlist_free(nvl); } /* * Write out a history event. */ int spa_history_log(spa_t *spa, const char *msg) { 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, *in_nvl = NULL; if (spa_version(spa) < SPA_VERSION_ZPOOL_HISTORY || !spa_writeable(spa)) return (SET_ERROR(EINVAL)); err = nvlist_lookup_nvlist(nvl, ZPOOL_HIST_INPUT_NVL, &in_nvl); if (err == 0) { (void) nvlist_remove_all(in_nvl, ZPOOL_HIDDEN_ARGS); } tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir); err = dmu_tx_assign(tx, TXG_WAIT); if (err) { dmu_tx_abort(tx); return (err); } VERIFY0(nvlist_dup(nvl, &nvarg, KM_SLEEP)); 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_nowait(spa_get_dsl(spa), spa_history_log_sync, nvarg, tx); dmu_tx_commit(tx); /* spa_history_log_sync will free nvl */ return (err); } /* * Read out the command history. */ int spa_history_get(spa_t *spa, uint64_t *offp, uint64_t *len, char *buf) { objset_t *mos = spa->spa_meta_objset; dmu_buf_t *dbp; uint64_t read_len, phys_read_off, phys_eof; uint64_t leftover = 0; spa_history_phys_t *shpp; int err; /* * If the command history doesn't exist (older pool), * that's ok, just return ENOENT. */ if (!spa->spa_history) return (SET_ERROR(ENOENT)); /* * The history is logged asynchronously, so when they request * the first chunk of history, make sure everything has been * synced to disk so that we get it. */ if (*offp == 0 && spa_writeable(spa)) txg_wait_synced(spa_get_dsl(spa), 0); if ((err = dmu_bonus_hold(mos, spa->spa_history, FTAG, &dbp)) != 0) return (err); shpp = dbp->db_data; #ifdef ZFS_DEBUG { dmu_object_info_t doi; dmu_object_info_from_db(dbp, &doi); ASSERT3U(doi.doi_bonus_type, ==, DMU_OT_SPA_HISTORY_OFFSETS); } #endif mutex_enter(&spa->spa_history_lock); phys_eof = spa_history_log_to_phys(shpp->sh_eof, shpp); if (*offp < shpp->sh_pool_create_len) { /* read in just the zpool create history */ phys_read_off = *offp; read_len = MIN(*len, shpp->sh_pool_create_len - phys_read_off); } else { /* * Need to reset passed in offset to BOF if the passed in * offset has since been overwritten. */ *offp = MAX(*offp, shpp->sh_bof); phys_read_off = spa_history_log_to_phys(*offp, shpp); /* * Read up to the minimum of what the user passed down or * the EOF (physical or logical). If we hit physical EOF, * use 'leftover' to read from the physical BOF. */ if (phys_read_off <= phys_eof) { read_len = MIN(*len, phys_eof - phys_read_off); } else { read_len = MIN(*len, shpp->sh_phys_max_off - phys_read_off); if (phys_read_off + *len > shpp->sh_phys_max_off) { leftover = MIN(*len - read_len, phys_eof - shpp->sh_pool_create_len); } } } /* offset for consumer to use next */ *offp += read_len + leftover; /* tell the consumer how much you actually read */ *len = read_len + leftover; if (read_len == 0) { mutex_exit(&spa->spa_history_lock); dmu_buf_rele(dbp, FTAG); return (0); } err = dmu_read(mos, spa->spa_history, phys_read_off, read_len, buf, DMU_READ_PREFETCH); if (leftover && err == 0) { err = dmu_read(mos, spa->spa_history, shpp->sh_pool_create_len, leftover, buf + read_len, DMU_READ_PREFETCH); } mutex_exit(&spa->spa_history_lock); dmu_buf_rele(dbp, FTAG); return (err); } /* * The nvlist will be consumed by this call. */ static void log_internal(nvlist_t *nvl, const char *operation, spa_t *spa, dmu_tx_t *tx, const char *fmt, va_list adx) { char *msg; /* * If this is part of creating a pool, not everything is * initialized yet, so don't bother logging the internal events. * Likewise if the pool is not writeable. */ if (spa_is_initializing(spa) || !spa_writeable(spa)) { fnvlist_free(nvl); return; } msg = kmem_vasprintf(fmt, adx); fnvlist_add_string(nvl, ZPOOL_HIST_INT_STR, msg); kmem_strfree(msg); 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(nvl, tx); } else { dsl_sync_task_nowait(spa_get_dsl(spa), spa_history_log_sync, nvl, tx); } /* spa_history_log_sync() will free nvl */ } void spa_history_log_internal(spa_t *spa, const char *operation, dmu_tx_t *tx, const char *fmt, ...) { dmu_tx_t *htx = tx; va_list adx; /* create a tx if we didn't get one */ if (tx == NULL) { htx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir); if (dmu_tx_assign(htx, TXG_WAIT) != 0) { dmu_tx_abort(htx); return; } } va_start(adx, fmt); 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 */ if (tx == NULL) dmu_tx_commit(htx); } void spa_history_log_internal_ds(dsl_dataset_t *ds, const char *operation, dmu_tx_t *tx, const char *fmt, ...) { va_list adx; char namebuf[ZFS_MAX_DATASET_NAME_LEN]; 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[ZFS_MAX_DATASET_NAME_LEN]; 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, dsl_dir_phys(dd)->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, dmu_tx_t *tx) { utsname_t *u = utsname(); spa_history_log_internal(spa, operation, tx, "pool version %llu; software version %s; uts %s %s %s %s", (u_longlong_t)spa_version(spa), ZFS_META_GITREV, u->nodename, u->release, u->version, u->machine); } #ifndef _KERNEL const char * spa_history_zone(void) { return (NULL); } #endif #if defined(_KERNEL) EXPORT_SYMBOL(spa_history_create_obj); EXPORT_SYMBOL(spa_history_get); EXPORT_SYMBOL(spa_history_log); EXPORT_SYMBOL(spa_history_log_internal); EXPORT_SYMBOL(spa_history_log_version); #endif diff --git a/module/zfs/spa_misc.c b/module/zfs/spa_misc.c index 04210472886c..1640dceddc76 100644 --- a/module/zfs/spa_misc.c +++ b/module/zfs/spa_misc.c @@ -1,2922 +1,2922 @@ /* * 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) 2011, 2019 by Delphix. All rights reserved. * Copyright 2015 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved. * Copyright 2013 Saso Kiselkov. All rights reserved. * Copyright (c) 2017 Datto Inc. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2019, loli10K . All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_prop.h" #include #include #include #include /* * SPA locking * * There are three basic locks for managing spa_t structures: * * spa_namespace_lock (global mutex) * * This lock must be acquired to do any of the following: * * - Lookup a spa_t by name * - Add or remove a spa_t from the namespace * - Increase spa_refcount from non-zero * - Check if spa_refcount is zero * - Rename a spa_t * - add/remove/attach/detach devices * - Held for the duration of create/destroy/import/export * * It does not need to handle recursion. A create or destroy may * reference objects (files or zvols) in other pools, but by * definition they must have an existing reference, and will never need * to lookup a spa_t by name. * * spa_refcount (per-spa zfs_refcount_t protected by mutex) * * This reference count keep track of any active users of the spa_t. The * spa_t cannot be destroyed or freed while this is non-zero. Internally, * the refcount is never really 'zero' - opening a pool implicitly keeps * some references in the DMU. Internally we check against spa_minref, but * present the image of a zero/non-zero value to consumers. * * spa_config_lock[] (per-spa array of rwlocks) * * This protects the spa_t from config changes, and must be held in * the following circumstances: * * - RW_READER to perform I/O to the spa * - RW_WRITER to change the vdev config * * The locking order is fairly straightforward: * * spa_namespace_lock -> spa_refcount * * The namespace lock must be acquired to increase the refcount from 0 * or to check if it is zero. * * spa_refcount -> spa_config_lock[] * * There must be at least one valid reference on the spa_t to acquire * the config lock. * * spa_namespace_lock -> spa_config_lock[] * * The namespace lock must always be taken before the config lock. * * * The spa_namespace_lock can be acquired directly and is globally visible. * * The namespace is manipulated using the following functions, all of which * require the spa_namespace_lock to be held. * * spa_lookup() Lookup a spa_t by name. * * spa_add() Create a new spa_t in the namespace. * * spa_remove() Remove a spa_t from the namespace. This also * frees up any memory associated with the spa_t. * * spa_next() Returns the next spa_t in the system, or the * first if NULL is passed. * * spa_evict_all() Shutdown and remove all spa_t structures in * the system. * * spa_guid_exists() Determine whether a pool/device guid exists. * * The spa_refcount is manipulated using the following functions: * * spa_open_ref() Adds a reference to the given spa_t. Must be * called with spa_namespace_lock held if the * refcount is currently zero. * * spa_close() Remove a reference from the spa_t. This will * not free the spa_t or remove it from the * namespace. No locking is required. * * spa_refcount_zero() Returns true if the refcount is currently * zero. Must be called with spa_namespace_lock * held. * * The spa_config_lock[] is an array of rwlocks, ordered as follows: * SCL_CONFIG > SCL_STATE > SCL_ALLOC > SCL_ZIO > SCL_FREE > SCL_VDEV. * spa_config_lock[] is manipulated with spa_config_{enter,exit,held}(). * * To read the configuration, it suffices to hold one of these locks as reader. * To modify the configuration, you must hold all locks as writer. To modify * vdev state without altering the vdev tree's topology (e.g. online/offline), * you must hold SCL_STATE and SCL_ZIO as writer. * * We use these distinct config locks to avoid recursive lock entry. * For example, spa_sync() (which holds SCL_CONFIG as reader) induces * block allocations (SCL_ALLOC), which may require reading space maps * from disk (dmu_read() -> zio_read() -> SCL_ZIO). * * The spa config locks cannot be normal rwlocks because we need the * ability to hand off ownership. For example, SCL_ZIO is acquired * by the issuing thread and later released by an interrupt thread. * They do, however, obey the usual write-wanted semantics to prevent * writer (i.e. system administrator) starvation. * * The lock acquisition rules are as follows: * * SCL_CONFIG * Protects changes to the vdev tree topology, such as vdev * add/remove/attach/detach. Protects the dirty config list * (spa_config_dirty_list) and the set of spares and l2arc devices. * * SCL_STATE * Protects changes to pool state and vdev state, such as vdev * online/offline/fault/degrade/clear. Protects the dirty state list * (spa_state_dirty_list) and global pool state (spa_state). * * SCL_ALLOC * Protects changes to metaslab groups and classes. * Held as reader by metaslab_alloc() and metaslab_claim(). * * SCL_ZIO * Held by bp-level zios (those which have no io_vd upon entry) * to prevent changes to the vdev tree. The bp-level zio implicitly * protects all of its vdev child zios, which do not hold SCL_ZIO. * * SCL_FREE * Protects changes to metaslab groups and classes. * Held as reader by metaslab_free(). SCL_FREE is distinct from * SCL_ALLOC, and lower than SCL_ZIO, so that we can safely free * blocks in zio_done() while another i/o that holds either * SCL_ALLOC or SCL_ZIO is waiting for this i/o to complete. * * SCL_VDEV * Held as reader to prevent changes to the vdev tree during trivial * inquiries such as bp_get_dsize(). SCL_VDEV is distinct from the * other locks, and lower than all of them, to ensure that it's safe * to acquire regardless of caller context. * * In addition, the following rules apply: * * (a) spa_props_lock protects pool properties, spa_config and spa_config_list. * The lock ordering is SCL_CONFIG > spa_props_lock. * * (b) I/O operations on leaf vdevs. For any zio operation that takes * an explicit vdev_t argument -- such as zio_ioctl(), zio_read_phys(), * or zio_write_phys() -- the caller must ensure that the config cannot * cannot change in the interim, and that the vdev cannot be reopened. * SCL_STATE as reader suffices for both. * * The vdev configuration is protected by spa_vdev_enter() / spa_vdev_exit(). * * spa_vdev_enter() Acquire the namespace lock and the config lock * for writing. * * spa_vdev_exit() Release the config lock, wait for all I/O * to complete, sync the updated configs to the * cache, and release the namespace lock. * * vdev state is protected by spa_vdev_state_enter() / spa_vdev_state_exit(). * Like spa_vdev_enter/exit, these are convenience wrappers -- the actual * locking is, always, based on spa_namespace_lock and spa_config_lock[]. */ static avl_tree_t spa_namespace_avl; kmutex_t spa_namespace_lock; static kcondvar_t spa_namespace_cv; int spa_max_replication_override = SPA_DVAS_PER_BP; static kmutex_t spa_spare_lock; static avl_tree_t spa_spare_avl; static kmutex_t spa_l2cache_lock; static avl_tree_t spa_l2cache_avl; kmem_cache_t *spa_buffer_pool; spa_mode_t spa_mode_global = SPA_MODE_UNINIT; #ifdef ZFS_DEBUG /* * Everything except dprintf, set_error, spa, and indirect_remap is on * by default in debug builds. */ int zfs_flags = ~(ZFS_DEBUG_DPRINTF | ZFS_DEBUG_SET_ERROR | ZFS_DEBUG_INDIRECT_REMAP); #else int zfs_flags = 0; #endif /* * zfs_recover can be set to nonzero to attempt to recover from * otherwise-fatal errors, typically caused by on-disk corruption. When * set, calls to zfs_panic_recover() will turn into warning messages. * This should only be used as a last resort, as it typically results * in leaked space, or worse. */ int zfs_recover = B_FALSE; /* * If destroy encounters an EIO while reading metadata (e.g. indirect * blocks), space referenced by the missing metadata can not be freed. * Normally this causes the background destroy to become "stalled", as * it is unable to make forward progress. While in this stalled state, * all remaining space to free from the error-encountering filesystem is * "temporarily leaked". Set this flag to cause it to ignore the EIO, * permanently leak the space from indirect blocks that can not be read, * and continue to free everything else that it can. * * The default, "stalling" behavior is useful if the storage partially * fails (i.e. some but not all i/os fail), and then later recovers. In * this case, we will be able to continue pool operations while it is * partially failed, and when it recovers, we can continue to free the * space, with no leaks. However, note that this case is actually * fairly rare. * * Typically pools either (a) fail completely (but perhaps temporarily, * e.g. a top-level vdev going offline), or (b) have localized, * permanent errors (e.g. disk returns the wrong data due to bit flip or * firmware bug). In case (a), this setting does not matter because the * pool will be suspended and the sync thread will not be able to make * forward progress regardless. In case (b), because the error is * permanent, the best we can do is leak the minimum amount of space, * which is what setting this flag will do. Therefore, it is reasonable * for this flag to normally be set, but we chose the more conservative * approach of not setting it, so that there is no possibility of * leaking space in the "partial temporary" failure case. */ int zfs_free_leak_on_eio = B_FALSE; /* * Expiration time in milliseconds. This value has two meanings. First it is * used to determine when the spa_deadman() logic should fire. By default the * spa_deadman() will fire if spa_sync() has not completed in 600 seconds. * Secondly, the value determines if an I/O is considered "hung". Any I/O that * has not completed in zfs_deadman_synctime_ms is considered "hung" resulting * in one of three behaviors controlled by zfs_deadman_failmode. */ unsigned long zfs_deadman_synctime_ms = 600000UL; /* * This value controls the maximum amount of time zio_wait() will block for an * outstanding IO. By default this is 300 seconds at which point the "hung" * behavior will be applied as described for zfs_deadman_synctime_ms. */ unsigned long zfs_deadman_ziotime_ms = 300000UL; /* * Check time in milliseconds. This defines the frequency at which we check * for hung I/O. */ unsigned long zfs_deadman_checktime_ms = 60000UL; /* * By default the deadman is enabled. */ int zfs_deadman_enabled = 1; /* * Controls the behavior of the deadman when it detects a "hung" I/O. * Valid values are zfs_deadman_failmode=. * * wait - Wait for the "hung" I/O (default) * continue - Attempt to recover from a "hung" I/O * panic - Panic the system */ char *zfs_deadman_failmode = "wait"; /* * The worst case is single-sector max-parity RAID-Z blocks, in which * case the space requirement is exactly (VDEV_RAIDZ_MAXPARITY + 1) * times the size; so just assume that. Add to this the fact that * we can have up to 3 DVAs per bp, and one more factor of 2 because * the block may be dittoed with up to 3 DVAs by ddt_sync(). All together, * the worst case is: * (VDEV_RAIDZ_MAXPARITY + 1) * SPA_DVAS_PER_BP * 2 == 24 */ int spa_asize_inflation = 24; /* * Normally, we don't allow the last 3.2% (1/(2^spa_slop_shift)) of space in * the pool to be consumed. This ensures that we don't run the pool * completely out of space, due to unaccounted changes (e.g. to the MOS). * It also limits the worst-case time to allocate space. If we have * less than this amount of free space, most ZPL operations (e.g. write, * create) will return ENOSPC. * * Certain operations (e.g. file removal, most administrative actions) can * use half the slop space. They will only return ENOSPC if less than half * the slop space is free. Typically, once the pool has less than the slop * space free, the user will use these operations to free up space in the pool. * These are the operations that call dsl_pool_adjustedsize() with the netfree * argument set to TRUE. * * Operations that are almost guaranteed to free up space in the absence of * a pool checkpoint can use up to three quarters of the slop space * (e.g zfs destroy). * * A very restricted set of operations are always permitted, regardless of * the amount of free space. These are the operations that call * dsl_sync_task(ZFS_SPACE_CHECK_NONE). If these operations result in a net * increase in the amount of space used, it is possible to run the pool * completely out of space, causing it to be permanently read-only. * * Note that on very small pools, the slop space will be larger than * 3.2%, in an effort to have it be at least spa_min_slop (128MB), * but we never allow it to be more than half the pool size. * * See also the comments in zfs_space_check_t. */ int spa_slop_shift = 5; uint64_t spa_min_slop = 128 * 1024 * 1024; int spa_allocators = 4; /*PRINTFLIKE2*/ void spa_load_failed(spa_t *spa, const char *fmt, ...) { va_list adx; char buf[256]; va_start(adx, fmt); (void) vsnprintf(buf, sizeof (buf), fmt, adx); va_end(adx); zfs_dbgmsg("spa_load(%s, config %s): FAILED: %s", spa->spa_name, spa->spa_trust_config ? "trusted" : "untrusted", buf); } /*PRINTFLIKE2*/ void spa_load_note(spa_t *spa, const char *fmt, ...) { va_list adx; char buf[256]; va_start(adx, fmt); (void) vsnprintf(buf, sizeof (buf), fmt, adx); va_end(adx); zfs_dbgmsg("spa_load(%s, config %s): %s", spa->spa_name, spa->spa_trust_config ? "trusted" : "untrusted", buf); } /* * By default dedup and user data indirects land in the special class */ int zfs_ddt_data_is_special = B_TRUE; int zfs_user_indirect_is_special = B_TRUE; /* * The percentage of special class final space reserved for metadata only. * Once we allocate 100 - zfs_special_class_metadata_reserve_pct we only * let metadata into the class. */ int zfs_special_class_metadata_reserve_pct = 25; /* * ========================================================================== * SPA config locking * ========================================================================== */ static void spa_config_lock_init(spa_t *spa) { for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; mutex_init(&scl->scl_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&scl->scl_cv, NULL, CV_DEFAULT, NULL); zfs_refcount_create_untracked(&scl->scl_count); scl->scl_writer = NULL; scl->scl_write_wanted = 0; } } static void spa_config_lock_destroy(spa_t *spa) { for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; mutex_destroy(&scl->scl_lock); cv_destroy(&scl->scl_cv); zfs_refcount_destroy(&scl->scl_count); ASSERT(scl->scl_writer == NULL); ASSERT(scl->scl_write_wanted == 0); } } int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw) { for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (!(locks & (1 << i))) continue; mutex_enter(&scl->scl_lock); if (rw == RW_READER) { if (scl->scl_writer || scl->scl_write_wanted) { mutex_exit(&scl->scl_lock); spa_config_exit(spa, locks & ((1 << i) - 1), tag); return (0); } } else { ASSERT(scl->scl_writer != curthread); if (!zfs_refcount_is_zero(&scl->scl_count)) { mutex_exit(&scl->scl_lock); spa_config_exit(spa, locks & ((1 << i) - 1), tag); return (0); } scl->scl_writer = curthread; } (void) zfs_refcount_add(&scl->scl_count, tag); mutex_exit(&scl->scl_lock); } return (1); } void spa_config_enter(spa_t *spa, int locks, const void *tag, krw_t rw) { int wlocks_held = 0; ASSERT3U(SCL_LOCKS, <, sizeof (wlocks_held) * NBBY); for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (scl->scl_writer == curthread) wlocks_held |= (1 << i); if (!(locks & (1 << i))) continue; mutex_enter(&scl->scl_lock); if (rw == RW_READER) { while (scl->scl_writer || scl->scl_write_wanted) { cv_wait(&scl->scl_cv, &scl->scl_lock); } } else { ASSERT(scl->scl_writer != curthread); while (!zfs_refcount_is_zero(&scl->scl_count)) { scl->scl_write_wanted++; cv_wait(&scl->scl_cv, &scl->scl_lock); scl->scl_write_wanted--; } scl->scl_writer = curthread; } (void) zfs_refcount_add(&scl->scl_count, tag); mutex_exit(&scl->scl_lock); } ASSERT3U(wlocks_held, <=, locks); } void spa_config_exit(spa_t *spa, int locks, const void *tag) { for (int i = SCL_LOCKS - 1; i >= 0; i--) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (!(locks & (1 << i))) continue; mutex_enter(&scl->scl_lock); ASSERT(!zfs_refcount_is_zero(&scl->scl_count)); if (zfs_refcount_remove(&scl->scl_count, tag) == 0) { ASSERT(scl->scl_writer == NULL || scl->scl_writer == curthread); scl->scl_writer = NULL; /* OK in either case */ cv_broadcast(&scl->scl_cv); } mutex_exit(&scl->scl_lock); } } int spa_config_held(spa_t *spa, int locks, krw_t rw) { int locks_held = 0; for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (!(locks & (1 << i))) continue; if ((rw == RW_READER && !zfs_refcount_is_zero(&scl->scl_count)) || (rw == RW_WRITER && scl->scl_writer == curthread)) locks_held |= 1 << i; } return (locks_held); } /* * ========================================================================== * SPA namespace functions * ========================================================================== */ /* * Lookup the named spa_t in the AVL tree. The spa_namespace_lock must be held. * Returns NULL if no matching spa_t is found. */ spa_t * spa_lookup(const char *name) { static spa_t search; /* spa_t is large; don't allocate on stack */ spa_t *spa; avl_index_t where; char *cp; ASSERT(MUTEX_HELD(&spa_namespace_lock)); (void) strlcpy(search.spa_name, name, sizeof (search.spa_name)); /* * If it's a full dataset name, figure out the pool name and * just use that. */ cp = strpbrk(search.spa_name, "/@#"); if (cp != NULL) *cp = '\0'; spa = avl_find(&spa_namespace_avl, &search, &where); return (spa); } /* * Fires when spa_sync has not completed within zfs_deadman_synctime_ms. * If the zfs_deadman_enabled flag is set then it inspects all vdev queues * looking for potentially hung I/Os. */ void spa_deadman(void *arg) { spa_t *spa = arg; /* Disable the deadman if the pool is suspended. */ if (spa_suspended(spa)) return; zfs_dbgmsg("slow spa_sync: started %llu seconds ago, calls %llu", (gethrtime() - spa->spa_sync_starttime) / NANOSEC, ++spa->spa_deadman_calls); if (zfs_deadman_enabled) vdev_deadman(spa->spa_root_vdev, FTAG); spa->spa_deadman_tqid = taskq_dispatch_delay(system_delay_taskq, spa_deadman, spa, TQ_SLEEP, ddi_get_lbolt() + MSEC_TO_TICK(zfs_deadman_checktime_ms)); } static int spa_log_sm_sort_by_txg(const void *va, const void *vb) { const spa_log_sm_t *a = va; const spa_log_sm_t *b = vb; return (TREE_CMP(a->sls_txg, b->sls_txg)); } /* * Create an uninitialized spa_t with the given name. Requires * spa_namespace_lock. The caller must ensure that the spa_t doesn't already * exist by calling spa_lookup() first. */ spa_t * spa_add(const char *name, nvlist_t *config, const char *altroot) { spa_t *spa; spa_config_dirent_t *dp; ASSERT(MUTEX_HELD(&spa_namespace_lock)); spa = kmem_zalloc(sizeof (spa_t), KM_SLEEP); mutex_init(&spa->spa_async_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_errlist_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_errlog_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_evicting_os_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_history_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_proc_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_props_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_cksum_tmpls_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_scrub_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_suspend_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_vdev_top_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_feat_stats_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_flushed_ms_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_activities_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&spa->spa_async_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_evicting_os_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_proc_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_scrub_io_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_suspend_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_activities_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_waiters_cv, NULL, CV_DEFAULT, NULL); for (int t = 0; t < TXG_SIZE; t++) bplist_create(&spa->spa_free_bplist[t]); (void) strlcpy(spa->spa_name, name, sizeof (spa->spa_name)); spa->spa_state = POOL_STATE_UNINITIALIZED; spa->spa_freeze_txg = UINT64_MAX; spa->spa_final_txg = UINT64_MAX; spa->spa_load_max_txg = UINT64_MAX; spa->spa_proc = &p0; spa->spa_proc_state = SPA_PROC_NONE; spa->spa_trust_config = B_TRUE; spa->spa_hostid = zone_get_hostid(NULL); spa->spa_deadman_synctime = MSEC2NSEC(zfs_deadman_synctime_ms); spa->spa_deadman_ziotime = MSEC2NSEC(zfs_deadman_ziotime_ms); spa_set_deadman_failmode(spa, zfs_deadman_failmode); zfs_refcount_create(&spa->spa_refcount); spa_config_lock_init(spa); spa_stats_init(spa); avl_add(&spa_namespace_avl, spa); /* * Set the alternate root, if there is one. */ if (altroot) spa->spa_root = spa_strdup(altroot); spa->spa_alloc_count = spa_allocators; spa->spa_alloc_locks = kmem_zalloc(spa->spa_alloc_count * sizeof (kmutex_t), KM_SLEEP); spa->spa_alloc_trees = kmem_zalloc(spa->spa_alloc_count * sizeof (avl_tree_t), KM_SLEEP); for (int i = 0; i < spa->spa_alloc_count; i++) { mutex_init(&spa->spa_alloc_locks[i], NULL, MUTEX_DEFAULT, NULL); avl_create(&spa->spa_alloc_trees[i], zio_bookmark_compare, sizeof (zio_t), offsetof(zio_t, io_alloc_node)); } avl_create(&spa->spa_metaslabs_by_flushed, metaslab_sort_by_flushed, sizeof (metaslab_t), offsetof(metaslab_t, ms_spa_txg_node)); avl_create(&spa->spa_sm_logs_by_txg, spa_log_sm_sort_by_txg, sizeof (spa_log_sm_t), offsetof(spa_log_sm_t, sls_node)); list_create(&spa->spa_log_summary, sizeof (log_summary_entry_t), offsetof(log_summary_entry_t, lse_node)); /* * Every pool starts with the default cachefile */ list_create(&spa->spa_config_list, sizeof (spa_config_dirent_t), offsetof(spa_config_dirent_t, scd_link)); dp = kmem_zalloc(sizeof (spa_config_dirent_t), KM_SLEEP); dp->scd_path = altroot ? NULL : spa_strdup(spa_config_path); list_insert_head(&spa->spa_config_list, dp); VERIFY(nvlist_alloc(&spa->spa_load_info, NV_UNIQUE_NAME, KM_SLEEP) == 0); if (config != NULL) { nvlist_t *features; if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_FEATURES_FOR_READ, &features) == 0) { VERIFY(nvlist_dup(features, &spa->spa_label_features, 0) == 0); } VERIFY(nvlist_dup(config, &spa->spa_config, 0) == 0); } if (spa->spa_label_features == NULL) { VERIFY(nvlist_alloc(&spa->spa_label_features, NV_UNIQUE_NAME, KM_SLEEP) == 0); } spa->spa_min_ashift = INT_MAX; spa->spa_max_ashift = 0; /* Reset cached value */ spa->spa_dedup_dspace = ~0ULL; /* * As a pool is being created, treat all features as disabled by * setting SPA_FEATURE_DISABLED for all entries in the feature * refcount cache. */ for (int i = 0; i < SPA_FEATURES; i++) { spa->spa_feat_refcount_cache[i] = SPA_FEATURE_DISABLED; } list_create(&spa->spa_leaf_list, sizeof (vdev_t), offsetof(vdev_t, vdev_leaf_node)); return (spa); } /* * Removes a spa_t from the namespace, freeing up any memory used. Requires * spa_namespace_lock. This is called only after the spa_t has been closed and * deactivated. */ void spa_remove(spa_t *spa) { spa_config_dirent_t *dp; ASSERT(MUTEX_HELD(&spa_namespace_lock)); ASSERT(spa_state(spa) == POOL_STATE_UNINITIALIZED); ASSERT3U(zfs_refcount_count(&spa->spa_refcount), ==, 0); ASSERT0(spa->spa_waiters); nvlist_free(spa->spa_config_splitting); avl_remove(&spa_namespace_avl, spa); cv_broadcast(&spa_namespace_cv); if (spa->spa_root) spa_strfree(spa->spa_root); while ((dp = list_head(&spa->spa_config_list)) != NULL) { list_remove(&spa->spa_config_list, dp); if (dp->scd_path != NULL) spa_strfree(dp->scd_path); kmem_free(dp, sizeof (spa_config_dirent_t)); } for (int i = 0; i < spa->spa_alloc_count; i++) { avl_destroy(&spa->spa_alloc_trees[i]); mutex_destroy(&spa->spa_alloc_locks[i]); } kmem_free(spa->spa_alloc_locks, spa->spa_alloc_count * sizeof (kmutex_t)); kmem_free(spa->spa_alloc_trees, spa->spa_alloc_count * sizeof (avl_tree_t)); avl_destroy(&spa->spa_metaslabs_by_flushed); avl_destroy(&spa->spa_sm_logs_by_txg); list_destroy(&spa->spa_log_summary); list_destroy(&spa->spa_config_list); list_destroy(&spa->spa_leaf_list); nvlist_free(spa->spa_label_features); nvlist_free(spa->spa_load_info); nvlist_free(spa->spa_feat_stats); spa_config_set(spa, NULL); zfs_refcount_destroy(&spa->spa_refcount); spa_stats_destroy(spa); spa_config_lock_destroy(spa); for (int t = 0; t < TXG_SIZE; t++) bplist_destroy(&spa->spa_free_bplist[t]); zio_checksum_templates_free(spa); cv_destroy(&spa->spa_async_cv); cv_destroy(&spa->spa_evicting_os_cv); cv_destroy(&spa->spa_proc_cv); cv_destroy(&spa->spa_scrub_io_cv); cv_destroy(&spa->spa_suspend_cv); cv_destroy(&spa->spa_activities_cv); cv_destroy(&spa->spa_waiters_cv); mutex_destroy(&spa->spa_flushed_ms_lock); mutex_destroy(&spa->spa_async_lock); mutex_destroy(&spa->spa_errlist_lock); mutex_destroy(&spa->spa_errlog_lock); mutex_destroy(&spa->spa_evicting_os_lock); mutex_destroy(&spa->spa_history_lock); mutex_destroy(&spa->spa_proc_lock); mutex_destroy(&spa->spa_props_lock); mutex_destroy(&spa->spa_cksum_tmpls_lock); mutex_destroy(&spa->spa_scrub_lock); mutex_destroy(&spa->spa_suspend_lock); mutex_destroy(&spa->spa_vdev_top_lock); mutex_destroy(&spa->spa_feat_stats_lock); mutex_destroy(&spa->spa_activities_lock); kmem_free(spa, sizeof (spa_t)); } /* * Given a pool, return the next pool in the namespace, or NULL if there is * none. If 'prev' is NULL, return the first pool. */ spa_t * spa_next(spa_t *prev) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); if (prev) return (AVL_NEXT(&spa_namespace_avl, prev)); else return (avl_first(&spa_namespace_avl)); } /* * ========================================================================== * SPA refcount functions * ========================================================================== */ /* * Add a reference to the given spa_t. Must have at least one reference, or * have the namespace lock held. */ void spa_open_ref(spa_t *spa, void *tag) { ASSERT(zfs_refcount_count(&spa->spa_refcount) >= spa->spa_minref || MUTEX_HELD(&spa_namespace_lock)); (void) zfs_refcount_add(&spa->spa_refcount, tag); } /* * Remove a reference to the given spa_t. Must have at least one reference, or * have the namespace lock held. */ void spa_close(spa_t *spa, void *tag) { ASSERT(zfs_refcount_count(&spa->spa_refcount) > spa->spa_minref || MUTEX_HELD(&spa_namespace_lock)); (void) zfs_refcount_remove(&spa->spa_refcount, tag); } /* * Remove a reference to the given spa_t held by a dsl dir that is * being asynchronously released. Async releases occur from a taskq * performing eviction of dsl datasets and dirs. The namespace lock * isn't held and the hold by the object being evicted may contribute to * spa_minref (e.g. dataset or directory released during pool export), * so the asserts in spa_close() do not apply. */ void spa_async_close(spa_t *spa, void *tag) { (void) zfs_refcount_remove(&spa->spa_refcount, tag); } /* * Check to see if the spa refcount is zero. Must be called with * spa_namespace_lock held. We really compare against spa_minref, which is the * number of references acquired when opening a pool */ boolean_t spa_refcount_zero(spa_t *spa) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); return (zfs_refcount_count(&spa->spa_refcount) == spa->spa_minref); } /* * ========================================================================== * SPA spare and l2cache tracking * ========================================================================== */ /* * Hot spares and cache devices are tracked using the same code below, * for 'auxiliary' devices. */ typedef struct spa_aux { uint64_t aux_guid; uint64_t aux_pool; avl_node_t aux_avl; int aux_count; } spa_aux_t; static inline int spa_aux_compare(const void *a, const void *b) { const spa_aux_t *sa = (const spa_aux_t *)a; const spa_aux_t *sb = (const spa_aux_t *)b; return (TREE_CMP(sa->aux_guid, sb->aux_guid)); } static void spa_aux_add(vdev_t *vd, avl_tree_t *avl) { avl_index_t where; spa_aux_t search; spa_aux_t *aux; search.aux_guid = vd->vdev_guid; if ((aux = avl_find(avl, &search, &where)) != NULL) { aux->aux_count++; } else { aux = kmem_zalloc(sizeof (spa_aux_t), KM_SLEEP); aux->aux_guid = vd->vdev_guid; aux->aux_count = 1; avl_insert(avl, aux, where); } } static void spa_aux_remove(vdev_t *vd, avl_tree_t *avl) { spa_aux_t search; spa_aux_t *aux; avl_index_t where; search.aux_guid = vd->vdev_guid; aux = avl_find(avl, &search, &where); ASSERT(aux != NULL); if (--aux->aux_count == 0) { avl_remove(avl, aux); kmem_free(aux, sizeof (spa_aux_t)); } else if (aux->aux_pool == spa_guid(vd->vdev_spa)) { aux->aux_pool = 0ULL; } } static boolean_t spa_aux_exists(uint64_t guid, uint64_t *pool, int *refcnt, avl_tree_t *avl) { spa_aux_t search, *found; search.aux_guid = guid; found = avl_find(avl, &search, NULL); if (pool) { if (found) *pool = found->aux_pool; else *pool = 0ULL; } if (refcnt) { if (found) *refcnt = found->aux_count; else *refcnt = 0; } return (found != NULL); } static void spa_aux_activate(vdev_t *vd, avl_tree_t *avl) { spa_aux_t search, *found; avl_index_t where; search.aux_guid = vd->vdev_guid; found = avl_find(avl, &search, &where); ASSERT(found != NULL); ASSERT(found->aux_pool == 0ULL); found->aux_pool = spa_guid(vd->vdev_spa); } /* * Spares are tracked globally due to the following constraints: * * - A spare may be part of multiple pools. * - A spare may be added to a pool even if it's actively in use within * another pool. * - A spare in use in any pool can only be the source of a replacement if * the target is a spare in the same pool. * * We keep track of all spares on the system through the use of a reference * counted AVL tree. When a vdev is added as a spare, or used as a replacement * spare, then we bump the reference count in the AVL tree. In addition, we set * the 'vdev_isspare' member to indicate that the device is a spare (active or * inactive). When a spare is made active (used to replace a device in the * pool), we also keep track of which pool its been made a part of. * * The 'spa_spare_lock' protects the AVL tree. These functions are normally * called under the spa_namespace lock as part of vdev reconfiguration. The * separate spare lock exists for the status query path, which does not need to * be completely consistent with respect to other vdev configuration changes. */ static int spa_spare_compare(const void *a, const void *b) { return (spa_aux_compare(a, b)); } void spa_spare_add(vdev_t *vd) { mutex_enter(&spa_spare_lock); ASSERT(!vd->vdev_isspare); spa_aux_add(vd, &spa_spare_avl); vd->vdev_isspare = B_TRUE; mutex_exit(&spa_spare_lock); } void spa_spare_remove(vdev_t *vd) { mutex_enter(&spa_spare_lock); ASSERT(vd->vdev_isspare); spa_aux_remove(vd, &spa_spare_avl); vd->vdev_isspare = B_FALSE; mutex_exit(&spa_spare_lock); } boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt) { boolean_t found; mutex_enter(&spa_spare_lock); found = spa_aux_exists(guid, pool, refcnt, &spa_spare_avl); mutex_exit(&spa_spare_lock); return (found); } void spa_spare_activate(vdev_t *vd) { mutex_enter(&spa_spare_lock); ASSERT(vd->vdev_isspare); spa_aux_activate(vd, &spa_spare_avl); mutex_exit(&spa_spare_lock); } /* * Level 2 ARC devices are tracked globally for the same reasons as spares. * Cache devices currently only support one pool per cache device, and so * for these devices the aux reference count is currently unused beyond 1. */ static int spa_l2cache_compare(const void *a, const void *b) { return (spa_aux_compare(a, b)); } void spa_l2cache_add(vdev_t *vd) { mutex_enter(&spa_l2cache_lock); ASSERT(!vd->vdev_isl2cache); spa_aux_add(vd, &spa_l2cache_avl); vd->vdev_isl2cache = B_TRUE; mutex_exit(&spa_l2cache_lock); } void spa_l2cache_remove(vdev_t *vd) { mutex_enter(&spa_l2cache_lock); ASSERT(vd->vdev_isl2cache); spa_aux_remove(vd, &spa_l2cache_avl); vd->vdev_isl2cache = B_FALSE; mutex_exit(&spa_l2cache_lock); } boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool) { boolean_t found; mutex_enter(&spa_l2cache_lock); found = spa_aux_exists(guid, pool, NULL, &spa_l2cache_avl); mutex_exit(&spa_l2cache_lock); return (found); } void spa_l2cache_activate(vdev_t *vd) { mutex_enter(&spa_l2cache_lock); ASSERT(vd->vdev_isl2cache); spa_aux_activate(vd, &spa_l2cache_avl); mutex_exit(&spa_l2cache_lock); } /* * ========================================================================== * SPA vdev locking * ========================================================================== */ /* * Lock the given spa_t for the purpose of adding or removing a vdev. * Grabs the global spa_namespace_lock plus the spa config lock for writing. * It returns the next transaction group for the spa_t. */ uint64_t spa_vdev_enter(spa_t *spa) { mutex_enter(&spa->spa_vdev_top_lock); mutex_enter(&spa_namespace_lock); vdev_autotrim_stop_all(spa); return (spa_vdev_config_enter(spa)); } /* * The same as spa_vdev_enter() above but additionally takes the guid of * the vdev being detached. When there is a rebuild in process it will be * suspended while the vdev tree is modified then resumed by spa_vdev_exit(). * The rebuild is canceled if only a single child remains after the detach. */ uint64_t spa_vdev_detach_enter(spa_t *spa, uint64_t guid) { mutex_enter(&spa->spa_vdev_top_lock); mutex_enter(&spa_namespace_lock); vdev_autotrim_stop_all(spa); if (guid != 0) { vdev_t *vd = spa_lookup_by_guid(spa, guid, B_FALSE); if (vd) { vdev_rebuild_stop_wait(vd->vdev_top); } } return (spa_vdev_config_enter(spa)); } /* * Internal implementation for spa_vdev_enter(). Used when a vdev * operation requires multiple syncs (i.e. removing a device) while * keeping the spa_namespace_lock held. */ uint64_t spa_vdev_config_enter(spa_t *spa) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); spa_config_enter(spa, SCL_ALL, spa, RW_WRITER); return (spa_last_synced_txg(spa) + 1); } /* * Used in combination with spa_vdev_config_enter() to allow the syncing * of multiple transactions without releasing the spa_namespace_lock. */ void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error, char *tag) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); int config_changed = B_FALSE; ASSERT(txg > spa_last_synced_txg(spa)); spa->spa_pending_vdev = NULL; /* * Reassess the DTLs. */ vdev_dtl_reassess(spa->spa_root_vdev, 0, 0, B_FALSE, B_FALSE); if (error == 0 && !list_is_empty(&spa->spa_config_dirty_list)) { config_changed = B_TRUE; spa->spa_config_generation++; } /* * Verify the metaslab classes. */ ASSERT(metaslab_class_validate(spa_normal_class(spa)) == 0); ASSERT(metaslab_class_validate(spa_log_class(spa)) == 0); ASSERT(metaslab_class_validate(spa_special_class(spa)) == 0); ASSERT(metaslab_class_validate(spa_dedup_class(spa)) == 0); spa_config_exit(spa, SCL_ALL, spa); /* * Panic the system if the specified tag requires it. This * is useful for ensuring that configurations are updated * transactionally. */ if (zio_injection_enabled) zio_handle_panic_injection(spa, tag, 0); /* * Note: this txg_wait_synced() is important because it ensures * that there won't be more than one config change per txg. * This allows us to use the txg as the generation number. */ if (error == 0) txg_wait_synced(spa->spa_dsl_pool, txg); if (vd != NULL) { ASSERT(!vd->vdev_detached || vd->vdev_dtl_sm == NULL); if (vd->vdev_ops->vdev_op_leaf) { mutex_enter(&vd->vdev_initialize_lock); vdev_initialize_stop(vd, VDEV_INITIALIZE_CANCELED, NULL); mutex_exit(&vd->vdev_initialize_lock); mutex_enter(&vd->vdev_trim_lock); vdev_trim_stop(vd, VDEV_TRIM_CANCELED, NULL); mutex_exit(&vd->vdev_trim_lock); } /* * The vdev may be both a leaf and top-level device. */ vdev_autotrim_stop_wait(vd); spa_config_enter(spa, SCL_ALL, spa, RW_WRITER); vdev_free(vd); spa_config_exit(spa, SCL_ALL, spa); } /* * If the config changed, update the config cache. */ if (config_changed) spa_write_cachefile(spa, B_FALSE, B_TRUE); } /* * Unlock the spa_t after adding or removing a vdev. Besides undoing the * locking of spa_vdev_enter(), we also want make sure the transactions have * synced to disk, and then update the global configuration cache with the new * information. */ int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error) { vdev_autotrim_restart(spa); vdev_rebuild_restart(spa); spa_vdev_config_exit(spa, vd, txg, error, FTAG); mutex_exit(&spa_namespace_lock); mutex_exit(&spa->spa_vdev_top_lock); return (error); } /* * Lock the given spa_t for the purpose of changing vdev state. */ void spa_vdev_state_enter(spa_t *spa, int oplocks) { int locks = SCL_STATE_ALL | oplocks; /* * Root pools may need to read of the underlying devfs filesystem * when opening up a vdev. Unfortunately if we're holding the * SCL_ZIO lock it will result in a deadlock when we try to issue * the read from the root filesystem. Instead we "prefetch" * the associated vnodes that we need prior to opening the * underlying devices and cache them so that we can prevent * any I/O when we are doing the actual open. */ if (spa_is_root(spa)) { int low = locks & ~(SCL_ZIO - 1); int high = locks & ~low; spa_config_enter(spa, high, spa, RW_WRITER); vdev_hold(spa->spa_root_vdev); spa_config_enter(spa, low, spa, RW_WRITER); } else { spa_config_enter(spa, locks, spa, RW_WRITER); } spa->spa_vdev_locks = locks; } int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error) { boolean_t config_changed = B_FALSE; vdev_t *vdev_top; if (vd == NULL || vd == spa->spa_root_vdev) { vdev_top = spa->spa_root_vdev; } else { vdev_top = vd->vdev_top; } if (vd != NULL || error == 0) vdev_dtl_reassess(vdev_top, 0, 0, B_FALSE, B_FALSE); if (vd != NULL) { if (vd != spa->spa_root_vdev) vdev_state_dirty(vdev_top); config_changed = B_TRUE; spa->spa_config_generation++; } if (spa_is_root(spa)) vdev_rele(spa->spa_root_vdev); ASSERT3U(spa->spa_vdev_locks, >=, SCL_STATE_ALL); spa_config_exit(spa, spa->spa_vdev_locks, spa); /* * If anything changed, wait for it to sync. This ensures that, - * from the system administrator's perspective, zpool(1M) commands + * from the system administrator's perspective, zpool(8) commands * are synchronous. This is important for things like zpool offline: * when the command completes, you expect no further I/O from ZFS. */ if (vd != NULL) txg_wait_synced(spa->spa_dsl_pool, 0); /* * If the config changed, update the config cache. */ if (config_changed) { mutex_enter(&spa_namespace_lock); spa_write_cachefile(spa, B_FALSE, B_TRUE); mutex_exit(&spa_namespace_lock); } return (error); } /* * ========================================================================== * Miscellaneous functions * ========================================================================== */ void spa_activate_mos_feature(spa_t *spa, const char *feature, dmu_tx_t *tx) { if (!nvlist_exists(spa->spa_label_features, feature)) { fnvlist_add_boolean(spa->spa_label_features, feature); /* * When we are creating the pool (tx_txg==TXG_INITIAL), we can't * dirty the vdev config because lock SCL_CONFIG is not held. * Thankfully, in this case we don't need to dirty the config * because it will be written out anyway when we finish * creating the pool. */ if (tx->tx_txg != TXG_INITIAL) vdev_config_dirty(spa->spa_root_vdev); } } void spa_deactivate_mos_feature(spa_t *spa, const char *feature) { if (nvlist_remove_all(spa->spa_label_features, feature) == 0) vdev_config_dirty(spa->spa_root_vdev); } /* * Return the spa_t associated with given pool_guid, if it exists. If * device_guid is non-zero, determine whether the pool exists *and* contains * a device with the specified device_guid. */ spa_t * spa_by_guid(uint64_t pool_guid, uint64_t device_guid) { spa_t *spa; avl_tree_t *t = &spa_namespace_avl; ASSERT(MUTEX_HELD(&spa_namespace_lock)); for (spa = avl_first(t); spa != NULL; spa = AVL_NEXT(t, spa)) { if (spa->spa_state == POOL_STATE_UNINITIALIZED) continue; if (spa->spa_root_vdev == NULL) continue; if (spa_guid(spa) == pool_guid) { if (device_guid == 0) break; if (vdev_lookup_by_guid(spa->spa_root_vdev, device_guid) != NULL) break; /* * Check any devices we may be in the process of adding. */ if (spa->spa_pending_vdev) { if (vdev_lookup_by_guid(spa->spa_pending_vdev, device_guid) != NULL) break; } } } return (spa); } /* * Determine whether a pool with the given pool_guid exists. */ boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid) { return (spa_by_guid(pool_guid, device_guid) != NULL); } char * spa_strdup(const char *s) { size_t len; char *new; len = strlen(s); new = kmem_alloc(len + 1, KM_SLEEP); bcopy(s, new, len); new[len] = '\0'; return (new); } void spa_strfree(char *s) { kmem_free(s, strlen(s) + 1); } uint64_t spa_get_random(uint64_t range) { uint64_t r; ASSERT(range != 0); if (range == 1) return (0); (void) random_get_pseudo_bytes((void *)&r, sizeof (uint64_t)); return (r % range); } uint64_t spa_generate_guid(spa_t *spa) { uint64_t guid = spa_get_random(-1ULL); if (spa != NULL) { while (guid == 0 || spa_guid_exists(spa_guid(spa), guid)) guid = spa_get_random(-1ULL); } else { while (guid == 0 || spa_guid_exists(guid, 0)) guid = spa_get_random(-1ULL); } return (guid); } void snprintf_blkptr(char *buf, size_t buflen, const blkptr_t *bp) { char type[256]; char *checksum = NULL; char *compress = NULL; if (bp != NULL) { if (BP_GET_TYPE(bp) & DMU_OT_NEWTYPE) { dmu_object_byteswap_t bswap = DMU_OT_BYTESWAP(BP_GET_TYPE(bp)); (void) snprintf(type, sizeof (type), "bswap %s %s", DMU_OT_IS_METADATA(BP_GET_TYPE(bp)) ? "metadata" : "data", dmu_ot_byteswap[bswap].ob_name); } else { (void) strlcpy(type, dmu_ot[BP_GET_TYPE(bp)].ot_name, sizeof (type)); } if (!BP_IS_EMBEDDED(bp)) { checksum = zio_checksum_table[BP_GET_CHECKSUM(bp)].ci_name; } compress = zio_compress_table[BP_GET_COMPRESS(bp)].ci_name; } SNPRINTF_BLKPTR(snprintf, ' ', buf, buflen, bp, type, checksum, compress); } void spa_freeze(spa_t *spa) { uint64_t freeze_txg = 0; spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); if (spa->spa_freeze_txg == UINT64_MAX) { freeze_txg = spa_last_synced_txg(spa) + TXG_SIZE; spa->spa_freeze_txg = freeze_txg; } spa_config_exit(spa, SCL_ALL, FTAG); if (freeze_txg != 0) txg_wait_synced(spa_get_dsl(spa), freeze_txg); } void zfs_panic_recover(const char *fmt, ...) { va_list adx; va_start(adx, fmt); vcmn_err(zfs_recover ? CE_WARN : CE_PANIC, fmt, adx); va_end(adx); } /* * This is a stripped-down version of strtoull, suitable only for converting * lowercase hexadecimal numbers that don't overflow. */ uint64_t zfs_strtonum(const char *str, char **nptr) { uint64_t val = 0; char c; int digit; while ((c = *str) != '\0') { if (c >= '0' && c <= '9') digit = c - '0'; else if (c >= 'a' && c <= 'f') digit = 10 + c - 'a'; else break; val *= 16; val += digit; str++; } if (nptr) *nptr = (char *)str; return (val); } void spa_activate_allocation_classes(spa_t *spa, dmu_tx_t *tx) { /* * We bump the feature refcount for each special vdev added to the pool */ ASSERT(spa_feature_is_enabled(spa, SPA_FEATURE_ALLOCATION_CLASSES)); spa_feature_incr(spa, SPA_FEATURE_ALLOCATION_CLASSES, tx); } /* * ========================================================================== * Accessor functions * ========================================================================== */ boolean_t spa_shutting_down(spa_t *spa) { return (spa->spa_async_suspended); } dsl_pool_t * spa_get_dsl(spa_t *spa) { return (spa->spa_dsl_pool); } boolean_t spa_is_initializing(spa_t *spa) { return (spa->spa_is_initializing); } boolean_t spa_indirect_vdevs_loaded(spa_t *spa) { return (spa->spa_indirect_vdevs_loaded); } blkptr_t * spa_get_rootblkptr(spa_t *spa) { return (&spa->spa_ubsync.ub_rootbp); } void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp) { spa->spa_uberblock.ub_rootbp = *bp; } void spa_altroot(spa_t *spa, char *buf, size_t buflen) { if (spa->spa_root == NULL) buf[0] = '\0'; else (void) strncpy(buf, spa->spa_root, buflen); } int spa_sync_pass(spa_t *spa) { return (spa->spa_sync_pass); } char * spa_name(spa_t *spa) { return (spa->spa_name); } uint64_t spa_guid(spa_t *spa) { dsl_pool_t *dp = spa_get_dsl(spa); uint64_t guid; /* * If we fail to parse the config during spa_load(), we can go through * the error path (which posts an ereport) and end up here with no root * vdev. We stash the original pool guid in 'spa_config_guid' to handle * this case. */ if (spa->spa_root_vdev == NULL) return (spa->spa_config_guid); guid = spa->spa_last_synced_guid != 0 ? spa->spa_last_synced_guid : spa->spa_root_vdev->vdev_guid; /* * Return the most recently synced out guid unless we're * in syncing context. */ if (dp && dsl_pool_sync_context(dp)) return (spa->spa_root_vdev->vdev_guid); else return (guid); } uint64_t spa_load_guid(spa_t *spa) { /* * This is a GUID that exists solely as a reference for the * purposes of the arc. It is generated at load time, and * is never written to persistent storage. */ return (spa->spa_load_guid); } uint64_t spa_last_synced_txg(spa_t *spa) { return (spa->spa_ubsync.ub_txg); } uint64_t spa_first_txg(spa_t *spa) { return (spa->spa_first_txg); } uint64_t spa_syncing_txg(spa_t *spa) { return (spa->spa_syncing_txg); } /* * Return the last txg where data can be dirtied. The final txgs * will be used to just clear out any deferred frees that remain. */ uint64_t spa_final_dirty_txg(spa_t *spa) { return (spa->spa_final_txg - TXG_DEFER_SIZE); } pool_state_t spa_state(spa_t *spa) { return (spa->spa_state); } spa_load_state_t spa_load_state(spa_t *spa) { return (spa->spa_load_state); } uint64_t spa_freeze_txg(spa_t *spa) { return (spa->spa_freeze_txg); } /* * Return the inflated asize for a logical write in bytes. This is used by the * DMU to calculate the space a logical write will require on disk. * If lsize is smaller than the largest physical block size allocatable on this * pool we use its value instead, since the write will end up using the whole * block anyway. */ uint64_t spa_get_worst_case_asize(spa_t *spa, uint64_t lsize) { if (lsize == 0) return (0); /* No inflation needed */ return (MAX(lsize, 1 << spa->spa_max_ashift) * spa_asize_inflation); } /* * Return the amount of slop space in bytes. It is 1/32 of the pool (3.2%), * or at least 128MB, unless that would cause it to be more than half the * pool size. * * See the comment above spa_slop_shift for details. */ uint64_t spa_get_slop_space(spa_t *spa) { uint64_t space = spa_get_dspace(spa); return (MAX(space >> spa_slop_shift, MIN(space >> 1, spa_min_slop))); } uint64_t spa_get_dspace(spa_t *spa) { return (spa->spa_dspace); } uint64_t spa_get_checkpoint_space(spa_t *spa) { return (spa->spa_checkpoint_info.sci_dspace); } void spa_update_dspace(spa_t *spa) { spa->spa_dspace = metaslab_class_get_dspace(spa_normal_class(spa)) + ddt_get_dedup_dspace(spa); if (spa->spa_vdev_removal != NULL) { /* * We can't allocate from the removing device, so * subtract its size. This prevents the DMU/DSL from * filling up the (now smaller) pool while we are in the * middle of removing the device. * * Note that the DMU/DSL doesn't actually know or care * how much space is allocated (it does its own tracking * of how much space has been logically used). So it * doesn't matter that the data we are moving may be * allocated twice (on the old device and the new * device). */ spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); vdev_t *vd = vdev_lookup_top(spa, spa->spa_vdev_removal->svr_vdev_id); spa->spa_dspace -= spa_deflate(spa) ? vd->vdev_stat.vs_dspace : vd->vdev_stat.vs_space; spa_config_exit(spa, SCL_VDEV, FTAG); } } /* * Return the failure mode that has been set to this pool. The default * behavior will be to block all I/Os when a complete failure occurs. */ uint64_t spa_get_failmode(spa_t *spa) { return (spa->spa_failmode); } boolean_t spa_suspended(spa_t *spa) { return (spa->spa_suspended != ZIO_SUSPEND_NONE); } uint64_t spa_version(spa_t *spa) { return (spa->spa_ubsync.ub_version); } boolean_t spa_deflate(spa_t *spa) { return (spa->spa_deflate); } metaslab_class_t * spa_normal_class(spa_t *spa) { return (spa->spa_normal_class); } metaslab_class_t * spa_log_class(spa_t *spa) { return (spa->spa_log_class); } metaslab_class_t * spa_special_class(spa_t *spa) { return (spa->spa_special_class); } metaslab_class_t * spa_dedup_class(spa_t *spa) { return (spa->spa_dedup_class); } /* * Locate an appropriate allocation class */ metaslab_class_t * spa_preferred_class(spa_t *spa, uint64_t size, dmu_object_type_t objtype, uint_t level, uint_t special_smallblk) { if (DMU_OT_IS_ZIL(objtype)) { if (spa->spa_log_class->mc_groups != 0) return (spa_log_class(spa)); else return (spa_normal_class(spa)); } boolean_t has_special_class = spa->spa_special_class->mc_groups != 0; if (DMU_OT_IS_DDT(objtype)) { if (spa->spa_dedup_class->mc_groups != 0) return (spa_dedup_class(spa)); else if (has_special_class && zfs_ddt_data_is_special) return (spa_special_class(spa)); else return (spa_normal_class(spa)); } /* Indirect blocks for user data can land in special if allowed */ if (level > 0 && (DMU_OT_IS_FILE(objtype) || objtype == DMU_OT_ZVOL)) { if (has_special_class && zfs_user_indirect_is_special) return (spa_special_class(spa)); else return (spa_normal_class(spa)); } if (DMU_OT_IS_METADATA(objtype) || level > 0) { if (has_special_class) return (spa_special_class(spa)); else return (spa_normal_class(spa)); } /* * Allow small file blocks in special class in some cases (like * for the dRAID vdev feature). But always leave a reserve of * zfs_special_class_metadata_reserve_pct exclusively for metadata. */ if (DMU_OT_IS_FILE(objtype) && has_special_class && size <= special_smallblk) { metaslab_class_t *special = spa_special_class(spa); uint64_t alloc = metaslab_class_get_alloc(special); uint64_t space = metaslab_class_get_space(special); uint64_t limit = (space * (100 - zfs_special_class_metadata_reserve_pct)) / 100; if (alloc < limit) return (special); } return (spa_normal_class(spa)); } void spa_evicting_os_register(spa_t *spa, objset_t *os) { mutex_enter(&spa->spa_evicting_os_lock); list_insert_head(&spa->spa_evicting_os_list, os); mutex_exit(&spa->spa_evicting_os_lock); } void spa_evicting_os_deregister(spa_t *spa, objset_t *os) { mutex_enter(&spa->spa_evicting_os_lock); list_remove(&spa->spa_evicting_os_list, os); cv_broadcast(&spa->spa_evicting_os_cv); mutex_exit(&spa->spa_evicting_os_lock); } void spa_evicting_os_wait(spa_t *spa) { mutex_enter(&spa->spa_evicting_os_lock); while (!list_is_empty(&spa->spa_evicting_os_list)) cv_wait(&spa->spa_evicting_os_cv, &spa->spa_evicting_os_lock); mutex_exit(&spa->spa_evicting_os_lock); dmu_buf_user_evict_wait(); } int spa_max_replication(spa_t *spa) { /* * As of SPA_VERSION == SPA_VERSION_DITTO_BLOCKS, we are able to * handle BPs with more than one DVA allocated. Set our max * replication level accordingly. */ if (spa_version(spa) < SPA_VERSION_DITTO_BLOCKS) return (1); return (MIN(SPA_DVAS_PER_BP, spa_max_replication_override)); } int spa_prev_software_version(spa_t *spa) { return (spa->spa_prev_software_version); } uint64_t spa_deadman_synctime(spa_t *spa) { return (spa->spa_deadman_synctime); } spa_autotrim_t spa_get_autotrim(spa_t *spa) { return (spa->spa_autotrim); } uint64_t spa_deadman_ziotime(spa_t *spa) { return (spa->spa_deadman_ziotime); } uint64_t spa_get_deadman_failmode(spa_t *spa) { return (spa->spa_deadman_failmode); } void spa_set_deadman_failmode(spa_t *spa, const char *failmode) { if (strcmp(failmode, "wait") == 0) spa->spa_deadman_failmode = ZIO_FAILURE_MODE_WAIT; else if (strcmp(failmode, "continue") == 0) spa->spa_deadman_failmode = ZIO_FAILURE_MODE_CONTINUE; else if (strcmp(failmode, "panic") == 0) spa->spa_deadman_failmode = ZIO_FAILURE_MODE_PANIC; else spa->spa_deadman_failmode = ZIO_FAILURE_MODE_WAIT; } void spa_set_deadman_ziotime(hrtime_t ns) { spa_t *spa = NULL; if (spa_mode_global != SPA_MODE_UNINIT) { mutex_enter(&spa_namespace_lock); while ((spa = spa_next(spa)) != NULL) spa->spa_deadman_ziotime = ns; mutex_exit(&spa_namespace_lock); } } void spa_set_deadman_synctime(hrtime_t ns) { spa_t *spa = NULL; if (spa_mode_global != SPA_MODE_UNINIT) { mutex_enter(&spa_namespace_lock); while ((spa = spa_next(spa)) != NULL) spa->spa_deadman_synctime = ns; mutex_exit(&spa_namespace_lock); } } uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva) { uint64_t asize = DVA_GET_ASIZE(dva); uint64_t dsize = asize; ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0); if (asize != 0 && spa->spa_deflate) { vdev_t *vd = vdev_lookup_top(spa, DVA_GET_VDEV(dva)); if (vd != NULL) dsize = (asize >> SPA_MINBLOCKSHIFT) * vd->vdev_deflate_ratio; } return (dsize); } uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp) { uint64_t dsize = 0; for (int d = 0; d < BP_GET_NDVAS(bp); d++) dsize += dva_get_dsize_sync(spa, &bp->blk_dva[d]); return (dsize); } uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp) { uint64_t dsize = 0; spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); for (int d = 0; d < BP_GET_NDVAS(bp); d++) dsize += dva_get_dsize_sync(spa, &bp->blk_dva[d]); spa_config_exit(spa, SCL_VDEV, FTAG); return (dsize); } uint64_t spa_dirty_data(spa_t *spa) { return (spa->spa_dsl_pool->dp_dirty_total); } /* * ========================================================================== * SPA Import Progress Routines * ========================================================================== */ typedef struct spa_import_progress { uint64_t pool_guid; /* unique id for updates */ char *pool_name; spa_load_state_t spa_load_state; uint64_t mmp_sec_remaining; /* MMP activity check */ uint64_t spa_load_max_txg; /* rewind txg */ procfs_list_node_t smh_node; } spa_import_progress_t; spa_history_list_t *spa_import_progress_list = NULL; static int spa_import_progress_show_header(struct seq_file *f) { seq_printf(f, "%-20s %-14s %-14s %-12s %s\n", "pool_guid", "load_state", "multihost_secs", "max_txg", "pool_name"); return (0); } static int spa_import_progress_show(struct seq_file *f, void *data) { spa_import_progress_t *sip = (spa_import_progress_t *)data; seq_printf(f, "%-20llu %-14llu %-14llu %-12llu %s\n", (u_longlong_t)sip->pool_guid, (u_longlong_t)sip->spa_load_state, (u_longlong_t)sip->mmp_sec_remaining, (u_longlong_t)sip->spa_load_max_txg, (sip->pool_name ? sip->pool_name : "-")); return (0); } /* Remove oldest elements from list until there are no more than 'size' left */ static void spa_import_progress_truncate(spa_history_list_t *shl, unsigned int size) { spa_import_progress_t *sip; while (shl->size > size) { sip = list_remove_head(&shl->procfs_list.pl_list); if (sip->pool_name) spa_strfree(sip->pool_name); kmem_free(sip, sizeof (spa_import_progress_t)); shl->size--; } IMPLY(size == 0, list_is_empty(&shl->procfs_list.pl_list)); } static void spa_import_progress_init(void) { spa_import_progress_list = kmem_zalloc(sizeof (spa_history_list_t), KM_SLEEP); spa_import_progress_list->size = 0; spa_import_progress_list->procfs_list.pl_private = spa_import_progress_list; procfs_list_install("zfs", NULL, "import_progress", 0644, &spa_import_progress_list->procfs_list, spa_import_progress_show, spa_import_progress_show_header, NULL, offsetof(spa_import_progress_t, smh_node)); } static void spa_import_progress_destroy(void) { spa_history_list_t *shl = spa_import_progress_list; procfs_list_uninstall(&shl->procfs_list); spa_import_progress_truncate(shl, 0); procfs_list_destroy(&shl->procfs_list); kmem_free(shl, sizeof (spa_history_list_t)); } int spa_import_progress_set_state(uint64_t pool_guid, spa_load_state_t load_state) { spa_history_list_t *shl = spa_import_progress_list; spa_import_progress_t *sip; int error = ENOENT; if (shl->size == 0) return (0); mutex_enter(&shl->procfs_list.pl_lock); for (sip = list_tail(&shl->procfs_list.pl_list); sip != NULL; sip = list_prev(&shl->procfs_list.pl_list, sip)) { if (sip->pool_guid == pool_guid) { sip->spa_load_state = load_state; error = 0; break; } } mutex_exit(&shl->procfs_list.pl_lock); return (error); } int spa_import_progress_set_max_txg(uint64_t pool_guid, uint64_t load_max_txg) { spa_history_list_t *shl = spa_import_progress_list; spa_import_progress_t *sip; int error = ENOENT; if (shl->size == 0) return (0); mutex_enter(&shl->procfs_list.pl_lock); for (sip = list_tail(&shl->procfs_list.pl_list); sip != NULL; sip = list_prev(&shl->procfs_list.pl_list, sip)) { if (sip->pool_guid == pool_guid) { sip->spa_load_max_txg = load_max_txg; error = 0; break; } } mutex_exit(&shl->procfs_list.pl_lock); return (error); } int spa_import_progress_set_mmp_check(uint64_t pool_guid, uint64_t mmp_sec_remaining) { spa_history_list_t *shl = spa_import_progress_list; spa_import_progress_t *sip; int error = ENOENT; if (shl->size == 0) return (0); mutex_enter(&shl->procfs_list.pl_lock); for (sip = list_tail(&shl->procfs_list.pl_list); sip != NULL; sip = list_prev(&shl->procfs_list.pl_list, sip)) { if (sip->pool_guid == pool_guid) { sip->mmp_sec_remaining = mmp_sec_remaining; error = 0; break; } } mutex_exit(&shl->procfs_list.pl_lock); return (error); } /* * A new import is in progress, add an entry. */ void spa_import_progress_add(spa_t *spa) { spa_history_list_t *shl = spa_import_progress_list; spa_import_progress_t *sip; char *poolname = NULL; sip = kmem_zalloc(sizeof (spa_import_progress_t), KM_SLEEP); sip->pool_guid = spa_guid(spa); (void) nvlist_lookup_string(spa->spa_config, ZPOOL_CONFIG_POOL_NAME, &poolname); if (poolname == NULL) poolname = spa_name(spa); sip->pool_name = spa_strdup(poolname); sip->spa_load_state = spa_load_state(spa); mutex_enter(&shl->procfs_list.pl_lock); procfs_list_add(&shl->procfs_list, sip); shl->size++; mutex_exit(&shl->procfs_list.pl_lock); } void spa_import_progress_remove(uint64_t pool_guid) { spa_history_list_t *shl = spa_import_progress_list; spa_import_progress_t *sip; mutex_enter(&shl->procfs_list.pl_lock); for (sip = list_tail(&shl->procfs_list.pl_list); sip != NULL; sip = list_prev(&shl->procfs_list.pl_list, sip)) { if (sip->pool_guid == pool_guid) { if (sip->pool_name) spa_strfree(sip->pool_name); list_remove(&shl->procfs_list.pl_list, sip); shl->size--; kmem_free(sip, sizeof (spa_import_progress_t)); break; } } mutex_exit(&shl->procfs_list.pl_lock); } /* * ========================================================================== * Initialization and Termination * ========================================================================== */ static int spa_name_compare(const void *a1, const void *a2) { const spa_t *s1 = a1; const spa_t *s2 = a2; int s; s = strcmp(s1->spa_name, s2->spa_name); return (TREE_ISIGN(s)); } void spa_boot_init(void) { spa_config_load(); } void spa_init(spa_mode_t mode) { mutex_init(&spa_namespace_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa_spare_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa_l2cache_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&spa_namespace_cv, NULL, CV_DEFAULT, NULL); avl_create(&spa_namespace_avl, spa_name_compare, sizeof (spa_t), offsetof(spa_t, spa_avl)); avl_create(&spa_spare_avl, spa_spare_compare, sizeof (spa_aux_t), offsetof(spa_aux_t, aux_avl)); avl_create(&spa_l2cache_avl, spa_l2cache_compare, sizeof (spa_aux_t), offsetof(spa_aux_t, aux_avl)); spa_mode_global = mode; #ifndef _KERNEL if (spa_mode_global != SPA_MODE_READ && dprintf_find_string("watch")) { struct sigaction sa; sa.sa_flags = SA_SIGINFO; sigemptyset(&sa.sa_mask); sa.sa_sigaction = arc_buf_sigsegv; if (sigaction(SIGSEGV, &sa, NULL) == -1) { perror("could not enable watchpoints: " "sigaction(SIGSEGV, ...) = "); } else { arc_watch = B_TRUE; } } #endif fm_init(); zfs_refcount_init(); unique_init(); zfs_btree_init(); metaslab_stat_init(); ddt_init(); zio_init(); dmu_init(); zil_init(); vdev_cache_stat_init(); vdev_mirror_stat_init(); vdev_raidz_math_init(); vdev_file_init(); zfs_prop_init(); zpool_prop_init(); zpool_feature_init(); spa_config_load(); l2arc_start(); scan_init(); qat_init(); spa_import_progress_init(); } void spa_fini(void) { l2arc_stop(); spa_evict_all(); vdev_file_fini(); vdev_cache_stat_fini(); vdev_mirror_stat_fini(); vdev_raidz_math_fini(); zil_fini(); dmu_fini(); zio_fini(); ddt_fini(); metaslab_stat_fini(); zfs_btree_fini(); unique_fini(); zfs_refcount_fini(); fm_fini(); scan_fini(); qat_fini(); spa_import_progress_destroy(); avl_destroy(&spa_namespace_avl); avl_destroy(&spa_spare_avl); avl_destroy(&spa_l2cache_avl); cv_destroy(&spa_namespace_cv); mutex_destroy(&spa_namespace_lock); mutex_destroy(&spa_spare_lock); mutex_destroy(&spa_l2cache_lock); } /* * Return whether this pool has slogs. No locking needed. * It's not a problem if the wrong answer is returned as it's only for * performance and not correctness */ boolean_t spa_has_slogs(spa_t *spa) { return (spa->spa_log_class->mc_rotor != NULL); } spa_log_state_t spa_get_log_state(spa_t *spa) { return (spa->spa_log_state); } void spa_set_log_state(spa_t *spa, spa_log_state_t state) { spa->spa_log_state = state; } boolean_t spa_is_root(spa_t *spa) { return (spa->spa_is_root); } boolean_t spa_writeable(spa_t *spa) { return (!!(spa->spa_mode & SPA_MODE_WRITE) && spa->spa_trust_config); } /* * Returns true if there is a pending sync task in any of the current * syncing txg, the current quiescing txg, or the current open txg. */ boolean_t spa_has_pending_synctask(spa_t *spa) { return (!txg_all_lists_empty(&spa->spa_dsl_pool->dp_sync_tasks) || !txg_all_lists_empty(&spa->spa_dsl_pool->dp_early_sync_tasks)); } spa_mode_t spa_mode(spa_t *spa) { return (spa->spa_mode); } uint64_t spa_bootfs(spa_t *spa) { return (spa->spa_bootfs); } uint64_t spa_delegation(spa_t *spa) { return (spa->spa_delegation); } objset_t * spa_meta_objset(spa_t *spa) { return (spa->spa_meta_objset); } enum zio_checksum spa_dedup_checksum(spa_t *spa) { return (spa->spa_dedup_checksum); } /* * Reset pool scan stat per scan pass (or reboot). */ void spa_scan_stat_init(spa_t *spa) { /* data not stored on disk */ spa->spa_scan_pass_start = gethrestime_sec(); if (dsl_scan_is_paused_scrub(spa->spa_dsl_pool->dp_scan)) spa->spa_scan_pass_scrub_pause = spa->spa_scan_pass_start; else spa->spa_scan_pass_scrub_pause = 0; spa->spa_scan_pass_scrub_spent_paused = 0; spa->spa_scan_pass_exam = 0; spa->spa_scan_pass_issued = 0; vdev_scan_stat_init(spa->spa_root_vdev); } /* * Get scan stats for zpool status reports */ int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps) { dsl_scan_t *scn = spa->spa_dsl_pool ? spa->spa_dsl_pool->dp_scan : NULL; if (scn == NULL || scn->scn_phys.scn_func == POOL_SCAN_NONE) return (SET_ERROR(ENOENT)); bzero(ps, sizeof (pool_scan_stat_t)); /* data stored on disk */ ps->pss_func = scn->scn_phys.scn_func; ps->pss_state = scn->scn_phys.scn_state; ps->pss_start_time = scn->scn_phys.scn_start_time; ps->pss_end_time = scn->scn_phys.scn_end_time; ps->pss_to_examine = scn->scn_phys.scn_to_examine; ps->pss_examined = scn->scn_phys.scn_examined; ps->pss_to_process = scn->scn_phys.scn_to_process; ps->pss_processed = scn->scn_phys.scn_processed; ps->pss_errors = scn->scn_phys.scn_errors; /* data not stored on disk */ ps->pss_pass_exam = spa->spa_scan_pass_exam; ps->pss_pass_start = spa->spa_scan_pass_start; ps->pss_pass_scrub_pause = spa->spa_scan_pass_scrub_pause; ps->pss_pass_scrub_spent_paused = spa->spa_scan_pass_scrub_spent_paused; ps->pss_pass_issued = spa->spa_scan_pass_issued; ps->pss_issued = scn->scn_issued_before_pass + spa->spa_scan_pass_issued; return (0); } int spa_maxblocksize(spa_t *spa) { if (spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_BLOCKS)) return (SPA_MAXBLOCKSIZE); else return (SPA_OLD_MAXBLOCKSIZE); } /* * Returns the txg that the last device removal completed. No indirect mappings * have been added since this txg. */ uint64_t spa_get_last_removal_txg(spa_t *spa) { uint64_t vdevid; uint64_t ret = -1ULL; spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); /* * sr_prev_indirect_vdev is only modified while holding all the * config locks, so it is sufficient to hold SCL_VDEV as reader when * examining it. */ vdevid = spa->spa_removing_phys.sr_prev_indirect_vdev; while (vdevid != -1ULL) { vdev_t *vd = vdev_lookup_top(spa, vdevid); vdev_indirect_births_t *vib = vd->vdev_indirect_births; ASSERT3P(vd->vdev_ops, ==, &vdev_indirect_ops); /* * If the removal did not remap any data, we don't care. */ if (vdev_indirect_births_count(vib) != 0) { ret = vdev_indirect_births_last_entry_txg(vib); break; } vdevid = vd->vdev_indirect_config.vic_prev_indirect_vdev; } spa_config_exit(spa, SCL_VDEV, FTAG); IMPLY(ret != -1ULL, spa_feature_is_active(spa, SPA_FEATURE_DEVICE_REMOVAL)); return (ret); } int spa_maxdnodesize(spa_t *spa) { if (spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_DNODE)) return (DNODE_MAX_SIZE); else return (DNODE_MIN_SIZE); } boolean_t spa_multihost(spa_t *spa) { return (spa->spa_multihost ? B_TRUE : B_FALSE); } uint32_t spa_get_hostid(spa_t *spa) { return (spa->spa_hostid); } boolean_t spa_trust_config(spa_t *spa) { return (spa->spa_trust_config); } uint64_t spa_missing_tvds_allowed(spa_t *spa) { return (spa->spa_missing_tvds_allowed); } space_map_t * spa_syncing_log_sm(spa_t *spa) { return (spa->spa_syncing_log_sm); } void spa_set_missing_tvds(spa_t *spa, uint64_t missing) { spa->spa_missing_tvds = missing; } /* * Return the pool state string ("ONLINE", "DEGRADED", "SUSPENDED", etc). */ const char * spa_state_to_name(spa_t *spa) { ASSERT3P(spa, !=, NULL); /* * it is possible for the spa to exist, without root vdev * as the spa transitions during import/export */ vdev_t *rvd = spa->spa_root_vdev; if (rvd == NULL) { return ("TRANSITIONING"); } vdev_state_t state = rvd->vdev_state; vdev_aux_t aux = rvd->vdev_stat.vs_aux; if (spa_suspended(spa) && (spa_get_failmode(spa) != ZIO_FAILURE_MODE_CONTINUE)) return ("SUSPENDED"); switch (state) { case VDEV_STATE_CLOSED: case VDEV_STATE_OFFLINE: return ("OFFLINE"); case VDEV_STATE_REMOVED: return ("REMOVED"); case VDEV_STATE_CANT_OPEN: if (aux == VDEV_AUX_CORRUPT_DATA || aux == VDEV_AUX_BAD_LOG) return ("FAULTED"); else if (aux == VDEV_AUX_SPLIT_POOL) return ("SPLIT"); else return ("UNAVAIL"); case VDEV_STATE_FAULTED: return ("FAULTED"); case VDEV_STATE_DEGRADED: return ("DEGRADED"); case VDEV_STATE_HEALTHY: return ("ONLINE"); default: break; } return ("UNKNOWN"); } boolean_t spa_top_vdevs_spacemap_addressable(spa_t *spa) { vdev_t *rvd = spa->spa_root_vdev; for (uint64_t c = 0; c < rvd->vdev_children; c++) { if (!vdev_is_spacemap_addressable(rvd->vdev_child[c])) return (B_FALSE); } return (B_TRUE); } boolean_t spa_has_checkpoint(spa_t *spa) { return (spa->spa_checkpoint_txg != 0); } boolean_t spa_importing_readonly_checkpoint(spa_t *spa) { return ((spa->spa_import_flags & ZFS_IMPORT_CHECKPOINT) && spa->spa_mode == SPA_MODE_READ); } uint64_t spa_min_claim_txg(spa_t *spa) { uint64_t checkpoint_txg = spa->spa_uberblock.ub_checkpoint_txg; if (checkpoint_txg != 0) return (checkpoint_txg + 1); return (spa->spa_first_txg); } /* * If there is a checkpoint, async destroys may consume more space from * the pool instead of freeing it. In an attempt to save the pool from * getting suspended when it is about to run out of space, we stop * processing async destroys. */ boolean_t spa_suspend_async_destroy(spa_t *spa) { dsl_pool_t *dp = spa_get_dsl(spa); uint64_t unreserved = dsl_pool_unreserved_space(dp, ZFS_SPACE_CHECK_EXTRA_RESERVED); uint64_t used = dsl_dir_phys(dp->dp_root_dir)->dd_used_bytes; uint64_t avail = (unreserved > used) ? (unreserved - used) : 0; if (spa_has_checkpoint(spa) && avail == 0) return (B_TRUE); return (B_FALSE); } #if defined(_KERNEL) int param_set_deadman_failmode_common(const char *val) { spa_t *spa = NULL; char *p; if (val == NULL) return (SET_ERROR(EINVAL)); if ((p = strchr(val, '\n')) != NULL) *p = '\0'; if (strcmp(val, "wait") != 0 && strcmp(val, "continue") != 0 && strcmp(val, "panic")) return (SET_ERROR(EINVAL)); if (spa_mode_global != SPA_MODE_UNINIT) { mutex_enter(&spa_namespace_lock); while ((spa = spa_next(spa)) != NULL) spa_set_deadman_failmode(spa, val); mutex_exit(&spa_namespace_lock); } return (0); } #endif /* Namespace manipulation */ EXPORT_SYMBOL(spa_lookup); EXPORT_SYMBOL(spa_add); EXPORT_SYMBOL(spa_remove); EXPORT_SYMBOL(spa_next); /* Refcount functions */ EXPORT_SYMBOL(spa_open_ref); EXPORT_SYMBOL(spa_close); EXPORT_SYMBOL(spa_refcount_zero); /* Pool configuration lock */ EXPORT_SYMBOL(spa_config_tryenter); EXPORT_SYMBOL(spa_config_enter); EXPORT_SYMBOL(spa_config_exit); EXPORT_SYMBOL(spa_config_held); /* Pool vdev add/remove lock */ EXPORT_SYMBOL(spa_vdev_enter); EXPORT_SYMBOL(spa_vdev_exit); /* Pool vdev state change lock */ EXPORT_SYMBOL(spa_vdev_state_enter); EXPORT_SYMBOL(spa_vdev_state_exit); /* Accessor functions */ EXPORT_SYMBOL(spa_shutting_down); EXPORT_SYMBOL(spa_get_dsl); EXPORT_SYMBOL(spa_get_rootblkptr); EXPORT_SYMBOL(spa_set_rootblkptr); EXPORT_SYMBOL(spa_altroot); EXPORT_SYMBOL(spa_sync_pass); EXPORT_SYMBOL(spa_name); EXPORT_SYMBOL(spa_guid); EXPORT_SYMBOL(spa_last_synced_txg); EXPORT_SYMBOL(spa_first_txg); EXPORT_SYMBOL(spa_syncing_txg); EXPORT_SYMBOL(spa_version); EXPORT_SYMBOL(spa_state); EXPORT_SYMBOL(spa_load_state); EXPORT_SYMBOL(spa_freeze_txg); EXPORT_SYMBOL(spa_get_dspace); EXPORT_SYMBOL(spa_update_dspace); EXPORT_SYMBOL(spa_deflate); EXPORT_SYMBOL(spa_normal_class); EXPORT_SYMBOL(spa_log_class); EXPORT_SYMBOL(spa_special_class); EXPORT_SYMBOL(spa_preferred_class); EXPORT_SYMBOL(spa_max_replication); EXPORT_SYMBOL(spa_prev_software_version); EXPORT_SYMBOL(spa_get_failmode); EXPORT_SYMBOL(spa_suspended); EXPORT_SYMBOL(spa_bootfs); EXPORT_SYMBOL(spa_delegation); EXPORT_SYMBOL(spa_meta_objset); EXPORT_SYMBOL(spa_maxblocksize); EXPORT_SYMBOL(spa_maxdnodesize); /* Miscellaneous support routines */ EXPORT_SYMBOL(spa_guid_exists); EXPORT_SYMBOL(spa_strdup); EXPORT_SYMBOL(spa_strfree); EXPORT_SYMBOL(spa_get_random); EXPORT_SYMBOL(spa_generate_guid); EXPORT_SYMBOL(snprintf_blkptr); EXPORT_SYMBOL(spa_freeze); EXPORT_SYMBOL(spa_upgrade); EXPORT_SYMBOL(spa_evict_all); EXPORT_SYMBOL(spa_lookup_by_guid); EXPORT_SYMBOL(spa_has_spare); EXPORT_SYMBOL(dva_get_dsize_sync); EXPORT_SYMBOL(bp_get_dsize_sync); EXPORT_SYMBOL(bp_get_dsize); EXPORT_SYMBOL(spa_has_slogs); EXPORT_SYMBOL(spa_is_root); EXPORT_SYMBOL(spa_writeable); EXPORT_SYMBOL(spa_mode); EXPORT_SYMBOL(spa_namespace_lock); EXPORT_SYMBOL(spa_trust_config); EXPORT_SYMBOL(spa_missing_tvds_allowed); EXPORT_SYMBOL(spa_set_missing_tvds); EXPORT_SYMBOL(spa_state_to_name); EXPORT_SYMBOL(spa_importing_readonly_checkpoint); EXPORT_SYMBOL(spa_min_claim_txg); EXPORT_SYMBOL(spa_suspend_async_destroy); EXPORT_SYMBOL(spa_has_checkpoint); EXPORT_SYMBOL(spa_top_vdevs_spacemap_addressable); ZFS_MODULE_PARAM(zfs, zfs_, flags, UINT, ZMOD_RW, "Set additional debugging flags"); ZFS_MODULE_PARAM(zfs, zfs_, recover, INT, ZMOD_RW, "Set to attempt to recover from fatal errors"); ZFS_MODULE_PARAM(zfs, zfs_, free_leak_on_eio, INT, ZMOD_RW, "Set to ignore IO errors during free and permanently leak the space"); ZFS_MODULE_PARAM(zfs, zfs_, deadman_checktime_ms, ULONG, ZMOD_RW, "Dead I/O check interval in milliseconds"); ZFS_MODULE_PARAM(zfs, zfs_, deadman_enabled, INT, ZMOD_RW, "Enable deadman timer"); ZFS_MODULE_PARAM(zfs_spa, spa_, asize_inflation, INT, ZMOD_RW, "SPA size estimate multiplication factor"); ZFS_MODULE_PARAM(zfs, zfs_, ddt_data_is_special, INT, ZMOD_RW, "Place DDT data into the special class"); ZFS_MODULE_PARAM(zfs, zfs_, user_indirect_is_special, INT, ZMOD_RW, "Place user data indirect blocks into the special class"); /* BEGIN CSTYLED */ ZFS_MODULE_PARAM_CALL(zfs_deadman, zfs_deadman_, failmode, param_set_deadman_failmode, param_get_charp, ZMOD_RW, "Failmode for deadman timer"); ZFS_MODULE_PARAM_CALL(zfs_deadman, zfs_deadman_, synctime_ms, param_set_deadman_synctime, param_get_ulong, ZMOD_RW, "Pool sync expiration time in milliseconds"); ZFS_MODULE_PARAM_CALL(zfs_deadman, zfs_deadman_, ziotime_ms, param_set_deadman_ziotime, param_get_ulong, ZMOD_RW, "IO expiration time in milliseconds"); ZFS_MODULE_PARAM(zfs, zfs_, special_class_metadata_reserve_pct, INT, ZMOD_RW, "Small file blocks in special vdevs depends on this much " "free space available"); /* END CSTYLED */ ZFS_MODULE_PARAM_CALL(zfs_spa, spa_, slop_shift, param_set_slop_shift, param_get_int, ZMOD_RW, "Reserved free space in pool"); diff --git a/module/zfs/vdev.c b/module/zfs/vdev.c index 6af61cdcd9bf..e41e79ab8a18 100644 --- a/module/zfs/vdev.c +++ b/module/zfs/vdev.c @@ -1,5100 +1,5100 @@ /* * 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) 2011, 2020 by Delphix. All rights reserved. * Copyright 2017 Nexenta Systems, Inc. * Copyright (c) 2014 Integros [integros.com] * Copyright 2016 Toomas Soome * Copyright 2017 Joyent, Inc. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2019, Datto Inc. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* default target for number of metaslabs per top-level vdev */ int zfs_vdev_default_ms_count = 200; /* minimum number of metaslabs per top-level vdev */ int zfs_vdev_min_ms_count = 16; /* practical upper limit of total metaslabs per top-level vdev */ int zfs_vdev_ms_count_limit = 1ULL << 17; /* lower limit for metaslab size (512M) */ int zfs_vdev_default_ms_shift = 29; /* upper limit for metaslab size (16G) */ int zfs_vdev_max_ms_shift = 34; int vdev_validate_skip = B_FALSE; /* * Since the DTL space map of a vdev is not expected to have a lot of * entries, we default its block size to 4K. */ int zfs_vdev_dtl_sm_blksz = (1 << 12); /* * Rate limit slow IO (delay) events to this many per second. */ unsigned int zfs_slow_io_events_per_second = 20; /* * Rate limit checksum events after this many checksum errors per second. */ unsigned int zfs_checksum_events_per_second = 20; /* * Ignore errors during scrub/resilver. Allows to work around resilver * upon import when there are pool errors. */ int zfs_scan_ignore_errors = 0; /* * vdev-wide space maps that have lots of entries written to them at * the end of each transaction can benefit from a higher I/O bandwidth * (e.g. vdev_obsolete_sm), thus we default their block size to 128K. */ int zfs_vdev_standard_sm_blksz = (1 << 17); /* * Tunable parameter for debugging or performance analysis. Setting this * will cause pool corruption on power loss if a volatile out-of-order * write cache is enabled. */ int zfs_nocacheflush = 0; uint64_t zfs_vdev_max_auto_ashift = ASHIFT_MAX; uint64_t zfs_vdev_min_auto_ashift = ASHIFT_MIN; /*PRINTFLIKE2*/ void vdev_dbgmsg(vdev_t *vd, const char *fmt, ...) { va_list adx; char buf[256]; va_start(adx, fmt); (void) vsnprintf(buf, sizeof (buf), fmt, adx); va_end(adx); if (vd->vdev_path != NULL) { zfs_dbgmsg("%s vdev '%s': %s", vd->vdev_ops->vdev_op_type, vd->vdev_path, buf); } else { zfs_dbgmsg("%s-%llu vdev (guid %llu): %s", vd->vdev_ops->vdev_op_type, (u_longlong_t)vd->vdev_id, (u_longlong_t)vd->vdev_guid, buf); } } void vdev_dbgmsg_print_tree(vdev_t *vd, int indent) { char state[20]; if (vd->vdev_ishole || vd->vdev_ops == &vdev_missing_ops) { zfs_dbgmsg("%*svdev %u: %s", indent, "", vd->vdev_id, vd->vdev_ops->vdev_op_type); return; } switch (vd->vdev_state) { case VDEV_STATE_UNKNOWN: (void) snprintf(state, sizeof (state), "unknown"); break; case VDEV_STATE_CLOSED: (void) snprintf(state, sizeof (state), "closed"); break; case VDEV_STATE_OFFLINE: (void) snprintf(state, sizeof (state), "offline"); break; case VDEV_STATE_REMOVED: (void) snprintf(state, sizeof (state), "removed"); break; case VDEV_STATE_CANT_OPEN: (void) snprintf(state, sizeof (state), "can't open"); break; case VDEV_STATE_FAULTED: (void) snprintf(state, sizeof (state), "faulted"); break; case VDEV_STATE_DEGRADED: (void) snprintf(state, sizeof (state), "degraded"); break; case VDEV_STATE_HEALTHY: (void) snprintf(state, sizeof (state), "healthy"); break; default: (void) snprintf(state, sizeof (state), "", (uint_t)vd->vdev_state); } zfs_dbgmsg("%*svdev %u: %s%s, guid: %llu, path: %s, %s", indent, "", (int)vd->vdev_id, vd->vdev_ops->vdev_op_type, vd->vdev_islog ? " (log)" : "", (u_longlong_t)vd->vdev_guid, vd->vdev_path ? vd->vdev_path : "N/A", state); for (uint64_t i = 0; i < vd->vdev_children; i++) vdev_dbgmsg_print_tree(vd->vdev_child[i], indent + 2); } /* * Virtual device management. */ static vdev_ops_t *vdev_ops_table[] = { &vdev_root_ops, &vdev_raidz_ops, &vdev_mirror_ops, &vdev_replacing_ops, &vdev_spare_ops, &vdev_disk_ops, &vdev_file_ops, &vdev_missing_ops, &vdev_hole_ops, &vdev_indirect_ops, NULL }; /* * Given a vdev type, return the appropriate ops vector. */ static vdev_ops_t * vdev_getops(const char *type) { vdev_ops_t *ops, **opspp; for (opspp = vdev_ops_table; (ops = *opspp) != NULL; opspp++) if (strcmp(ops->vdev_op_type, type) == 0) break; return (ops); } /* ARGSUSED */ void vdev_default_xlate(vdev_t *vd, const range_seg64_t *in, range_seg64_t *res) { res->rs_start = in->rs_start; res->rs_end = in->rs_end; } /* * Derive the enumerated allocation bias from string input. - * String origin is either the per-vdev zap or zpool(1M). + * String origin is either the per-vdev zap or zpool(8). */ static vdev_alloc_bias_t vdev_derive_alloc_bias(const char *bias) { vdev_alloc_bias_t alloc_bias = VDEV_BIAS_NONE; if (strcmp(bias, VDEV_ALLOC_BIAS_LOG) == 0) alloc_bias = VDEV_BIAS_LOG; else if (strcmp(bias, VDEV_ALLOC_BIAS_SPECIAL) == 0) alloc_bias = VDEV_BIAS_SPECIAL; else if (strcmp(bias, VDEV_ALLOC_BIAS_DEDUP) == 0) alloc_bias = VDEV_BIAS_DEDUP; return (alloc_bias); } /* * Default asize function: return the MAX of psize with the asize of * all children. This is what's used by anything other than RAID-Z. */ uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize) { uint64_t asize = P2ROUNDUP(psize, 1ULL << vd->vdev_top->vdev_ashift); uint64_t csize; for (int c = 0; c < vd->vdev_children; c++) { csize = vdev_psize_to_asize(vd->vdev_child[c], psize); asize = MAX(asize, csize); } return (asize); } /* * Get the minimum allocatable size. We define the allocatable size as * the vdev's asize rounded to the nearest metaslab. This allows us to * replace or attach devices which don't have the same physical size but * can still satisfy the same number of allocations. */ uint64_t vdev_get_min_asize(vdev_t *vd) { vdev_t *pvd = vd->vdev_parent; /* * If our parent is NULL (inactive spare or cache) or is the root, * just return our own asize. */ if (pvd == NULL) return (vd->vdev_asize); /* * The top-level vdev just returns the allocatable size rounded * to the nearest metaslab. */ if (vd == vd->vdev_top) return (P2ALIGN(vd->vdev_asize, 1ULL << vd->vdev_ms_shift)); /* * The allocatable space for a raidz vdev is N * sizeof(smallest child), * so each child must provide at least 1/Nth of its asize. */ if (pvd->vdev_ops == &vdev_raidz_ops) return ((pvd->vdev_min_asize + pvd->vdev_children - 1) / pvd->vdev_children); return (pvd->vdev_min_asize); } void vdev_set_min_asize(vdev_t *vd) { vd->vdev_min_asize = vdev_get_min_asize(vd); for (int c = 0; c < vd->vdev_children; c++) vdev_set_min_asize(vd->vdev_child[c]); } vdev_t * vdev_lookup_top(spa_t *spa, uint64_t vdev) { vdev_t *rvd = spa->spa_root_vdev; ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0); if (vdev < rvd->vdev_children) { ASSERT(rvd->vdev_child[vdev] != NULL); return (rvd->vdev_child[vdev]); } return (NULL); } vdev_t * vdev_lookup_by_guid(vdev_t *vd, uint64_t guid) { vdev_t *mvd; if (vd->vdev_guid == guid) return (vd); for (int c = 0; c < vd->vdev_children; c++) if ((mvd = vdev_lookup_by_guid(vd->vdev_child[c], guid)) != NULL) return (mvd); return (NULL); } static int vdev_count_leaves_impl(vdev_t *vd) { int n = 0; if (vd->vdev_ops->vdev_op_leaf) return (1); for (int c = 0; c < vd->vdev_children; c++) n += vdev_count_leaves_impl(vd->vdev_child[c]); return (n); } int vdev_count_leaves(spa_t *spa) { int rc; spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); rc = vdev_count_leaves_impl(spa->spa_root_vdev); spa_config_exit(spa, SCL_VDEV, FTAG); return (rc); } void vdev_add_child(vdev_t *pvd, vdev_t *cvd) { size_t oldsize, newsize; uint64_t id = cvd->vdev_id; vdev_t **newchild; ASSERT(spa_config_held(cvd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL); ASSERT(cvd->vdev_parent == NULL); cvd->vdev_parent = pvd; if (pvd == NULL) return; ASSERT(id >= pvd->vdev_children || pvd->vdev_child[id] == NULL); oldsize = pvd->vdev_children * sizeof (vdev_t *); pvd->vdev_children = MAX(pvd->vdev_children, id + 1); newsize = pvd->vdev_children * sizeof (vdev_t *); newchild = kmem_alloc(newsize, KM_SLEEP); if (pvd->vdev_child != NULL) { bcopy(pvd->vdev_child, newchild, oldsize); kmem_free(pvd->vdev_child, oldsize); } pvd->vdev_child = newchild; pvd->vdev_child[id] = cvd; cvd->vdev_top = (pvd->vdev_top ? pvd->vdev_top: cvd); ASSERT(cvd->vdev_top->vdev_parent->vdev_parent == NULL); /* * Walk up all ancestors to update guid sum. */ for (; pvd != NULL; pvd = pvd->vdev_parent) pvd->vdev_guid_sum += cvd->vdev_guid_sum; if (cvd->vdev_ops->vdev_op_leaf) { list_insert_head(&cvd->vdev_spa->spa_leaf_list, cvd); cvd->vdev_spa->spa_leaf_list_gen++; } } void vdev_remove_child(vdev_t *pvd, vdev_t *cvd) { int c; uint_t id = cvd->vdev_id; ASSERT(cvd->vdev_parent == pvd); if (pvd == NULL) return; ASSERT(id < pvd->vdev_children); ASSERT(pvd->vdev_child[id] == cvd); pvd->vdev_child[id] = NULL; cvd->vdev_parent = NULL; for (c = 0; c < pvd->vdev_children; c++) if (pvd->vdev_child[c]) break; if (c == pvd->vdev_children) { kmem_free(pvd->vdev_child, c * sizeof (vdev_t *)); pvd->vdev_child = NULL; pvd->vdev_children = 0; } if (cvd->vdev_ops->vdev_op_leaf) { spa_t *spa = cvd->vdev_spa; list_remove(&spa->spa_leaf_list, cvd); spa->spa_leaf_list_gen++; } /* * Walk up all ancestors to update guid sum. */ for (; pvd != NULL; pvd = pvd->vdev_parent) pvd->vdev_guid_sum -= cvd->vdev_guid_sum; } /* * Remove any holes in the child array. */ void vdev_compact_children(vdev_t *pvd) { vdev_t **newchild, *cvd; int oldc = pvd->vdev_children; int newc; ASSERT(spa_config_held(pvd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL); if (oldc == 0) return; for (int c = newc = 0; c < oldc; c++) if (pvd->vdev_child[c]) newc++; if (newc > 0) { newchild = kmem_zalloc(newc * sizeof (vdev_t *), KM_SLEEP); for (int c = newc = 0; c < oldc; c++) { if ((cvd = pvd->vdev_child[c]) != NULL) { newchild[newc] = cvd; cvd->vdev_id = newc++; } } } else { newchild = NULL; } kmem_free(pvd->vdev_child, oldc * sizeof (vdev_t *)); pvd->vdev_child = newchild; pvd->vdev_children = newc; } /* * Allocate and minimally initialize a vdev_t. */ vdev_t * vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid, vdev_ops_t *ops) { vdev_t *vd; vdev_indirect_config_t *vic; vd = kmem_zalloc(sizeof (vdev_t), KM_SLEEP); vic = &vd->vdev_indirect_config; if (spa->spa_root_vdev == NULL) { ASSERT(ops == &vdev_root_ops); spa->spa_root_vdev = vd; spa->spa_load_guid = spa_generate_guid(NULL); } if (guid == 0 && ops != &vdev_hole_ops) { if (spa->spa_root_vdev == vd) { /* * The root vdev's guid will also be the pool guid, * which must be unique among all pools. */ guid = spa_generate_guid(NULL); } else { /* * Any other vdev's guid must be unique within the pool. */ guid = spa_generate_guid(spa); } ASSERT(!spa_guid_exists(spa_guid(spa), guid)); } vd->vdev_spa = spa; vd->vdev_id = id; vd->vdev_guid = guid; vd->vdev_guid_sum = guid; vd->vdev_ops = ops; vd->vdev_state = VDEV_STATE_CLOSED; vd->vdev_ishole = (ops == &vdev_hole_ops); vic->vic_prev_indirect_vdev = UINT64_MAX; rw_init(&vd->vdev_indirect_rwlock, NULL, RW_DEFAULT, NULL); mutex_init(&vd->vdev_obsolete_lock, NULL, MUTEX_DEFAULT, NULL); vd->vdev_obsolete_segments = range_tree_create(NULL, RANGE_SEG64, NULL, 0, 0); /* * Initialize rate limit structs for events. We rate limit ZIO delay * and checksum events so that we don't overwhelm ZED with thousands * of events when a disk is acting up. */ zfs_ratelimit_init(&vd->vdev_delay_rl, &zfs_slow_io_events_per_second, 1); zfs_ratelimit_init(&vd->vdev_checksum_rl, &zfs_checksum_events_per_second, 1); list_link_init(&vd->vdev_config_dirty_node); list_link_init(&vd->vdev_state_dirty_node); list_link_init(&vd->vdev_initialize_node); list_link_init(&vd->vdev_leaf_node); list_link_init(&vd->vdev_trim_node); mutex_init(&vd->vdev_dtl_lock, NULL, MUTEX_NOLOCKDEP, NULL); mutex_init(&vd->vdev_stat_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_probe_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_scan_io_queue_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_initialize_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_initialize_io_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&vd->vdev_initialize_cv, NULL, CV_DEFAULT, NULL); cv_init(&vd->vdev_initialize_io_cv, NULL, CV_DEFAULT, NULL); mutex_init(&vd->vdev_trim_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_autotrim_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_trim_io_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&vd->vdev_trim_cv, NULL, CV_DEFAULT, NULL); cv_init(&vd->vdev_autotrim_cv, NULL, CV_DEFAULT, NULL); cv_init(&vd->vdev_trim_io_cv, NULL, CV_DEFAULT, NULL); mutex_init(&vd->vdev_rebuild_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_rebuild_io_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&vd->vdev_rebuild_cv, NULL, CV_DEFAULT, NULL); cv_init(&vd->vdev_rebuild_io_cv, NULL, CV_DEFAULT, NULL); for (int t = 0; t < DTL_TYPES; t++) { vd->vdev_dtl[t] = range_tree_create(NULL, RANGE_SEG64, NULL, 0, 0); } txg_list_create(&vd->vdev_ms_list, spa, offsetof(struct metaslab, ms_txg_node)); txg_list_create(&vd->vdev_dtl_list, spa, offsetof(struct vdev, vdev_dtl_node)); vd->vdev_stat.vs_timestamp = gethrtime(); vdev_queue_init(vd); vdev_cache_init(vd); return (vd); } /* * Allocate a new vdev. The 'alloctype' is used to control whether we are * creating a new vdev or loading an existing one - the behavior is slightly * different for each case. */ int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent, uint_t id, int alloctype) { vdev_ops_t *ops; char *type; uint64_t guid = 0, islog, nparity; vdev_t *vd; vdev_indirect_config_t *vic; char *tmp = NULL; int rc; vdev_alloc_bias_t alloc_bias = VDEV_BIAS_NONE; boolean_t top_level = (parent && !parent->vdev_parent); ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); if (nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) != 0) return (SET_ERROR(EINVAL)); if ((ops = vdev_getops(type)) == NULL) return (SET_ERROR(EINVAL)); /* * If this is a load, get the vdev guid from the nvlist. * Otherwise, vdev_alloc_common() will generate one for us. */ if (alloctype == VDEV_ALLOC_LOAD) { uint64_t label_id; if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ID, &label_id) || label_id != id) return (SET_ERROR(EINVAL)); if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0) return (SET_ERROR(EINVAL)); } else if (alloctype == VDEV_ALLOC_SPARE) { if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0) return (SET_ERROR(EINVAL)); } else if (alloctype == VDEV_ALLOC_L2CACHE) { if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0) return (SET_ERROR(EINVAL)); } else if (alloctype == VDEV_ALLOC_ROOTPOOL) { if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0) return (SET_ERROR(EINVAL)); } /* * The first allocated vdev must be of type 'root'. */ if (ops != &vdev_root_ops && spa->spa_root_vdev == NULL) return (SET_ERROR(EINVAL)); /* * Determine whether we're a log vdev. */ islog = 0; (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_LOG, &islog); if (islog && spa_version(spa) < SPA_VERSION_SLOGS) return (SET_ERROR(ENOTSUP)); if (ops == &vdev_hole_ops && spa_version(spa) < SPA_VERSION_HOLES) return (SET_ERROR(ENOTSUP)); /* * Set the nparity property for RAID-Z vdevs. */ nparity = -1ULL; if (ops == &vdev_raidz_ops) { if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NPARITY, &nparity) == 0) { if (nparity == 0 || nparity > VDEV_RAIDZ_MAXPARITY) return (SET_ERROR(EINVAL)); /* * Previous versions could only support 1 or 2 parity * device. */ if (nparity > 1 && spa_version(spa) < SPA_VERSION_RAIDZ2) return (SET_ERROR(ENOTSUP)); if (nparity > 2 && spa_version(spa) < SPA_VERSION_RAIDZ3) return (SET_ERROR(ENOTSUP)); } else { /* * We require the parity to be specified for SPAs that * support multiple parity levels. */ if (spa_version(spa) >= SPA_VERSION_RAIDZ2) return (SET_ERROR(EINVAL)); /* * Otherwise, we default to 1 parity device for RAID-Z. */ nparity = 1; } } else { nparity = 0; } ASSERT(nparity != -1ULL); /* * If creating a top-level vdev, check for allocation classes input */ if (top_level && alloctype == VDEV_ALLOC_ADD) { char *bias; if (nvlist_lookup_string(nv, ZPOOL_CONFIG_ALLOCATION_BIAS, &bias) == 0) { alloc_bias = vdev_derive_alloc_bias(bias); /* spa_vdev_add() expects feature to be enabled */ if (spa->spa_load_state != SPA_LOAD_CREATE && !spa_feature_is_enabled(spa, SPA_FEATURE_ALLOCATION_CLASSES)) { return (SET_ERROR(ENOTSUP)); } } } vd = vdev_alloc_common(spa, id, guid, ops); vic = &vd->vdev_indirect_config; vd->vdev_islog = islog; vd->vdev_nparity = nparity; if (top_level && alloc_bias != VDEV_BIAS_NONE) vd->vdev_alloc_bias = alloc_bias; if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &vd->vdev_path) == 0) vd->vdev_path = spa_strdup(vd->vdev_path); /* * ZPOOL_CONFIG_AUX_STATE = "external" means we previously forced a * fault on a vdev and want it to persist across imports (like with * zpool offline -f). */ rc = nvlist_lookup_string(nv, ZPOOL_CONFIG_AUX_STATE, &tmp); if (rc == 0 && tmp != NULL && strcmp(tmp, "external") == 0) { vd->vdev_stat.vs_aux = VDEV_AUX_EXTERNAL; vd->vdev_faulted = 1; vd->vdev_label_aux = VDEV_AUX_EXTERNAL; } if (nvlist_lookup_string(nv, ZPOOL_CONFIG_DEVID, &vd->vdev_devid) == 0) vd->vdev_devid = spa_strdup(vd->vdev_devid); if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PHYS_PATH, &vd->vdev_physpath) == 0) vd->vdev_physpath = spa_strdup(vd->vdev_physpath); if (nvlist_lookup_string(nv, ZPOOL_CONFIG_VDEV_ENC_SYSFS_PATH, &vd->vdev_enc_sysfs_path) == 0) vd->vdev_enc_sysfs_path = spa_strdup(vd->vdev_enc_sysfs_path); if (nvlist_lookup_string(nv, ZPOOL_CONFIG_FRU, &vd->vdev_fru) == 0) vd->vdev_fru = spa_strdup(vd->vdev_fru); /* * Set the whole_disk property. If it's not specified, leave the value * as -1. */ if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK, &vd->vdev_wholedisk) != 0) vd->vdev_wholedisk = -1ULL; ASSERT0(vic->vic_mapping_object); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_INDIRECT_OBJECT, &vic->vic_mapping_object); ASSERT0(vic->vic_births_object); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_INDIRECT_BIRTHS, &vic->vic_births_object); ASSERT3U(vic->vic_prev_indirect_vdev, ==, UINT64_MAX); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_PREV_INDIRECT_VDEV, &vic->vic_prev_indirect_vdev); /* * Look for the 'not present' flag. This will only be set if the device * was not present at the time of import. */ (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NOT_PRESENT, &vd->vdev_not_present); /* * Get the alignment requirement. */ (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ASHIFT, &vd->vdev_ashift); /* * Retrieve the vdev creation time. */ (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_CREATE_TXG, &vd->vdev_crtxg); /* * If we're a top-level vdev, try to load the allocation parameters. */ if (top_level && (alloctype == VDEV_ALLOC_LOAD || alloctype == VDEV_ALLOC_SPLIT)) { (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_METASLAB_ARRAY, &vd->vdev_ms_array); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_METASLAB_SHIFT, &vd->vdev_ms_shift); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ASIZE, &vd->vdev_asize); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_REMOVING, &vd->vdev_removing); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_VDEV_TOP_ZAP, &vd->vdev_top_zap); } else { ASSERT0(vd->vdev_top_zap); } if (top_level && alloctype != VDEV_ALLOC_ATTACH) { ASSERT(alloctype == VDEV_ALLOC_LOAD || alloctype == VDEV_ALLOC_ADD || alloctype == VDEV_ALLOC_SPLIT || alloctype == VDEV_ALLOC_ROOTPOOL); /* Note: metaslab_group_create() is now deferred */ } if (vd->vdev_ops->vdev_op_leaf && (alloctype == VDEV_ALLOC_LOAD || alloctype == VDEV_ALLOC_SPLIT)) { (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_VDEV_LEAF_ZAP, &vd->vdev_leaf_zap); } else { ASSERT0(vd->vdev_leaf_zap); } /* * If we're a leaf vdev, try to load the DTL object and other state. */ if (vd->vdev_ops->vdev_op_leaf && (alloctype == VDEV_ALLOC_LOAD || alloctype == VDEV_ALLOC_L2CACHE || alloctype == VDEV_ALLOC_ROOTPOOL)) { if (alloctype == VDEV_ALLOC_LOAD) { (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_DTL, &vd->vdev_dtl_object); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_UNSPARE, &vd->vdev_unspare); } if (alloctype == VDEV_ALLOC_ROOTPOOL) { uint64_t spare = 0; if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_SPARE, &spare) == 0 && spare) spa_spare_add(vd); } (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_OFFLINE, &vd->vdev_offline); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_RESILVER_TXG, &vd->vdev_resilver_txg); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_REBUILD_TXG, &vd->vdev_rebuild_txg); if (nvlist_exists(nv, ZPOOL_CONFIG_RESILVER_DEFER)) vdev_defer_resilver(vd); /* * In general, when importing a pool we want to ignore the * persistent fault state, as the diagnosis made on another * system may not be valid in the current context. The only * exception is if we forced a vdev to a persistently faulted * state with 'zpool offline -f'. The persistent fault will * remain across imports until cleared. * * Local vdevs will remain in the faulted state. */ if (spa_load_state(spa) == SPA_LOAD_OPEN || spa_load_state(spa) == SPA_LOAD_IMPORT) { (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_FAULTED, &vd->vdev_faulted); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_DEGRADED, &vd->vdev_degraded); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_REMOVED, &vd->vdev_removed); if (vd->vdev_faulted || vd->vdev_degraded) { char *aux; vd->vdev_label_aux = VDEV_AUX_ERR_EXCEEDED; if (nvlist_lookup_string(nv, ZPOOL_CONFIG_AUX_STATE, &aux) == 0 && strcmp(aux, "external") == 0) vd->vdev_label_aux = VDEV_AUX_EXTERNAL; else vd->vdev_faulted = 0ULL; } } } /* * Add ourselves to the parent's list of children. */ vdev_add_child(parent, vd); *vdp = vd; return (0); } void vdev_free(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT3P(vd->vdev_initialize_thread, ==, NULL); ASSERT3P(vd->vdev_trim_thread, ==, NULL); ASSERT3P(vd->vdev_autotrim_thread, ==, NULL); ASSERT3P(vd->vdev_rebuild_thread, ==, NULL); /* * Scan queues are normally destroyed at the end of a scan. If the * queue exists here, that implies the vdev is being removed while * the scan is still running. */ if (vd->vdev_scan_io_queue != NULL) { mutex_enter(&vd->vdev_scan_io_queue_lock); dsl_scan_io_queue_destroy(vd->vdev_scan_io_queue); vd->vdev_scan_io_queue = NULL; mutex_exit(&vd->vdev_scan_io_queue_lock); } /* * vdev_free() implies closing the vdev first. This is simpler than * trying to ensure complicated semantics for all callers. */ vdev_close(vd); ASSERT(!list_link_active(&vd->vdev_config_dirty_node)); ASSERT(!list_link_active(&vd->vdev_state_dirty_node)); /* * Free all children. */ for (int c = 0; c < vd->vdev_children; c++) vdev_free(vd->vdev_child[c]); ASSERT(vd->vdev_child == NULL); ASSERT(vd->vdev_guid_sum == vd->vdev_guid); /* * Discard allocation state. */ if (vd->vdev_mg != NULL) { vdev_metaslab_fini(vd); metaslab_group_destroy(vd->vdev_mg); vd->vdev_mg = NULL; } ASSERT0(vd->vdev_stat.vs_space); ASSERT0(vd->vdev_stat.vs_dspace); ASSERT0(vd->vdev_stat.vs_alloc); /* * Remove this vdev from its parent's child list. */ vdev_remove_child(vd->vdev_parent, vd); ASSERT(vd->vdev_parent == NULL); ASSERT(!list_link_active(&vd->vdev_leaf_node)); /* * Clean up vdev structure. */ vdev_queue_fini(vd); vdev_cache_fini(vd); if (vd->vdev_path) spa_strfree(vd->vdev_path); if (vd->vdev_devid) spa_strfree(vd->vdev_devid); if (vd->vdev_physpath) spa_strfree(vd->vdev_physpath); if (vd->vdev_enc_sysfs_path) spa_strfree(vd->vdev_enc_sysfs_path); if (vd->vdev_fru) spa_strfree(vd->vdev_fru); if (vd->vdev_isspare) spa_spare_remove(vd); if (vd->vdev_isl2cache) spa_l2cache_remove(vd); txg_list_destroy(&vd->vdev_ms_list); txg_list_destroy(&vd->vdev_dtl_list); mutex_enter(&vd->vdev_dtl_lock); space_map_close(vd->vdev_dtl_sm); for (int t = 0; t < DTL_TYPES; t++) { range_tree_vacate(vd->vdev_dtl[t], NULL, NULL); range_tree_destroy(vd->vdev_dtl[t]); } mutex_exit(&vd->vdev_dtl_lock); EQUIV(vd->vdev_indirect_births != NULL, vd->vdev_indirect_mapping != NULL); if (vd->vdev_indirect_births != NULL) { vdev_indirect_mapping_close(vd->vdev_indirect_mapping); vdev_indirect_births_close(vd->vdev_indirect_births); } if (vd->vdev_obsolete_sm != NULL) { ASSERT(vd->vdev_removing || vd->vdev_ops == &vdev_indirect_ops); space_map_close(vd->vdev_obsolete_sm); vd->vdev_obsolete_sm = NULL; } range_tree_destroy(vd->vdev_obsolete_segments); rw_destroy(&vd->vdev_indirect_rwlock); mutex_destroy(&vd->vdev_obsolete_lock); mutex_destroy(&vd->vdev_dtl_lock); mutex_destroy(&vd->vdev_stat_lock); mutex_destroy(&vd->vdev_probe_lock); mutex_destroy(&vd->vdev_scan_io_queue_lock); mutex_destroy(&vd->vdev_initialize_lock); mutex_destroy(&vd->vdev_initialize_io_lock); cv_destroy(&vd->vdev_initialize_io_cv); cv_destroy(&vd->vdev_initialize_cv); mutex_destroy(&vd->vdev_trim_lock); mutex_destroy(&vd->vdev_autotrim_lock); mutex_destroy(&vd->vdev_trim_io_lock); cv_destroy(&vd->vdev_trim_cv); cv_destroy(&vd->vdev_autotrim_cv); cv_destroy(&vd->vdev_trim_io_cv); mutex_destroy(&vd->vdev_rebuild_lock); mutex_destroy(&vd->vdev_rebuild_io_lock); cv_destroy(&vd->vdev_rebuild_cv); cv_destroy(&vd->vdev_rebuild_io_cv); zfs_ratelimit_fini(&vd->vdev_delay_rl); zfs_ratelimit_fini(&vd->vdev_checksum_rl); if (vd == spa->spa_root_vdev) spa->spa_root_vdev = NULL; kmem_free(vd, sizeof (vdev_t)); } /* * Transfer top-level vdev state from svd to tvd. */ static void vdev_top_transfer(vdev_t *svd, vdev_t *tvd) { spa_t *spa = svd->vdev_spa; metaslab_t *msp; vdev_t *vd; int t; ASSERT(tvd == tvd->vdev_top); tvd->vdev_pending_fastwrite = svd->vdev_pending_fastwrite; tvd->vdev_ms_array = svd->vdev_ms_array; tvd->vdev_ms_shift = svd->vdev_ms_shift; tvd->vdev_ms_count = svd->vdev_ms_count; tvd->vdev_top_zap = svd->vdev_top_zap; svd->vdev_ms_array = 0; svd->vdev_ms_shift = 0; svd->vdev_ms_count = 0; svd->vdev_top_zap = 0; if (tvd->vdev_mg) ASSERT3P(tvd->vdev_mg, ==, svd->vdev_mg); tvd->vdev_mg = svd->vdev_mg; tvd->vdev_ms = svd->vdev_ms; svd->vdev_mg = NULL; svd->vdev_ms = NULL; if (tvd->vdev_mg != NULL) tvd->vdev_mg->mg_vd = tvd; tvd->vdev_checkpoint_sm = svd->vdev_checkpoint_sm; svd->vdev_checkpoint_sm = NULL; tvd->vdev_alloc_bias = svd->vdev_alloc_bias; svd->vdev_alloc_bias = VDEV_BIAS_NONE; tvd->vdev_stat.vs_alloc = svd->vdev_stat.vs_alloc; tvd->vdev_stat.vs_space = svd->vdev_stat.vs_space; tvd->vdev_stat.vs_dspace = svd->vdev_stat.vs_dspace; svd->vdev_stat.vs_alloc = 0; svd->vdev_stat.vs_space = 0; svd->vdev_stat.vs_dspace = 0; /* * State which may be set on a top-level vdev that's in the * process of being removed. */ ASSERT0(tvd->vdev_indirect_config.vic_births_object); ASSERT0(tvd->vdev_indirect_config.vic_mapping_object); ASSERT3U(tvd->vdev_indirect_config.vic_prev_indirect_vdev, ==, -1ULL); ASSERT3P(tvd->vdev_indirect_mapping, ==, NULL); ASSERT3P(tvd->vdev_indirect_births, ==, NULL); ASSERT3P(tvd->vdev_obsolete_sm, ==, NULL); ASSERT0(tvd->vdev_removing); ASSERT0(tvd->vdev_rebuilding); tvd->vdev_removing = svd->vdev_removing; tvd->vdev_rebuilding = svd->vdev_rebuilding; tvd->vdev_rebuild_config = svd->vdev_rebuild_config; tvd->vdev_indirect_config = svd->vdev_indirect_config; tvd->vdev_indirect_mapping = svd->vdev_indirect_mapping; tvd->vdev_indirect_births = svd->vdev_indirect_births; range_tree_swap(&svd->vdev_obsolete_segments, &tvd->vdev_obsolete_segments); tvd->vdev_obsolete_sm = svd->vdev_obsolete_sm; svd->vdev_indirect_config.vic_mapping_object = 0; svd->vdev_indirect_config.vic_births_object = 0; svd->vdev_indirect_config.vic_prev_indirect_vdev = -1ULL; svd->vdev_indirect_mapping = NULL; svd->vdev_indirect_births = NULL; svd->vdev_obsolete_sm = NULL; svd->vdev_removing = 0; svd->vdev_rebuilding = 0; for (t = 0; t < TXG_SIZE; t++) { while ((msp = txg_list_remove(&svd->vdev_ms_list, t)) != NULL) (void) txg_list_add(&tvd->vdev_ms_list, msp, t); while ((vd = txg_list_remove(&svd->vdev_dtl_list, t)) != NULL) (void) txg_list_add(&tvd->vdev_dtl_list, vd, t); if (txg_list_remove_this(&spa->spa_vdev_txg_list, svd, t)) (void) txg_list_add(&spa->spa_vdev_txg_list, tvd, t); } if (list_link_active(&svd->vdev_config_dirty_node)) { vdev_config_clean(svd); vdev_config_dirty(tvd); } if (list_link_active(&svd->vdev_state_dirty_node)) { vdev_state_clean(svd); vdev_state_dirty(tvd); } tvd->vdev_deflate_ratio = svd->vdev_deflate_ratio; svd->vdev_deflate_ratio = 0; tvd->vdev_islog = svd->vdev_islog; svd->vdev_islog = 0; dsl_scan_io_queue_vdev_xfer(svd, tvd); } static void vdev_top_update(vdev_t *tvd, vdev_t *vd) { if (vd == NULL) return; vd->vdev_top = tvd; for (int c = 0; c < vd->vdev_children; c++) vdev_top_update(tvd, vd->vdev_child[c]); } /* * Add a mirror/replacing vdev above an existing vdev. */ vdev_t * vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops) { spa_t *spa = cvd->vdev_spa; vdev_t *pvd = cvd->vdev_parent; vdev_t *mvd; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); mvd = vdev_alloc_common(spa, cvd->vdev_id, 0, ops); mvd->vdev_asize = cvd->vdev_asize; mvd->vdev_min_asize = cvd->vdev_min_asize; mvd->vdev_max_asize = cvd->vdev_max_asize; mvd->vdev_psize = cvd->vdev_psize; mvd->vdev_ashift = cvd->vdev_ashift; mvd->vdev_logical_ashift = cvd->vdev_logical_ashift; mvd->vdev_physical_ashift = cvd->vdev_physical_ashift; mvd->vdev_state = cvd->vdev_state; mvd->vdev_crtxg = cvd->vdev_crtxg; vdev_remove_child(pvd, cvd); vdev_add_child(pvd, mvd); cvd->vdev_id = mvd->vdev_children; vdev_add_child(mvd, cvd); vdev_top_update(cvd->vdev_top, cvd->vdev_top); if (mvd == mvd->vdev_top) vdev_top_transfer(cvd, mvd); return (mvd); } /* * Remove a 1-way mirror/replacing vdev from the tree. */ void vdev_remove_parent(vdev_t *cvd) { vdev_t *mvd = cvd->vdev_parent; vdev_t *pvd = mvd->vdev_parent; ASSERT(spa_config_held(cvd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL); ASSERT(mvd->vdev_children == 1); ASSERT(mvd->vdev_ops == &vdev_mirror_ops || mvd->vdev_ops == &vdev_replacing_ops || mvd->vdev_ops == &vdev_spare_ops); cvd->vdev_ashift = mvd->vdev_ashift; cvd->vdev_logical_ashift = mvd->vdev_logical_ashift; cvd->vdev_physical_ashift = mvd->vdev_physical_ashift; vdev_remove_child(mvd, cvd); vdev_remove_child(pvd, mvd); /* * If cvd will replace mvd as a top-level vdev, preserve mvd's guid. * Otherwise, we could have detached an offline device, and when we * go to import the pool we'll think we have two top-level vdevs, * instead of a different version of the same top-level vdev. */ if (mvd->vdev_top == mvd) { uint64_t guid_delta = mvd->vdev_guid - cvd->vdev_guid; cvd->vdev_orig_guid = cvd->vdev_guid; cvd->vdev_guid += guid_delta; cvd->vdev_guid_sum += guid_delta; /* * If pool not set for autoexpand, we need to also preserve * mvd's asize to prevent automatic expansion of cvd. * Otherwise if we are adjusting the mirror by attaching and * detaching children of non-uniform sizes, the mirror could * autoexpand, unexpectedly requiring larger devices to * re-establish the mirror. */ if (!cvd->vdev_spa->spa_autoexpand) cvd->vdev_asize = mvd->vdev_asize; } cvd->vdev_id = mvd->vdev_id; vdev_add_child(pvd, cvd); vdev_top_update(cvd->vdev_top, cvd->vdev_top); if (cvd == cvd->vdev_top) vdev_top_transfer(mvd, cvd); ASSERT(mvd->vdev_children == 0); vdev_free(mvd); } static void vdev_metaslab_group_create(vdev_t *vd) { spa_t *spa = vd->vdev_spa; /* * metaslab_group_create was delayed until allocation bias was available */ if (vd->vdev_mg == NULL) { metaslab_class_t *mc; if (vd->vdev_islog && vd->vdev_alloc_bias == VDEV_BIAS_NONE) vd->vdev_alloc_bias = VDEV_BIAS_LOG; ASSERT3U(vd->vdev_islog, ==, (vd->vdev_alloc_bias == VDEV_BIAS_LOG)); switch (vd->vdev_alloc_bias) { case VDEV_BIAS_LOG: mc = spa_log_class(spa); break; case VDEV_BIAS_SPECIAL: mc = spa_special_class(spa); break; case VDEV_BIAS_DEDUP: mc = spa_dedup_class(spa); break; default: mc = spa_normal_class(spa); } vd->vdev_mg = metaslab_group_create(mc, vd, spa->spa_alloc_count); /* * The spa ashift min/max only apply for the normal metaslab * class. Class destination is late binding so ashift boundry * setting had to wait until now. */ if (vd->vdev_top == vd && vd->vdev_ashift != 0 && mc == spa_normal_class(spa) && vd->vdev_aux == NULL) { if (vd->vdev_ashift > spa->spa_max_ashift) spa->spa_max_ashift = vd->vdev_ashift; if (vd->vdev_ashift < spa->spa_min_ashift) spa->spa_min_ashift = vd->vdev_ashift; } } } int vdev_metaslab_init(vdev_t *vd, uint64_t txg) { spa_t *spa = vd->vdev_spa; objset_t *mos = spa->spa_meta_objset; uint64_t m; uint64_t oldc = vd->vdev_ms_count; uint64_t newc = vd->vdev_asize >> vd->vdev_ms_shift; metaslab_t **mspp; int error; boolean_t expanding = (oldc != 0); ASSERT(txg == 0 || spa_config_held(spa, SCL_ALLOC, RW_WRITER)); /* * This vdev is not being allocated from yet or is a hole. */ if (vd->vdev_ms_shift == 0) return (0); ASSERT(!vd->vdev_ishole); ASSERT(oldc <= newc); mspp = vmem_zalloc(newc * sizeof (*mspp), KM_SLEEP); if (expanding) { bcopy(vd->vdev_ms, mspp, oldc * sizeof (*mspp)); vmem_free(vd->vdev_ms, oldc * sizeof (*mspp)); } vd->vdev_ms = mspp; vd->vdev_ms_count = newc; for (m = oldc; m < newc; m++) { uint64_t object = 0; /* * vdev_ms_array may be 0 if we are creating the "fake" * metaslabs for an indirect vdev for zdb's leak detection. * See zdb_leak_init(). */ if (txg == 0 && vd->vdev_ms_array != 0) { error = dmu_read(mos, vd->vdev_ms_array, m * sizeof (uint64_t), sizeof (uint64_t), &object, DMU_READ_PREFETCH); if (error != 0) { vdev_dbgmsg(vd, "unable to read the metaslab " "array [error=%d]", error); return (error); } } #ifndef _KERNEL /* * To accommodate zdb_leak_init() fake indirect * metaslabs, we allocate a metaslab group for * indirect vdevs which normally don't have one. */ if (vd->vdev_mg == NULL) { ASSERT0(vdev_is_concrete(vd)); vdev_metaslab_group_create(vd); } #endif error = metaslab_init(vd->vdev_mg, m, object, txg, &(vd->vdev_ms[m])); if (error != 0) { vdev_dbgmsg(vd, "metaslab_init failed [error=%d]", error); return (error); } } if (txg == 0) spa_config_enter(spa, SCL_ALLOC, FTAG, RW_WRITER); /* * If the vdev is being removed we don't activate * the metaslabs since we want to ensure that no new * allocations are performed on this device. */ if (!expanding && !vd->vdev_removing) { metaslab_group_activate(vd->vdev_mg); } if (txg == 0) spa_config_exit(spa, SCL_ALLOC, FTAG); /* * Regardless whether this vdev was just added or it is being * expanded, the metaslab count has changed. Recalculate the * block limit. */ spa_log_sm_set_blocklimit(spa); return (0); } void vdev_metaslab_fini(vdev_t *vd) { if (vd->vdev_checkpoint_sm != NULL) { ASSERT(spa_feature_is_active(vd->vdev_spa, SPA_FEATURE_POOL_CHECKPOINT)); space_map_close(vd->vdev_checkpoint_sm); /* * Even though we close the space map, we need to set its * pointer to NULL. The reason is that vdev_metaslab_fini() * may be called multiple times for certain operations * (i.e. when destroying a pool) so we need to ensure that * this clause never executes twice. This logic is similar * to the one used for the vdev_ms clause below. */ vd->vdev_checkpoint_sm = NULL; } if (vd->vdev_ms != NULL) { metaslab_group_t *mg = vd->vdev_mg; metaslab_group_passivate(mg); uint64_t count = vd->vdev_ms_count; for (uint64_t m = 0; m < count; m++) { metaslab_t *msp = vd->vdev_ms[m]; if (msp != NULL) metaslab_fini(msp); } vmem_free(vd->vdev_ms, count * sizeof (metaslab_t *)); vd->vdev_ms = NULL; vd->vdev_ms_count = 0; for (int i = 0; i < RANGE_TREE_HISTOGRAM_SIZE; i++) ASSERT0(mg->mg_histogram[i]); } ASSERT0(vd->vdev_ms_count); ASSERT3U(vd->vdev_pending_fastwrite, ==, 0); } typedef struct vdev_probe_stats { boolean_t vps_readable; boolean_t vps_writeable; int vps_flags; } vdev_probe_stats_t; static void vdev_probe_done(zio_t *zio) { spa_t *spa = zio->io_spa; vdev_t *vd = zio->io_vd; vdev_probe_stats_t *vps = zio->io_private; ASSERT(vd->vdev_probe_zio != NULL); if (zio->io_type == ZIO_TYPE_READ) { if (zio->io_error == 0) vps->vps_readable = 1; if (zio->io_error == 0 && spa_writeable(spa)) { zio_nowait(zio_write_phys(vd->vdev_probe_zio, vd, zio->io_offset, zio->io_size, zio->io_abd, ZIO_CHECKSUM_OFF, vdev_probe_done, vps, ZIO_PRIORITY_SYNC_WRITE, vps->vps_flags, B_TRUE)); } else { abd_free(zio->io_abd); } } else if (zio->io_type == ZIO_TYPE_WRITE) { if (zio->io_error == 0) vps->vps_writeable = 1; abd_free(zio->io_abd); } else if (zio->io_type == ZIO_TYPE_NULL) { zio_t *pio; zio_link_t *zl; vd->vdev_cant_read |= !vps->vps_readable; vd->vdev_cant_write |= !vps->vps_writeable; if (vdev_readable(vd) && (vdev_writeable(vd) || !spa_writeable(spa))) { zio->io_error = 0; } else { ASSERT(zio->io_error != 0); vdev_dbgmsg(vd, "failed probe"); (void) zfs_ereport_post(FM_EREPORT_ZFS_PROBE_FAILURE, spa, vd, NULL, NULL, 0); zio->io_error = SET_ERROR(ENXIO); } mutex_enter(&vd->vdev_probe_lock); ASSERT(vd->vdev_probe_zio == zio); vd->vdev_probe_zio = NULL; mutex_exit(&vd->vdev_probe_lock); zl = NULL; while ((pio = zio_walk_parents(zio, &zl)) != NULL) if (!vdev_accessible(vd, pio)) pio->io_error = SET_ERROR(ENXIO); kmem_free(vps, sizeof (*vps)); } } /* * Determine whether this device is accessible. * * Read and write to several known locations: the pad regions of each * vdev label but the first, which we leave alone in case it contains * a VTOC. */ zio_t * vdev_probe(vdev_t *vd, zio_t *zio) { spa_t *spa = vd->vdev_spa; vdev_probe_stats_t *vps = NULL; zio_t *pio; ASSERT(vd->vdev_ops->vdev_op_leaf); /* * Don't probe the probe. */ if (zio && (zio->io_flags & ZIO_FLAG_PROBE)) return (NULL); /* * To prevent 'probe storms' when a device fails, we create * just one probe i/o at a time. All zios that want to probe * this vdev will become parents of the probe io. */ mutex_enter(&vd->vdev_probe_lock); if ((pio = vd->vdev_probe_zio) == NULL) { vps = kmem_zalloc(sizeof (*vps), KM_SLEEP); vps->vps_flags = ZIO_FLAG_CANFAIL | ZIO_FLAG_PROBE | ZIO_FLAG_DONT_CACHE | ZIO_FLAG_DONT_AGGREGATE | ZIO_FLAG_TRYHARD; if (spa_config_held(spa, SCL_ZIO, RW_WRITER)) { /* * vdev_cant_read and vdev_cant_write can only * transition from TRUE to FALSE when we have the * SCL_ZIO lock as writer; otherwise they can only * transition from FALSE to TRUE. This ensures that * any zio looking at these values can assume that * failures persist for the life of the I/O. That's * important because when a device has intermittent * connectivity problems, we want to ensure that * they're ascribed to the device (ENXIO) and not * the zio (EIO). * * Since we hold SCL_ZIO as writer here, clear both * values so the probe can reevaluate from first * principles. */ vps->vps_flags |= ZIO_FLAG_CONFIG_WRITER; vd->vdev_cant_read = B_FALSE; vd->vdev_cant_write = B_FALSE; } vd->vdev_probe_zio = pio = zio_null(NULL, spa, vd, vdev_probe_done, vps, vps->vps_flags | ZIO_FLAG_DONT_PROPAGATE); /* * We can't change the vdev state in this context, so we * kick off an async task to do it on our behalf. */ if (zio != NULL) { vd->vdev_probe_wanted = B_TRUE; spa_async_request(spa, SPA_ASYNC_PROBE); } } if (zio != NULL) zio_add_child(zio, pio); mutex_exit(&vd->vdev_probe_lock); if (vps == NULL) { ASSERT(zio != NULL); return (NULL); } for (int l = 1; l < VDEV_LABELS; l++) { zio_nowait(zio_read_phys(pio, vd, vdev_label_offset(vd->vdev_psize, l, offsetof(vdev_label_t, vl_be)), VDEV_PAD_SIZE, abd_alloc_for_io(VDEV_PAD_SIZE, B_TRUE), ZIO_CHECKSUM_OFF, vdev_probe_done, vps, ZIO_PRIORITY_SYNC_READ, vps->vps_flags, B_TRUE)); } if (zio == NULL) return (pio); zio_nowait(pio); return (NULL); } static void vdev_open_child(void *arg) { vdev_t *vd = arg; vd->vdev_open_thread = curthread; vd->vdev_open_error = vdev_open(vd); vd->vdev_open_thread = NULL; } static boolean_t vdev_uses_zvols(vdev_t *vd) { #ifdef _KERNEL if (zvol_is_zvol(vd->vdev_path)) return (B_TRUE); #endif for (int c = 0; c < vd->vdev_children; c++) if (vdev_uses_zvols(vd->vdev_child[c])) return (B_TRUE); return (B_FALSE); } void vdev_open_children(vdev_t *vd) { taskq_t *tq; int children = vd->vdev_children; /* * in order to handle pools on top of zvols, do the opens * in a single thread so that the same thread holds the * spa_namespace_lock */ if (vdev_uses_zvols(vd)) { retry_sync: for (int c = 0; c < children; c++) vd->vdev_child[c]->vdev_open_error = vdev_open(vd->vdev_child[c]); } else { tq = taskq_create("vdev_open", children, minclsyspri, children, children, TASKQ_PREPOPULATE); if (tq == NULL) goto retry_sync; for (int c = 0; c < children; c++) VERIFY(taskq_dispatch(tq, vdev_open_child, vd->vdev_child[c], TQ_SLEEP) != TASKQID_INVALID); taskq_destroy(tq); } vd->vdev_nonrot = B_TRUE; for (int c = 0; c < children; c++) vd->vdev_nonrot &= vd->vdev_child[c]->vdev_nonrot; } /* * Compute the raidz-deflation ratio. Note, we hard-code * in 128k (1 << 17) because it is the "typical" blocksize. * Even though SPA_MAXBLOCKSIZE changed, this algorithm can not change, * otherwise it would inconsistently account for existing bp's. */ static void vdev_set_deflate_ratio(vdev_t *vd) { if (vd == vd->vdev_top && !vd->vdev_ishole && vd->vdev_ashift != 0) { vd->vdev_deflate_ratio = (1 << 17) / (vdev_psize_to_asize(vd, 1 << 17) >> SPA_MINBLOCKSHIFT); } } /* * Maximize performance by inflating the configured ashift for top level * vdevs to be as close to the physical ashift as possible while maintaining * administrator defined limits and ensuring it doesn't go below the * logical ashift. */ static void vdev_ashift_optimize(vdev_t *vd) { ASSERT(vd == vd->vdev_top); if (vd->vdev_ashift < vd->vdev_physical_ashift) { vd->vdev_ashift = MIN( MAX(zfs_vdev_max_auto_ashift, vd->vdev_ashift), MAX(zfs_vdev_min_auto_ashift, vd->vdev_physical_ashift)); } else { /* * If the logical and physical ashifts are the same, then * we ensure that the top-level vdev's ashift is not smaller * than our minimum ashift value. For the unusual case * where logical ashift > physical ashift, we can't cap * the calculated ashift based on max ashift as that * would cause failures. * We still check if we need to increase it to match * the min ashift. */ vd->vdev_ashift = MAX(zfs_vdev_min_auto_ashift, vd->vdev_ashift); } } /* * Prepare a virtual device for access. */ int vdev_open(vdev_t *vd) { spa_t *spa = vd->vdev_spa; int error; uint64_t osize = 0; uint64_t max_osize = 0; uint64_t asize, max_asize, psize; uint64_t logical_ashift = 0; uint64_t physical_ashift = 0; ASSERT(vd->vdev_open_thread == curthread || spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); ASSERT(vd->vdev_state == VDEV_STATE_CLOSED || vd->vdev_state == VDEV_STATE_CANT_OPEN || vd->vdev_state == VDEV_STATE_OFFLINE); vd->vdev_stat.vs_aux = VDEV_AUX_NONE; vd->vdev_cant_read = B_FALSE; vd->vdev_cant_write = B_FALSE; vd->vdev_min_asize = vdev_get_min_asize(vd); /* * If this vdev is not removed, check its fault status. If it's * faulted, bail out of the open. */ if (!vd->vdev_removed && vd->vdev_faulted) { ASSERT(vd->vdev_children == 0); ASSERT(vd->vdev_label_aux == VDEV_AUX_ERR_EXCEEDED || vd->vdev_label_aux == VDEV_AUX_EXTERNAL); vdev_set_state(vd, B_TRUE, VDEV_STATE_FAULTED, vd->vdev_label_aux); return (SET_ERROR(ENXIO)); } else if (vd->vdev_offline) { ASSERT(vd->vdev_children == 0); vdev_set_state(vd, B_TRUE, VDEV_STATE_OFFLINE, VDEV_AUX_NONE); return (SET_ERROR(ENXIO)); } error = vd->vdev_ops->vdev_op_open(vd, &osize, &max_osize, &logical_ashift, &physical_ashift); /* * Physical volume size should never be larger than its max size, unless * the disk has shrunk while we were reading it or the device is buggy * or damaged: either way it's not safe for use, bail out of the open. */ if (osize > max_osize) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_OPEN_FAILED); return (SET_ERROR(ENXIO)); } /* * Reset the vdev_reopening flag so that we actually close * the vdev on error. */ vd->vdev_reopening = B_FALSE; if (zio_injection_enabled && error == 0) error = zio_handle_device_injection(vd, NULL, SET_ERROR(ENXIO)); if (error) { if (vd->vdev_removed && vd->vdev_stat.vs_aux != VDEV_AUX_OPEN_FAILED) vd->vdev_removed = B_FALSE; if (vd->vdev_stat.vs_aux == VDEV_AUX_CHILDREN_OFFLINE) { vdev_set_state(vd, B_TRUE, VDEV_STATE_OFFLINE, vd->vdev_stat.vs_aux); } else { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, vd->vdev_stat.vs_aux); } return (error); } vd->vdev_removed = B_FALSE; /* * Recheck the faulted flag now that we have confirmed that * the vdev is accessible. If we're faulted, bail. */ if (vd->vdev_faulted) { ASSERT(vd->vdev_children == 0); ASSERT(vd->vdev_label_aux == VDEV_AUX_ERR_EXCEEDED || vd->vdev_label_aux == VDEV_AUX_EXTERNAL); vdev_set_state(vd, B_TRUE, VDEV_STATE_FAULTED, vd->vdev_label_aux); return (SET_ERROR(ENXIO)); } if (vd->vdev_degraded) { ASSERT(vd->vdev_children == 0); vdev_set_state(vd, B_TRUE, VDEV_STATE_DEGRADED, VDEV_AUX_ERR_EXCEEDED); } else { vdev_set_state(vd, B_TRUE, VDEV_STATE_HEALTHY, 0); } /* * For hole or missing vdevs we just return success. */ if (vd->vdev_ishole || vd->vdev_ops == &vdev_missing_ops) return (0); for (int c = 0; c < vd->vdev_children; c++) { if (vd->vdev_child[c]->vdev_state != VDEV_STATE_HEALTHY) { vdev_set_state(vd, B_TRUE, VDEV_STATE_DEGRADED, VDEV_AUX_NONE); break; } } osize = P2ALIGN(osize, (uint64_t)sizeof (vdev_label_t)); max_osize = P2ALIGN(max_osize, (uint64_t)sizeof (vdev_label_t)); if (vd->vdev_children == 0) { if (osize < SPA_MINDEVSIZE) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_TOO_SMALL); return (SET_ERROR(EOVERFLOW)); } psize = osize; asize = osize - (VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE); max_asize = max_osize - (VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE); } else { if (vd->vdev_parent != NULL && osize < SPA_MINDEVSIZE - (VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE)) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_TOO_SMALL); return (SET_ERROR(EOVERFLOW)); } psize = 0; asize = osize; max_asize = max_osize; } /* * If the vdev was expanded, record this so that we can re-create the * uberblock rings in labels {2,3}, during the next sync. */ if ((psize > vd->vdev_psize) && (vd->vdev_psize != 0)) vd->vdev_copy_uberblocks = B_TRUE; vd->vdev_psize = psize; /* * Make sure the allocatable size hasn't shrunk too much. */ if (asize < vd->vdev_min_asize) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_BAD_LABEL); return (SET_ERROR(EINVAL)); } /* * We can always set the logical/physical ashift members since * their values are only used to calculate the vdev_ashift when * the device is first added to the config. These values should * not be used for anything else since they may change whenever * the device is reopened and we don't store them in the label. */ vd->vdev_physical_ashift = MAX(physical_ashift, vd->vdev_physical_ashift); vd->vdev_logical_ashift = MAX(logical_ashift, vd->vdev_logical_ashift); if (vd->vdev_asize == 0) { /* * This is the first-ever open, so use the computed values. * For compatibility, a different ashift can be requested. */ vd->vdev_asize = asize; vd->vdev_max_asize = max_asize; /* * If the vdev_ashift was not overriden at creation time, * then set it the logical ashift and optimize the ashift. */ if (vd->vdev_ashift == 0) { vd->vdev_ashift = vd->vdev_logical_ashift; if (vd->vdev_logical_ashift > ASHIFT_MAX) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_ASHIFT_TOO_BIG); return (SET_ERROR(EDOM)); } if (vd->vdev_top == vd) { vdev_ashift_optimize(vd); } } if (vd->vdev_ashift != 0 && (vd->vdev_ashift < ASHIFT_MIN || vd->vdev_ashift > ASHIFT_MAX)) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_BAD_ASHIFT); return (SET_ERROR(EDOM)); } } else { /* * Make sure the alignment required hasn't increased. */ if (vd->vdev_ashift > vd->vdev_top->vdev_ashift && vd->vdev_ops->vdev_op_leaf) { (void) zfs_ereport_post( FM_EREPORT_ZFS_DEVICE_BAD_ASHIFT, spa, vd, NULL, NULL, 0); vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_BAD_LABEL); return (SET_ERROR(EDOM)); } vd->vdev_max_asize = max_asize; } /* * If all children are healthy we update asize if either: * The asize has increased, due to a device expansion caused by dynamic * LUN growth or vdev replacement, and automatic expansion is enabled; * making the additional space available. * * The asize has decreased, due to a device shrink usually caused by a * vdev replace with a smaller device. This ensures that calculations * based of max_asize and asize e.g. esize are always valid. It's safe * to do this as we've already validated that asize is greater than * vdev_min_asize. */ if (vd->vdev_state == VDEV_STATE_HEALTHY && ((asize > vd->vdev_asize && (vd->vdev_expanding || spa->spa_autoexpand)) || (asize < vd->vdev_asize))) vd->vdev_asize = asize; vdev_set_min_asize(vd); /* * Ensure we can issue some IO before declaring the * vdev open for business. */ if (vd->vdev_ops->vdev_op_leaf && (error = zio_wait(vdev_probe(vd, NULL))) != 0) { vdev_set_state(vd, B_TRUE, VDEV_STATE_FAULTED, VDEV_AUX_ERR_EXCEEDED); return (error); } /* * If this is a leaf vdev, assess whether a resilver is needed. * But don't do this if we are doing a reopen for a scrub, since * this would just restart the scrub we are already doing. */ if (vd->vdev_ops->vdev_op_leaf && !spa->spa_scrub_reopen) dsl_scan_assess_vdev(spa->spa_dsl_pool, vd); return (0); } /* * Called once the vdevs are all opened, this routine validates the label * contents. This needs to be done before vdev_load() so that we don't * inadvertently do repair I/Os to the wrong device. * * This function will only return failure if one of the vdevs indicates that it * has since been destroyed or exported. This is only possible if * /etc/zfs/zpool.cache was readonly at the time. Otherwise, the vdev state * will be updated but the function will return 0. */ int vdev_validate(vdev_t *vd) { spa_t *spa = vd->vdev_spa; nvlist_t *label; uint64_t guid = 0, aux_guid = 0, top_guid; uint64_t state; nvlist_t *nvl; uint64_t txg; if (vdev_validate_skip) return (0); for (uint64_t c = 0; c < vd->vdev_children; c++) if (vdev_validate(vd->vdev_child[c]) != 0) return (SET_ERROR(EBADF)); /* * If the device has already failed, or was marked offline, don't do * any further validation. Otherwise, label I/O will fail and we will * overwrite the previous state. */ if (!vd->vdev_ops->vdev_op_leaf || !vdev_readable(vd)) return (0); /* * If we are performing an extreme rewind, we allow for a label that * was modified at a point after the current txg. * If config lock is not held do not check for the txg. spa_sync could * be updating the vdev's label before updating spa_last_synced_txg. */ if (spa->spa_extreme_rewind || spa_last_synced_txg(spa) == 0 || spa_config_held(spa, SCL_CONFIG, RW_WRITER) != SCL_CONFIG) txg = UINT64_MAX; else txg = spa_last_synced_txg(spa); if ((label = vdev_label_read_config(vd, txg)) == NULL) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_BAD_LABEL); vdev_dbgmsg(vd, "vdev_validate: failed reading config for " "txg %llu", (u_longlong_t)txg); return (0); } /* * Determine if this vdev has been split off into another * pool. If so, then refuse to open it. */ if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_SPLIT_GUID, &aux_guid) == 0 && aux_guid == spa_guid(spa)) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_SPLIT_POOL); nvlist_free(label); vdev_dbgmsg(vd, "vdev_validate: vdev split into other pool"); return (0); } if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_GUID, &guid) != 0) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); vdev_dbgmsg(vd, "vdev_validate: '%s' missing from label", ZPOOL_CONFIG_POOL_GUID); return (0); } /* * If config is not trusted then ignore the spa guid check. This is * necessary because if the machine crashed during a re-guid the new * guid might have been written to all of the vdev labels, but not the * cached config. The check will be performed again once we have the * trusted config from the MOS. */ if (spa->spa_trust_config && guid != spa_guid(spa)) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); vdev_dbgmsg(vd, "vdev_validate: vdev label pool_guid doesn't " "match config (%llu != %llu)", (u_longlong_t)guid, (u_longlong_t)spa_guid(spa)); return (0); } if (nvlist_lookup_nvlist(label, ZPOOL_CONFIG_VDEV_TREE, &nvl) != 0 || nvlist_lookup_uint64(nvl, ZPOOL_CONFIG_ORIG_GUID, &aux_guid) != 0) aux_guid = 0; if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_GUID, &guid) != 0) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); vdev_dbgmsg(vd, "vdev_validate: '%s' missing from label", ZPOOL_CONFIG_GUID); return (0); } if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_TOP_GUID, &top_guid) != 0) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); vdev_dbgmsg(vd, "vdev_validate: '%s' missing from label", ZPOOL_CONFIG_TOP_GUID); return (0); } /* * If this vdev just became a top-level vdev because its sibling was * detached, it will have adopted the parent's vdev guid -- but the * label may or may not be on disk yet. Fortunately, either version * of the label will have the same top guid, so if we're a top-level * vdev, we can safely compare to that instead. * However, if the config comes from a cachefile that failed to update * after the detach, a top-level vdev will appear as a non top-level * vdev in the config. Also relax the constraints if we perform an * extreme rewind. * * If we split this vdev off instead, then we also check the * original pool's guid. We don't want to consider the vdev * corrupt if it is partway through a split operation. */ if (vd->vdev_guid != guid && vd->vdev_guid != aux_guid) { boolean_t mismatch = B_FALSE; if (spa->spa_trust_config && !spa->spa_extreme_rewind) { if (vd != vd->vdev_top || vd->vdev_guid != top_guid) mismatch = B_TRUE; } else { if (vd->vdev_guid != top_guid && vd->vdev_top->vdev_guid != guid) mismatch = B_TRUE; } if (mismatch) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); vdev_dbgmsg(vd, "vdev_validate: config guid " "doesn't match label guid"); vdev_dbgmsg(vd, "CONFIG: guid %llu, top_guid %llu", (u_longlong_t)vd->vdev_guid, (u_longlong_t)vd->vdev_top->vdev_guid); vdev_dbgmsg(vd, "LABEL: guid %llu, top_guid %llu, " "aux_guid %llu", (u_longlong_t)guid, (u_longlong_t)top_guid, (u_longlong_t)aux_guid); return (0); } } if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_STATE, &state) != 0) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); vdev_dbgmsg(vd, "vdev_validate: '%s' missing from label", ZPOOL_CONFIG_POOL_STATE); return (0); } nvlist_free(label); /* * If this is a verbatim import, no need to check the * state of the pool. */ if (!(spa->spa_import_flags & ZFS_IMPORT_VERBATIM) && spa_load_state(spa) == SPA_LOAD_OPEN && state != POOL_STATE_ACTIVE) { vdev_dbgmsg(vd, "vdev_validate: invalid pool state (%llu) " "for spa %s", (u_longlong_t)state, spa->spa_name); return (SET_ERROR(EBADF)); } /* * If we were able to open and validate a vdev that was * previously marked permanently unavailable, clear that state * now. */ if (vd->vdev_not_present) vd->vdev_not_present = 0; return (0); } static void vdev_copy_path_impl(vdev_t *svd, vdev_t *dvd) { if (svd->vdev_path != NULL && dvd->vdev_path != NULL) { if (strcmp(svd->vdev_path, dvd->vdev_path) != 0) { zfs_dbgmsg("vdev_copy_path: vdev %llu: path changed " "from '%s' to '%s'", (u_longlong_t)dvd->vdev_guid, dvd->vdev_path, svd->vdev_path); spa_strfree(dvd->vdev_path); dvd->vdev_path = spa_strdup(svd->vdev_path); } } else if (svd->vdev_path != NULL) { dvd->vdev_path = spa_strdup(svd->vdev_path); zfs_dbgmsg("vdev_copy_path: vdev %llu: path set to '%s'", (u_longlong_t)dvd->vdev_guid, dvd->vdev_path); } } /* * Recursively copy vdev paths from one vdev to another. Source and destination * vdev trees must have same geometry otherwise return error. Intended to copy * paths from userland config into MOS config. */ int vdev_copy_path_strict(vdev_t *svd, vdev_t *dvd) { if ((svd->vdev_ops == &vdev_missing_ops) || (svd->vdev_ishole && dvd->vdev_ishole) || (dvd->vdev_ops == &vdev_indirect_ops)) return (0); if (svd->vdev_ops != dvd->vdev_ops) { vdev_dbgmsg(svd, "vdev_copy_path: vdev type mismatch: %s != %s", svd->vdev_ops->vdev_op_type, dvd->vdev_ops->vdev_op_type); return (SET_ERROR(EINVAL)); } if (svd->vdev_guid != dvd->vdev_guid) { vdev_dbgmsg(svd, "vdev_copy_path: guids mismatch (%llu != " "%llu)", (u_longlong_t)svd->vdev_guid, (u_longlong_t)dvd->vdev_guid); return (SET_ERROR(EINVAL)); } if (svd->vdev_children != dvd->vdev_children) { vdev_dbgmsg(svd, "vdev_copy_path: children count mismatch: " "%llu != %llu", (u_longlong_t)svd->vdev_children, (u_longlong_t)dvd->vdev_children); return (SET_ERROR(EINVAL)); } for (uint64_t i = 0; i < svd->vdev_children; i++) { int error = vdev_copy_path_strict(svd->vdev_child[i], dvd->vdev_child[i]); if (error != 0) return (error); } if (svd->vdev_ops->vdev_op_leaf) vdev_copy_path_impl(svd, dvd); return (0); } static void vdev_copy_path_search(vdev_t *stvd, vdev_t *dvd) { ASSERT(stvd->vdev_top == stvd); ASSERT3U(stvd->vdev_id, ==, dvd->vdev_top->vdev_id); for (uint64_t i = 0; i < dvd->vdev_children; i++) { vdev_copy_path_search(stvd, dvd->vdev_child[i]); } if (!dvd->vdev_ops->vdev_op_leaf || !vdev_is_concrete(dvd)) return; /* * The idea here is that while a vdev can shift positions within * a top vdev (when replacing, attaching mirror, etc.) it cannot * step outside of it. */ vdev_t *vd = vdev_lookup_by_guid(stvd, dvd->vdev_guid); if (vd == NULL || vd->vdev_ops != dvd->vdev_ops) return; ASSERT(vd->vdev_ops->vdev_op_leaf); vdev_copy_path_impl(vd, dvd); } /* * Recursively copy vdev paths from one root vdev to another. Source and * destination vdev trees may differ in geometry. For each destination leaf * vdev, search a vdev with the same guid and top vdev id in the source. * Intended to copy paths from userland config into MOS config. */ void vdev_copy_path_relaxed(vdev_t *srvd, vdev_t *drvd) { uint64_t children = MIN(srvd->vdev_children, drvd->vdev_children); ASSERT(srvd->vdev_ops == &vdev_root_ops); ASSERT(drvd->vdev_ops == &vdev_root_ops); for (uint64_t i = 0; i < children; i++) { vdev_copy_path_search(srvd->vdev_child[i], drvd->vdev_child[i]); } } /* * Close a virtual device. */ void vdev_close(vdev_t *vd) { vdev_t *pvd = vd->vdev_parent; spa_t *spa __maybe_unused = vd->vdev_spa; ASSERT(spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); /* * If our parent is reopening, then we are as well, unless we are * going offline. */ if (pvd != NULL && pvd->vdev_reopening) vd->vdev_reopening = (pvd->vdev_reopening && !vd->vdev_offline); vd->vdev_ops->vdev_op_close(vd); vdev_cache_purge(vd); /* * We record the previous state before we close it, so that if we are * doing a reopen(), we don't generate FMA ereports if we notice that * it's still faulted. */ vd->vdev_prevstate = vd->vdev_state; if (vd->vdev_offline) vd->vdev_state = VDEV_STATE_OFFLINE; else vd->vdev_state = VDEV_STATE_CLOSED; vd->vdev_stat.vs_aux = VDEV_AUX_NONE; } void vdev_hold(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT(spa_is_root(spa)); if (spa->spa_state == POOL_STATE_UNINITIALIZED) return; for (int c = 0; c < vd->vdev_children; c++) vdev_hold(vd->vdev_child[c]); if (vd->vdev_ops->vdev_op_leaf) vd->vdev_ops->vdev_op_hold(vd); } void vdev_rele(vdev_t *vd) { ASSERT(spa_is_root(vd->vdev_spa)); for (int c = 0; c < vd->vdev_children; c++) vdev_rele(vd->vdev_child[c]); if (vd->vdev_ops->vdev_op_leaf) vd->vdev_ops->vdev_op_rele(vd); } /* * Reopen all interior vdevs and any unopened leaves. We don't actually * reopen leaf vdevs which had previously been opened as they might deadlock * on the spa_config_lock. Instead we only obtain the leaf's physical size. * If the leaf has never been opened then open it, as usual. */ void vdev_reopen(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT(spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); /* set the reopening flag unless we're taking the vdev offline */ vd->vdev_reopening = !vd->vdev_offline; vdev_close(vd); (void) vdev_open(vd); /* * Call vdev_validate() here to make sure we have the same device. * Otherwise, a device with an invalid label could be successfully * opened in response to vdev_reopen(). */ if (vd->vdev_aux) { (void) vdev_validate_aux(vd); if (vdev_readable(vd) && vdev_writeable(vd) && vd->vdev_aux == &spa->spa_l2cache) { /* * In case the vdev is present we should evict all ARC * buffers and pointers to log blocks and reclaim their * space before restoring its contents to L2ARC. */ if (l2arc_vdev_present(vd)) { l2arc_rebuild_vdev(vd, B_TRUE); } else { l2arc_add_vdev(spa, vd); } spa_async_request(spa, SPA_ASYNC_L2CACHE_REBUILD); spa_async_request(spa, SPA_ASYNC_L2CACHE_TRIM); } } else { (void) vdev_validate(vd); } /* * Reassess parent vdev's health. */ vdev_propagate_state(vd); } int vdev_create(vdev_t *vd, uint64_t txg, boolean_t isreplacing) { int error; /* * Normally, partial opens (e.g. of a mirror) are allowed. * For a create, however, we want to fail the request if * there are any components we can't open. */ error = vdev_open(vd); if (error || vd->vdev_state != VDEV_STATE_HEALTHY) { vdev_close(vd); return (error ? error : SET_ERROR(ENXIO)); } /* * Recursively load DTLs and initialize all labels. */ if ((error = vdev_dtl_load(vd)) != 0 || (error = vdev_label_init(vd, txg, isreplacing ? VDEV_LABEL_REPLACE : VDEV_LABEL_CREATE)) != 0) { vdev_close(vd); return (error); } return (0); } void vdev_metaslab_set_size(vdev_t *vd) { uint64_t asize = vd->vdev_asize; uint64_t ms_count = asize >> zfs_vdev_default_ms_shift; uint64_t ms_shift; /* * There are two dimensions to the metaslab sizing calculation: * the size of the metaslab and the count of metaslabs per vdev. * * The default values used below are a good balance between memory * usage (larger metaslab size means more memory needed for loaded * metaslabs; more metaslabs means more memory needed for the * metaslab_t structs), metaslab load time (larger metaslabs take * longer to load), and metaslab sync time (more metaslabs means * more time spent syncing all of them). * * In general, we aim for zfs_vdev_default_ms_count (200) metaslabs. * The range of the dimensions are as follows: * * 2^29 <= ms_size <= 2^34 * 16 <= ms_count <= 131,072 * * On the lower end of vdev sizes, we aim for metaslabs sizes of * at least 512MB (2^29) to minimize fragmentation effects when * testing with smaller devices. However, the count constraint * of at least 16 metaslabs will override this minimum size goal. * * On the upper end of vdev sizes, we aim for a maximum metaslab * size of 16GB. However, we will cap the total count to 2^17 * metaslabs to keep our memory footprint in check and let the * metaslab size grow from there if that limit is hit. * * The net effect of applying above constrains is summarized below. * * vdev size metaslab count * --------------|----------------- * < 8GB ~16 * 8GB - 100GB one per 512MB * 100GB - 3TB ~200 * 3TB - 2PB one per 16GB * > 2PB ~131,072 * -------------------------------- * * Finally, note that all of the above calculate the initial * number of metaslabs. Expanding a top-level vdev will result * in additional metaslabs being allocated making it possible * to exceed the zfs_vdev_ms_count_limit. */ if (ms_count < zfs_vdev_min_ms_count) ms_shift = highbit64(asize / zfs_vdev_min_ms_count); else if (ms_count > zfs_vdev_default_ms_count) ms_shift = highbit64(asize / zfs_vdev_default_ms_count); else ms_shift = zfs_vdev_default_ms_shift; if (ms_shift < SPA_MAXBLOCKSHIFT) { ms_shift = SPA_MAXBLOCKSHIFT; } else if (ms_shift > zfs_vdev_max_ms_shift) { ms_shift = zfs_vdev_max_ms_shift; /* cap the total count to constrain memory footprint */ if ((asize >> ms_shift) > zfs_vdev_ms_count_limit) ms_shift = highbit64(asize / zfs_vdev_ms_count_limit); } vd->vdev_ms_shift = ms_shift; ASSERT3U(vd->vdev_ms_shift, >=, SPA_MAXBLOCKSHIFT); } void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg) { ASSERT(vd == vd->vdev_top); /* indirect vdevs don't have metaslabs or dtls */ ASSERT(vdev_is_concrete(vd) || flags == 0); ASSERT(ISP2(flags)); ASSERT(spa_writeable(vd->vdev_spa)); if (flags & VDD_METASLAB) (void) txg_list_add(&vd->vdev_ms_list, arg, txg); if (flags & VDD_DTL) (void) txg_list_add(&vd->vdev_dtl_list, arg, txg); (void) txg_list_add(&vd->vdev_spa->spa_vdev_txg_list, vd, txg); } void vdev_dirty_leaves(vdev_t *vd, int flags, uint64_t txg) { for (int c = 0; c < vd->vdev_children; c++) vdev_dirty_leaves(vd->vdev_child[c], flags, txg); if (vd->vdev_ops->vdev_op_leaf) vdev_dirty(vd->vdev_top, flags, vd, txg); } /* * DTLs. * * A vdev's DTL (dirty time log) is the set of transaction groups for which * the vdev has less than perfect replication. There are four kinds of DTL: * * DTL_MISSING: txgs for which the vdev has no valid copies of the data * * DTL_PARTIAL: txgs for which data is available, but not fully replicated * * DTL_SCRUB: the txgs that could not be repaired by the last scrub; upon * scrub completion, DTL_SCRUB replaces DTL_MISSING in the range of * txgs that was scrubbed. * * DTL_OUTAGE: txgs which cannot currently be read, whether due to * persistent errors or just some device being offline. * Unlike the other three, the DTL_OUTAGE map is not generally * maintained; it's only computed when needed, typically to * determine whether a device can be detached. * * For leaf vdevs, DTL_MISSING and DTL_PARTIAL are identical: the device * either has the data or it doesn't. * * For interior vdevs such as mirror and RAID-Z the picture is more complex. * A vdev's DTL_PARTIAL is the union of its children's DTL_PARTIALs, because * if any child is less than fully replicated, then so is its parent. * A vdev's DTL_MISSING is a modified union of its children's DTL_MISSINGs, * comprising only those txgs which appear in 'maxfaults' or more children; * those are the txgs we don't have enough replication to read. For example, * double-parity RAID-Z can tolerate up to two missing devices (maxfaults == 2); * thus, its DTL_MISSING consists of the set of txgs that appear in more than * two child DTL_MISSING maps. * * It should be clear from the above that to compute the DTLs and outage maps * for all vdevs, it suffices to know just the leaf vdevs' DTL_MISSING maps. * Therefore, that is all we keep on disk. When loading the pool, or after * a configuration change, we generate all other DTLs from first principles. */ void vdev_dtl_dirty(vdev_t *vd, vdev_dtl_type_t t, uint64_t txg, uint64_t size) { range_tree_t *rt = vd->vdev_dtl[t]; ASSERT(t < DTL_TYPES); ASSERT(vd != vd->vdev_spa->spa_root_vdev); ASSERT(spa_writeable(vd->vdev_spa)); mutex_enter(&vd->vdev_dtl_lock); if (!range_tree_contains(rt, txg, size)) range_tree_add(rt, txg, size); mutex_exit(&vd->vdev_dtl_lock); } boolean_t vdev_dtl_contains(vdev_t *vd, vdev_dtl_type_t t, uint64_t txg, uint64_t size) { range_tree_t *rt = vd->vdev_dtl[t]; boolean_t dirty = B_FALSE; ASSERT(t < DTL_TYPES); ASSERT(vd != vd->vdev_spa->spa_root_vdev); /* * While we are loading the pool, the DTLs have not been loaded yet. * Ignore the DTLs and try all devices. This avoids a recursive * mutex enter on the vdev_dtl_lock, and also makes us try hard * when loading the pool (relying on the checksum to ensure that * we get the right data -- note that we while loading, we are * only reading the MOS, which is always checksummed). */ if (vd->vdev_spa->spa_load_state != SPA_LOAD_NONE) return (B_FALSE); mutex_enter(&vd->vdev_dtl_lock); if (!range_tree_is_empty(rt)) dirty = range_tree_contains(rt, txg, size); mutex_exit(&vd->vdev_dtl_lock); return (dirty); } boolean_t vdev_dtl_empty(vdev_t *vd, vdev_dtl_type_t t) { range_tree_t *rt = vd->vdev_dtl[t]; boolean_t empty; mutex_enter(&vd->vdev_dtl_lock); empty = range_tree_is_empty(rt); mutex_exit(&vd->vdev_dtl_lock); return (empty); } /* * Returns B_TRUE if vdev determines offset needs to be resilvered. */ boolean_t vdev_dtl_need_resilver(vdev_t *vd, uint64_t offset, size_t psize) { ASSERT(vd != vd->vdev_spa->spa_root_vdev); if (vd->vdev_ops->vdev_op_need_resilver == NULL || vd->vdev_ops->vdev_op_leaf) return (B_TRUE); return (vd->vdev_ops->vdev_op_need_resilver(vd, offset, psize)); } /* * Returns the lowest txg in the DTL range. */ static uint64_t vdev_dtl_min(vdev_t *vd) { ASSERT(MUTEX_HELD(&vd->vdev_dtl_lock)); ASSERT3U(range_tree_space(vd->vdev_dtl[DTL_MISSING]), !=, 0); ASSERT0(vd->vdev_children); return (range_tree_min(vd->vdev_dtl[DTL_MISSING]) - 1); } /* * Returns the highest txg in the DTL. */ static uint64_t vdev_dtl_max(vdev_t *vd) { ASSERT(MUTEX_HELD(&vd->vdev_dtl_lock)); ASSERT3U(range_tree_space(vd->vdev_dtl[DTL_MISSING]), !=, 0); ASSERT0(vd->vdev_children); return (range_tree_max(vd->vdev_dtl[DTL_MISSING])); } /* * Determine if a resilvering vdev should remove any DTL entries from * its range. If the vdev was resilvering for the entire duration of the * scan then it should excise that range from its DTLs. Otherwise, this * vdev is considered partially resilvered and should leave its DTL * entries intact. The comment in vdev_dtl_reassess() describes how we * excise the DTLs. */ static boolean_t vdev_dtl_should_excise(vdev_t *vd, boolean_t rebuild_done) { ASSERT0(vd->vdev_children); if (vd->vdev_state < VDEV_STATE_DEGRADED) return (B_FALSE); if (vd->vdev_resilver_deferred) return (B_FALSE); if (range_tree_is_empty(vd->vdev_dtl[DTL_MISSING])) return (B_TRUE); if (rebuild_done) { vdev_rebuild_t *vr = &vd->vdev_top->vdev_rebuild_config; vdev_rebuild_phys_t *vrp = &vr->vr_rebuild_phys; /* Rebuild not initiated by attach */ if (vd->vdev_rebuild_txg == 0) return (B_TRUE); /* * When a rebuild completes without error then all missing data * up to the rebuild max txg has been reconstructed and the DTL * is eligible for excision. */ if (vrp->vrp_rebuild_state == VDEV_REBUILD_COMPLETE && vdev_dtl_max(vd) <= vrp->vrp_max_txg) { ASSERT3U(vrp->vrp_min_txg, <=, vdev_dtl_min(vd)); ASSERT3U(vrp->vrp_min_txg, <, vd->vdev_rebuild_txg); ASSERT3U(vd->vdev_rebuild_txg, <=, vrp->vrp_max_txg); return (B_TRUE); } } else { dsl_scan_t *scn = vd->vdev_spa->spa_dsl_pool->dp_scan; dsl_scan_phys_t *scnp __maybe_unused = &scn->scn_phys; /* Resilver not initiated by attach */ if (vd->vdev_resilver_txg == 0) return (B_TRUE); /* * When a resilver is initiated the scan will assign the * scn_max_txg value to the highest txg value that exists * in all DTLs. If this device's max DTL is not part of this * scan (i.e. it is not in the range (scn_min_txg, scn_max_txg] * then it is not eligible for excision. */ if (vdev_dtl_max(vd) <= scn->scn_phys.scn_max_txg) { ASSERT3U(scnp->scn_min_txg, <=, vdev_dtl_min(vd)); ASSERT3U(scnp->scn_min_txg, <, vd->vdev_resilver_txg); ASSERT3U(vd->vdev_resilver_txg, <=, scnp->scn_max_txg); return (B_TRUE); } } return (B_FALSE); } /* * Reassess DTLs after a config change or scrub completion. If txg == 0 no * write operations will be issued to the pool. */ void vdev_dtl_reassess(vdev_t *vd, uint64_t txg, uint64_t scrub_txg, boolean_t scrub_done, boolean_t rebuild_done) { spa_t *spa = vd->vdev_spa; avl_tree_t reftree; int minref; ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0); for (int c = 0; c < vd->vdev_children; c++) vdev_dtl_reassess(vd->vdev_child[c], txg, scrub_txg, scrub_done, rebuild_done); if (vd == spa->spa_root_vdev || !vdev_is_concrete(vd) || vd->vdev_aux) return; if (vd->vdev_ops->vdev_op_leaf) { dsl_scan_t *scn = spa->spa_dsl_pool->dp_scan; vdev_rebuild_t *vr = &vd->vdev_top->vdev_rebuild_config; boolean_t check_excise = B_FALSE; boolean_t wasempty = B_TRUE; mutex_enter(&vd->vdev_dtl_lock); /* * If requested, pretend the scan or rebuild completed cleanly. */ if (zfs_scan_ignore_errors) { if (scn != NULL) scn->scn_phys.scn_errors = 0; if (vr != NULL) vr->vr_rebuild_phys.vrp_errors = 0; } if (scrub_txg != 0 && !range_tree_is_empty(vd->vdev_dtl[DTL_MISSING])) { wasempty = B_FALSE; zfs_dbgmsg("guid:%llu txg:%llu scrub:%llu started:%d " "dtl:%llu/%llu errors:%llu", (u_longlong_t)vd->vdev_guid, (u_longlong_t)txg, (u_longlong_t)scrub_txg, spa->spa_scrub_started, (u_longlong_t)vdev_dtl_min(vd), (u_longlong_t)vdev_dtl_max(vd), (u_longlong_t)(scn ? scn->scn_phys.scn_errors : 0)); } /* * If we've completed a scrub/resilver or a rebuild cleanly * then determine if this vdev should remove any DTLs. We * only want to excise regions on vdevs that were available * during the entire duration of this scan. */ if (rebuild_done && vr != NULL && vr->vr_rebuild_phys.vrp_errors == 0) { check_excise = B_TRUE; } else { if (spa->spa_scrub_started || (scn != NULL && scn->scn_phys.scn_errors == 0)) { check_excise = B_TRUE; } } if (scrub_txg && check_excise && vdev_dtl_should_excise(vd, rebuild_done)) { /* * We completed a scrub, resilver or rebuild up to * scrub_txg. If we did it without rebooting, then * the scrub dtl will be valid, so excise the old * region and fold in the scrub dtl. Otherwise, * leave the dtl as-is if there was an error. * * There's little trick here: to excise the beginning * of the DTL_MISSING map, we put it into a reference * tree and then add a segment with refcnt -1 that * covers the range [0, scrub_txg). This means * that each txg in that range has refcnt -1 or 0. * We then add DTL_SCRUB with a refcnt of 2, so that * entries in the range [0, scrub_txg) will have a * positive refcnt -- either 1 or 2. We then convert * the reference tree into the new DTL_MISSING map. */ space_reftree_create(&reftree); space_reftree_add_map(&reftree, vd->vdev_dtl[DTL_MISSING], 1); space_reftree_add_seg(&reftree, 0, scrub_txg, -1); space_reftree_add_map(&reftree, vd->vdev_dtl[DTL_SCRUB], 2); space_reftree_generate_map(&reftree, vd->vdev_dtl[DTL_MISSING], 1); space_reftree_destroy(&reftree); if (!range_tree_is_empty(vd->vdev_dtl[DTL_MISSING])) { zfs_dbgmsg("update DTL_MISSING:%llu/%llu", (u_longlong_t)vdev_dtl_min(vd), (u_longlong_t)vdev_dtl_max(vd)); } else if (!wasempty) { zfs_dbgmsg("DTL_MISSING is now empty"); } } range_tree_vacate(vd->vdev_dtl[DTL_PARTIAL], NULL, NULL); range_tree_walk(vd->vdev_dtl[DTL_MISSING], range_tree_add, vd->vdev_dtl[DTL_PARTIAL]); if (scrub_done) range_tree_vacate(vd->vdev_dtl[DTL_SCRUB], NULL, NULL); range_tree_vacate(vd->vdev_dtl[DTL_OUTAGE], NULL, NULL); if (!vdev_readable(vd)) range_tree_add(vd->vdev_dtl[DTL_OUTAGE], 0, -1ULL); else range_tree_walk(vd->vdev_dtl[DTL_MISSING], range_tree_add, vd->vdev_dtl[DTL_OUTAGE]); /* * If the vdev was resilvering or rebuilding and no longer * has any DTLs then reset the appropriate flag and dirty * the top level so that we persist the change. */ if (txg != 0 && range_tree_is_empty(vd->vdev_dtl[DTL_MISSING]) && range_tree_is_empty(vd->vdev_dtl[DTL_OUTAGE])) { if (vd->vdev_rebuild_txg != 0) { vd->vdev_rebuild_txg = 0; vdev_config_dirty(vd->vdev_top); } else if (vd->vdev_resilver_txg != 0) { vd->vdev_resilver_txg = 0; vdev_config_dirty(vd->vdev_top); } } mutex_exit(&vd->vdev_dtl_lock); if (txg != 0) vdev_dirty(vd->vdev_top, VDD_DTL, vd, txg); return; } mutex_enter(&vd->vdev_dtl_lock); for (int t = 0; t < DTL_TYPES; t++) { /* account for child's outage in parent's missing map */ int s = (t == DTL_MISSING) ? DTL_OUTAGE: t; if (t == DTL_SCRUB) continue; /* leaf vdevs only */ if (t == DTL_PARTIAL) minref = 1; /* i.e. non-zero */ else if (vd->vdev_nparity != 0) minref = vd->vdev_nparity + 1; /* RAID-Z */ else minref = vd->vdev_children; /* any kind of mirror */ space_reftree_create(&reftree); for (int c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; mutex_enter(&cvd->vdev_dtl_lock); space_reftree_add_map(&reftree, cvd->vdev_dtl[s], 1); mutex_exit(&cvd->vdev_dtl_lock); } space_reftree_generate_map(&reftree, vd->vdev_dtl[t], minref); space_reftree_destroy(&reftree); } mutex_exit(&vd->vdev_dtl_lock); } int vdev_dtl_load(vdev_t *vd) { spa_t *spa = vd->vdev_spa; objset_t *mos = spa->spa_meta_objset; int error = 0; if (vd->vdev_ops->vdev_op_leaf && vd->vdev_dtl_object != 0) { ASSERT(vdev_is_concrete(vd)); error = space_map_open(&vd->vdev_dtl_sm, mos, vd->vdev_dtl_object, 0, -1ULL, 0); if (error) return (error); ASSERT(vd->vdev_dtl_sm != NULL); mutex_enter(&vd->vdev_dtl_lock); error = space_map_load(vd->vdev_dtl_sm, vd->vdev_dtl[DTL_MISSING], SM_ALLOC); mutex_exit(&vd->vdev_dtl_lock); return (error); } for (int c = 0; c < vd->vdev_children; c++) { error = vdev_dtl_load(vd->vdev_child[c]); if (error != 0) break; } return (error); } static void vdev_zap_allocation_data(vdev_t *vd, dmu_tx_t *tx) { spa_t *spa = vd->vdev_spa; objset_t *mos = spa->spa_meta_objset; vdev_alloc_bias_t alloc_bias = vd->vdev_alloc_bias; const char *string; ASSERT(alloc_bias != VDEV_BIAS_NONE); string = (alloc_bias == VDEV_BIAS_LOG) ? VDEV_ALLOC_BIAS_LOG : (alloc_bias == VDEV_BIAS_SPECIAL) ? VDEV_ALLOC_BIAS_SPECIAL : (alloc_bias == VDEV_BIAS_DEDUP) ? VDEV_ALLOC_BIAS_DEDUP : NULL; ASSERT(string != NULL); VERIFY0(zap_add(mos, vd->vdev_top_zap, VDEV_TOP_ZAP_ALLOCATION_BIAS, 1, strlen(string) + 1, string, tx)); if (alloc_bias == VDEV_BIAS_SPECIAL || alloc_bias == VDEV_BIAS_DEDUP) { spa_activate_allocation_classes(spa, tx); } } void vdev_destroy_unlink_zap(vdev_t *vd, uint64_t zapobj, dmu_tx_t *tx) { spa_t *spa = vd->vdev_spa; VERIFY0(zap_destroy(spa->spa_meta_objset, zapobj, tx)); VERIFY0(zap_remove_int(spa->spa_meta_objset, spa->spa_all_vdev_zaps, zapobj, tx)); } uint64_t vdev_create_link_zap(vdev_t *vd, dmu_tx_t *tx) { spa_t *spa = vd->vdev_spa; uint64_t zap = zap_create(spa->spa_meta_objset, DMU_OTN_ZAP_METADATA, DMU_OT_NONE, 0, tx); ASSERT(zap != 0); VERIFY0(zap_add_int(spa->spa_meta_objset, spa->spa_all_vdev_zaps, zap, tx)); return (zap); } void vdev_construct_zaps(vdev_t *vd, dmu_tx_t *tx) { if (vd->vdev_ops != &vdev_hole_ops && vd->vdev_ops != &vdev_missing_ops && vd->vdev_ops != &vdev_root_ops && !vd->vdev_top->vdev_removing) { if (vd->vdev_ops->vdev_op_leaf && vd->vdev_leaf_zap == 0) { vd->vdev_leaf_zap = vdev_create_link_zap(vd, tx); } if (vd == vd->vdev_top && vd->vdev_top_zap == 0) { vd->vdev_top_zap = vdev_create_link_zap(vd, tx); if (vd->vdev_alloc_bias != VDEV_BIAS_NONE) vdev_zap_allocation_data(vd, tx); } } for (uint64_t i = 0; i < vd->vdev_children; i++) { vdev_construct_zaps(vd->vdev_child[i], tx); } } static void vdev_dtl_sync(vdev_t *vd, uint64_t txg) { spa_t *spa = vd->vdev_spa; range_tree_t *rt = vd->vdev_dtl[DTL_MISSING]; objset_t *mos = spa->spa_meta_objset; range_tree_t *rtsync; dmu_tx_t *tx; uint64_t object = space_map_object(vd->vdev_dtl_sm); ASSERT(vdev_is_concrete(vd)); ASSERT(vd->vdev_ops->vdev_op_leaf); tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg); if (vd->vdev_detached || vd->vdev_top->vdev_removing) { mutex_enter(&vd->vdev_dtl_lock); space_map_free(vd->vdev_dtl_sm, tx); space_map_close(vd->vdev_dtl_sm); vd->vdev_dtl_sm = NULL; mutex_exit(&vd->vdev_dtl_lock); /* * We only destroy the leaf ZAP for detached leaves or for * removed log devices. Removed data devices handle leaf ZAP * cleanup later, once cancellation is no longer possible. */ if (vd->vdev_leaf_zap != 0 && (vd->vdev_detached || vd->vdev_top->vdev_islog)) { vdev_destroy_unlink_zap(vd, vd->vdev_leaf_zap, tx); vd->vdev_leaf_zap = 0; } dmu_tx_commit(tx); return; } if (vd->vdev_dtl_sm == NULL) { uint64_t new_object; new_object = space_map_alloc(mos, zfs_vdev_dtl_sm_blksz, tx); VERIFY3U(new_object, !=, 0); VERIFY0(space_map_open(&vd->vdev_dtl_sm, mos, new_object, 0, -1ULL, 0)); ASSERT(vd->vdev_dtl_sm != NULL); } rtsync = range_tree_create(NULL, RANGE_SEG64, NULL, 0, 0); mutex_enter(&vd->vdev_dtl_lock); range_tree_walk(rt, range_tree_add, rtsync); mutex_exit(&vd->vdev_dtl_lock); space_map_truncate(vd->vdev_dtl_sm, zfs_vdev_dtl_sm_blksz, tx); space_map_write(vd->vdev_dtl_sm, rtsync, SM_ALLOC, SM_NO_VDEVID, tx); range_tree_vacate(rtsync, NULL, NULL); range_tree_destroy(rtsync); /* * If the object for the space map has changed then dirty * the top level so that we update the config. */ if (object != space_map_object(vd->vdev_dtl_sm)) { vdev_dbgmsg(vd, "txg %llu, spa %s, DTL old object %llu, " "new object %llu", (u_longlong_t)txg, spa_name(spa), (u_longlong_t)object, (u_longlong_t)space_map_object(vd->vdev_dtl_sm)); vdev_config_dirty(vd->vdev_top); } dmu_tx_commit(tx); } /* * Determine whether the specified vdev can be offlined/detached/removed * without losing data. */ boolean_t vdev_dtl_required(vdev_t *vd) { spa_t *spa = vd->vdev_spa; vdev_t *tvd = vd->vdev_top; uint8_t cant_read = vd->vdev_cant_read; boolean_t required; ASSERT(spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); if (vd == spa->spa_root_vdev || vd == tvd) return (B_TRUE); /* * Temporarily mark the device as unreadable, and then determine * whether this results in any DTL outages in the top-level vdev. * If not, we can safely offline/detach/remove the device. */ vd->vdev_cant_read = B_TRUE; vdev_dtl_reassess(tvd, 0, 0, B_FALSE, B_FALSE); required = !vdev_dtl_empty(tvd, DTL_OUTAGE); vd->vdev_cant_read = cant_read; vdev_dtl_reassess(tvd, 0, 0, B_FALSE, B_FALSE); if (!required && zio_injection_enabled) { required = !!zio_handle_device_injection(vd, NULL, SET_ERROR(ECHILD)); } return (required); } /* * Determine if resilver is needed, and if so the txg range. */ boolean_t vdev_resilver_needed(vdev_t *vd, uint64_t *minp, uint64_t *maxp) { boolean_t needed = B_FALSE; uint64_t thismin = UINT64_MAX; uint64_t thismax = 0; if (vd->vdev_children == 0) { mutex_enter(&vd->vdev_dtl_lock); if (!range_tree_is_empty(vd->vdev_dtl[DTL_MISSING]) && vdev_writeable(vd)) { thismin = vdev_dtl_min(vd); thismax = vdev_dtl_max(vd); needed = B_TRUE; } mutex_exit(&vd->vdev_dtl_lock); } else { for (int c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; uint64_t cmin, cmax; if (vdev_resilver_needed(cvd, &cmin, &cmax)) { thismin = MIN(thismin, cmin); thismax = MAX(thismax, cmax); needed = B_TRUE; } } } if (needed && minp) { *minp = thismin; *maxp = thismax; } return (needed); } /* * Gets the checkpoint space map object from the vdev's ZAP. On success sm_obj * will contain either the checkpoint spacemap object or zero if none exists. * All other errors are returned to the caller. */ int vdev_checkpoint_sm_object(vdev_t *vd, uint64_t *sm_obj) { ASSERT0(spa_config_held(vd->vdev_spa, SCL_ALL, RW_WRITER)); if (vd->vdev_top_zap == 0) { *sm_obj = 0; return (0); } int error = zap_lookup(spa_meta_objset(vd->vdev_spa), vd->vdev_top_zap, VDEV_TOP_ZAP_POOL_CHECKPOINT_SM, sizeof (uint64_t), 1, sm_obj); if (error == ENOENT) { *sm_obj = 0; error = 0; } return (error); } int vdev_load(vdev_t *vd) { int error = 0; /* * Recursively load all children. */ for (int c = 0; c < vd->vdev_children; c++) { error = vdev_load(vd->vdev_child[c]); if (error != 0) { return (error); } } vdev_set_deflate_ratio(vd); /* * On spa_load path, grab the allocation bias from our zap */ if (vd == vd->vdev_top && vd->vdev_top_zap != 0) { spa_t *spa = vd->vdev_spa; char bias_str[64]; error = zap_lookup(spa->spa_meta_objset, vd->vdev_top_zap, VDEV_TOP_ZAP_ALLOCATION_BIAS, 1, sizeof (bias_str), bias_str); if (error == 0) { ASSERT(vd->vdev_alloc_bias == VDEV_BIAS_NONE); vd->vdev_alloc_bias = vdev_derive_alloc_bias(bias_str); } else if (error != ENOENT) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); vdev_dbgmsg(vd, "vdev_load: zap_lookup(top_zap=%llu) " "failed [error=%d]", vd->vdev_top_zap, error); return (error); } } /* * Load any rebuild state from the top-level vdev zap. */ if (vd == vd->vdev_top && vd->vdev_top_zap != 0) { error = vdev_rebuild_load(vd); if (error && error != ENOTSUP) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); vdev_dbgmsg(vd, "vdev_load: vdev_rebuild_load " "failed [error=%d]", error); return (error); } } /* * If this is a top-level vdev, initialize its metaslabs. */ if (vd == vd->vdev_top && vdev_is_concrete(vd)) { vdev_metaslab_group_create(vd); if (vd->vdev_ashift == 0 || vd->vdev_asize == 0) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); vdev_dbgmsg(vd, "vdev_load: invalid size. ashift=%llu, " "asize=%llu", (u_longlong_t)vd->vdev_ashift, (u_longlong_t)vd->vdev_asize); return (SET_ERROR(ENXIO)); } error = vdev_metaslab_init(vd, 0); if (error != 0) { vdev_dbgmsg(vd, "vdev_load: metaslab_init failed " "[error=%d]", error); vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); return (error); } uint64_t checkpoint_sm_obj; error = vdev_checkpoint_sm_object(vd, &checkpoint_sm_obj); if (error == 0 && checkpoint_sm_obj != 0) { objset_t *mos = spa_meta_objset(vd->vdev_spa); ASSERT(vd->vdev_asize != 0); ASSERT3P(vd->vdev_checkpoint_sm, ==, NULL); error = space_map_open(&vd->vdev_checkpoint_sm, mos, checkpoint_sm_obj, 0, vd->vdev_asize, vd->vdev_ashift); if (error != 0) { vdev_dbgmsg(vd, "vdev_load: space_map_open " "failed for checkpoint spacemap (obj %llu) " "[error=%d]", (u_longlong_t)checkpoint_sm_obj, error); return (error); } ASSERT3P(vd->vdev_checkpoint_sm, !=, NULL); /* * Since the checkpoint_sm contains free entries * exclusively we can use space_map_allocated() to * indicate the cumulative checkpointed space that * has been freed. */ vd->vdev_stat.vs_checkpoint_space = -space_map_allocated(vd->vdev_checkpoint_sm); vd->vdev_spa->spa_checkpoint_info.sci_dspace += vd->vdev_stat.vs_checkpoint_space; } else if (error != 0) { vdev_dbgmsg(vd, "vdev_load: failed to retrieve " "checkpoint space map object from vdev ZAP " "[error=%d]", error); return (error); } } /* * If this is a leaf vdev, load its DTL. */ if (vd->vdev_ops->vdev_op_leaf && (error = vdev_dtl_load(vd)) != 0) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); vdev_dbgmsg(vd, "vdev_load: vdev_dtl_load failed " "[error=%d]", error); return (error); } uint64_t obsolete_sm_object; error = vdev_obsolete_sm_object(vd, &obsolete_sm_object); if (error == 0 && obsolete_sm_object != 0) { objset_t *mos = vd->vdev_spa->spa_meta_objset; ASSERT(vd->vdev_asize != 0); ASSERT3P(vd->vdev_obsolete_sm, ==, NULL); if ((error = space_map_open(&vd->vdev_obsolete_sm, mos, obsolete_sm_object, 0, vd->vdev_asize, 0))) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); vdev_dbgmsg(vd, "vdev_load: space_map_open failed for " "obsolete spacemap (obj %llu) [error=%d]", (u_longlong_t)obsolete_sm_object, error); return (error); } } else if (error != 0) { vdev_dbgmsg(vd, "vdev_load: failed to retrieve obsolete " "space map object from vdev ZAP [error=%d]", error); return (error); } return (0); } /* * The special vdev case is used for hot spares and l2cache devices. Its * sole purpose it to set the vdev state for the associated vdev. To do this, * we make sure that we can open the underlying device, then try to read the * label, and make sure that the label is sane and that it hasn't been * repurposed to another pool. */ int vdev_validate_aux(vdev_t *vd) { nvlist_t *label; uint64_t guid, version; uint64_t state; if (!vdev_readable(vd)) return (0); if ((label = vdev_label_read_config(vd, -1ULL)) == NULL) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); return (-1); } if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_VERSION, &version) != 0 || !SPA_VERSION_IS_SUPPORTED(version) || nvlist_lookup_uint64(label, ZPOOL_CONFIG_GUID, &guid) != 0 || guid != vd->vdev_guid || nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_STATE, &state) != 0) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); return (-1); } /* * We don't actually check the pool state here. If it's in fact in * use by another pool, we update this fact on the fly when requested. */ nvlist_free(label); return (0); } static void vdev_destroy_ms_flush_data(vdev_t *vd, dmu_tx_t *tx) { objset_t *mos = spa_meta_objset(vd->vdev_spa); if (vd->vdev_top_zap == 0) return; uint64_t object = 0; int err = zap_lookup(mos, vd->vdev_top_zap, VDEV_TOP_ZAP_MS_UNFLUSHED_PHYS_TXGS, sizeof (uint64_t), 1, &object); if (err == ENOENT) return; VERIFY0(err); VERIFY0(dmu_object_free(mos, object, tx)); VERIFY0(zap_remove(mos, vd->vdev_top_zap, VDEV_TOP_ZAP_MS_UNFLUSHED_PHYS_TXGS, tx)); } /* * Free the objects used to store this vdev's spacemaps, and the array * that points to them. */ void vdev_destroy_spacemaps(vdev_t *vd, dmu_tx_t *tx) { if (vd->vdev_ms_array == 0) return; objset_t *mos = vd->vdev_spa->spa_meta_objset; uint64_t array_count = vd->vdev_asize >> vd->vdev_ms_shift; size_t array_bytes = array_count * sizeof (uint64_t); uint64_t *smobj_array = kmem_alloc(array_bytes, KM_SLEEP); VERIFY0(dmu_read(mos, vd->vdev_ms_array, 0, array_bytes, smobj_array, 0)); for (uint64_t i = 0; i < array_count; i++) { uint64_t smobj = smobj_array[i]; if (smobj == 0) continue; space_map_free_obj(mos, smobj, tx); } kmem_free(smobj_array, array_bytes); VERIFY0(dmu_object_free(mos, vd->vdev_ms_array, tx)); vdev_destroy_ms_flush_data(vd, tx); vd->vdev_ms_array = 0; } static void vdev_remove_empty_log(vdev_t *vd, uint64_t txg) { spa_t *spa = vd->vdev_spa; ASSERT(vd->vdev_islog); ASSERT(vd == vd->vdev_top); ASSERT3U(txg, ==, spa_syncing_txg(spa)); dmu_tx_t *tx = dmu_tx_create_assigned(spa_get_dsl(spa), txg); vdev_destroy_spacemaps(vd, tx); if (vd->vdev_top_zap != 0) { vdev_destroy_unlink_zap(vd, vd->vdev_top_zap, tx); vd->vdev_top_zap = 0; } dmu_tx_commit(tx); } void vdev_sync_done(vdev_t *vd, uint64_t txg) { metaslab_t *msp; boolean_t reassess = !txg_list_empty(&vd->vdev_ms_list, TXG_CLEAN(txg)); ASSERT(vdev_is_concrete(vd)); while ((msp = txg_list_remove(&vd->vdev_ms_list, TXG_CLEAN(txg))) != NULL) metaslab_sync_done(msp, txg); if (reassess) metaslab_sync_reassess(vd->vdev_mg); } void vdev_sync(vdev_t *vd, uint64_t txg) { spa_t *spa = vd->vdev_spa; vdev_t *lvd; metaslab_t *msp; ASSERT3U(txg, ==, spa->spa_syncing_txg); dmu_tx_t *tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg); if (range_tree_space(vd->vdev_obsolete_segments) > 0) { ASSERT(vd->vdev_removing || vd->vdev_ops == &vdev_indirect_ops); vdev_indirect_sync_obsolete(vd, tx); /* * If the vdev is indirect, it can't have dirty * metaslabs or DTLs. */ if (vd->vdev_ops == &vdev_indirect_ops) { ASSERT(txg_list_empty(&vd->vdev_ms_list, txg)); ASSERT(txg_list_empty(&vd->vdev_dtl_list, txg)); dmu_tx_commit(tx); return; } } ASSERT(vdev_is_concrete(vd)); if (vd->vdev_ms_array == 0 && vd->vdev_ms_shift != 0 && !vd->vdev_removing) { ASSERT(vd == vd->vdev_top); ASSERT0(vd->vdev_indirect_config.vic_mapping_object); vd->vdev_ms_array = dmu_object_alloc(spa->spa_meta_objset, DMU_OT_OBJECT_ARRAY, 0, DMU_OT_NONE, 0, tx); ASSERT(vd->vdev_ms_array != 0); vdev_config_dirty(vd); } while ((msp = txg_list_remove(&vd->vdev_ms_list, txg)) != NULL) { metaslab_sync(msp, txg); (void) txg_list_add(&vd->vdev_ms_list, msp, TXG_CLEAN(txg)); } while ((lvd = txg_list_remove(&vd->vdev_dtl_list, txg)) != NULL) vdev_dtl_sync(lvd, txg); /* * If this is an empty log device being removed, destroy the * metadata associated with it. */ if (vd->vdev_islog && vd->vdev_stat.vs_alloc == 0 && vd->vdev_removing) vdev_remove_empty_log(vd, txg); (void) txg_list_add(&spa->spa_vdev_txg_list, vd, TXG_CLEAN(txg)); dmu_tx_commit(tx); } uint64_t vdev_psize_to_asize(vdev_t *vd, uint64_t psize) { return (vd->vdev_ops->vdev_op_asize(vd, psize)); } /* * Mark the given vdev faulted. A faulted vdev behaves as if the device could * not be opened, and no I/O is attempted. */ int vdev_fault(spa_t *spa, uint64_t guid, vdev_aux_t aux) { vdev_t *vd, *tvd; spa_vdev_state_enter(spa, SCL_NONE); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENODEV))); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENOTSUP))); tvd = vd->vdev_top; /* * If user did a 'zpool offline -f' then make the fault persist across * reboots. */ if (aux == VDEV_AUX_EXTERNAL_PERSIST) { /* * There are two kinds of forced faults: temporary and * persistent. Temporary faults go away at pool import, while * persistent faults stay set. Both types of faults can be * cleared with a zpool clear. * * We tell if a vdev is persistently faulted by looking at the * ZPOOL_CONFIG_AUX_STATE nvpair. If it's set to "external" at * import then it's a persistent fault. Otherwise, it's * temporary. We get ZPOOL_CONFIG_AUX_STATE set to "external" * by setting vd.vdev_stat.vs_aux to VDEV_AUX_EXTERNAL. This * tells vdev_config_generate() (which gets run later) to set * ZPOOL_CONFIG_AUX_STATE to "external" in the nvlist. */ vd->vdev_stat.vs_aux = VDEV_AUX_EXTERNAL; vd->vdev_tmpoffline = B_FALSE; aux = VDEV_AUX_EXTERNAL; } else { vd->vdev_tmpoffline = B_TRUE; } /* * We don't directly use the aux state here, but if we do a * vdev_reopen(), we need this value to be present to remember why we * were faulted. */ vd->vdev_label_aux = aux; /* * Faulted state takes precedence over degraded. */ vd->vdev_delayed_close = B_FALSE; vd->vdev_faulted = 1ULL; vd->vdev_degraded = 0ULL; vdev_set_state(vd, B_FALSE, VDEV_STATE_FAULTED, aux); /* * If this device has the only valid copy of the data, then * back off and simply mark the vdev as degraded instead. */ if (!tvd->vdev_islog && vd->vdev_aux == NULL && vdev_dtl_required(vd)) { vd->vdev_degraded = 1ULL; vd->vdev_faulted = 0ULL; /* * If we reopen the device and it's not dead, only then do we * mark it degraded. */ vdev_reopen(tvd); if (vdev_readable(vd)) vdev_set_state(vd, B_FALSE, VDEV_STATE_DEGRADED, aux); } return (spa_vdev_state_exit(spa, vd, 0)); } /* * Mark the given vdev degraded. A degraded vdev is purely an indication to the * user that something is wrong. The vdev continues to operate as normal as far * as I/O is concerned. */ int vdev_degrade(spa_t *spa, uint64_t guid, vdev_aux_t aux) { vdev_t *vd; spa_vdev_state_enter(spa, SCL_NONE); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENODEV))); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENOTSUP))); /* * If the vdev is already faulted, then don't do anything. */ if (vd->vdev_faulted || vd->vdev_degraded) return (spa_vdev_state_exit(spa, NULL, 0)); vd->vdev_degraded = 1ULL; if (!vdev_is_dead(vd)) vdev_set_state(vd, B_FALSE, VDEV_STATE_DEGRADED, aux); return (spa_vdev_state_exit(spa, vd, 0)); } /* * Online the given vdev. * * If 'ZFS_ONLINE_UNSPARE' is set, it implies two things. First, any attached * spare device should be detached when the device finishes resilvering. * Second, the online should be treated like a 'test' online case, so no FMA * events are generated if the device fails to open. */ int vdev_online(spa_t *spa, uint64_t guid, uint64_t flags, vdev_state_t *newstate) { vdev_t *vd, *tvd, *pvd, *rvd = spa->spa_root_vdev; boolean_t wasoffline; vdev_state_t oldstate; spa_vdev_state_enter(spa, SCL_NONE); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENODEV))); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENOTSUP))); wasoffline = (vd->vdev_offline || vd->vdev_tmpoffline); oldstate = vd->vdev_state; tvd = vd->vdev_top; vd->vdev_offline = B_FALSE; vd->vdev_tmpoffline = B_FALSE; vd->vdev_checkremove = !!(flags & ZFS_ONLINE_CHECKREMOVE); vd->vdev_forcefault = !!(flags & ZFS_ONLINE_FORCEFAULT); /* XXX - L2ARC 1.0 does not support expansion */ if (!vd->vdev_aux) { for (pvd = vd; pvd != rvd; pvd = pvd->vdev_parent) pvd->vdev_expanding = !!((flags & ZFS_ONLINE_EXPAND) || spa->spa_autoexpand); vd->vdev_expansion_time = gethrestime_sec(); } vdev_reopen(tvd); vd->vdev_checkremove = vd->vdev_forcefault = B_FALSE; if (!vd->vdev_aux) { for (pvd = vd; pvd != rvd; pvd = pvd->vdev_parent) pvd->vdev_expanding = B_FALSE; } if (newstate) *newstate = vd->vdev_state; if ((flags & ZFS_ONLINE_UNSPARE) && !vdev_is_dead(vd) && vd->vdev_parent && vd->vdev_parent->vdev_ops == &vdev_spare_ops && vd->vdev_parent->vdev_child[0] == vd) vd->vdev_unspare = B_TRUE; if ((flags & ZFS_ONLINE_EXPAND) || spa->spa_autoexpand) { /* XXX - L2ARC 1.0 does not support expansion */ if (vd->vdev_aux) return (spa_vdev_state_exit(spa, vd, ENOTSUP)); spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE); } /* Restart initializing if necessary */ mutex_enter(&vd->vdev_initialize_lock); if (vdev_writeable(vd) && vd->vdev_initialize_thread == NULL && vd->vdev_initialize_state == VDEV_INITIALIZE_ACTIVE) { (void) vdev_initialize(vd); } mutex_exit(&vd->vdev_initialize_lock); /* * Restart trimming if necessary. We do not restart trimming for cache * devices here. This is triggered by l2arc_rebuild_vdev() * asynchronously for the whole device or in l2arc_evict() as it evicts * space for upcoming writes. */ mutex_enter(&vd->vdev_trim_lock); if (vdev_writeable(vd) && !vd->vdev_isl2cache && vd->vdev_trim_thread == NULL && vd->vdev_trim_state == VDEV_TRIM_ACTIVE) { (void) vdev_trim(vd, vd->vdev_trim_rate, vd->vdev_trim_partial, vd->vdev_trim_secure); } mutex_exit(&vd->vdev_trim_lock); if (wasoffline || (oldstate < VDEV_STATE_DEGRADED && vd->vdev_state >= VDEV_STATE_DEGRADED)) spa_event_notify(spa, vd, NULL, ESC_ZFS_VDEV_ONLINE); return (spa_vdev_state_exit(spa, vd, 0)); } static int vdev_offline_locked(spa_t *spa, uint64_t guid, uint64_t flags) { vdev_t *vd, *tvd; int error = 0; uint64_t generation; metaslab_group_t *mg; top: spa_vdev_state_enter(spa, SCL_ALLOC); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENODEV))); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENOTSUP))); tvd = vd->vdev_top; mg = tvd->vdev_mg; generation = spa->spa_config_generation + 1; /* * If the device isn't already offline, try to offline it. */ if (!vd->vdev_offline) { /* * If this device has the only valid copy of some data, * don't allow it to be offlined. Log devices are always * expendable. */ if (!tvd->vdev_islog && vd->vdev_aux == NULL && vdev_dtl_required(vd)) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(EBUSY))); /* * If the top-level is a slog and it has had allocations * then proceed. We check that the vdev's metaslab group * is not NULL since it's possible that we may have just * added this vdev but not yet initialized its metaslabs. */ if (tvd->vdev_islog && mg != NULL) { /* * Prevent any future allocations. */ metaslab_group_passivate(mg); (void) spa_vdev_state_exit(spa, vd, 0); error = spa_reset_logs(spa); /* * If the log device was successfully reset but has * checkpointed data, do not offline it. */ if (error == 0 && tvd->vdev_checkpoint_sm != NULL) { ASSERT3U(space_map_allocated( tvd->vdev_checkpoint_sm), !=, 0); error = ZFS_ERR_CHECKPOINT_EXISTS; } spa_vdev_state_enter(spa, SCL_ALLOC); /* * Check to see if the config has changed. */ if (error || generation != spa->spa_config_generation) { metaslab_group_activate(mg); if (error) return (spa_vdev_state_exit(spa, vd, error)); (void) spa_vdev_state_exit(spa, vd, 0); goto top; } ASSERT0(tvd->vdev_stat.vs_alloc); } /* * Offline this device and reopen its top-level vdev. * If the top-level vdev is a log device then just offline * it. Otherwise, if this action results in the top-level * vdev becoming unusable, undo it and fail the request. */ vd->vdev_offline = B_TRUE; vdev_reopen(tvd); if (!tvd->vdev_islog && vd->vdev_aux == NULL && vdev_is_dead(tvd)) { vd->vdev_offline = B_FALSE; vdev_reopen(tvd); return (spa_vdev_state_exit(spa, NULL, SET_ERROR(EBUSY))); } /* * Add the device back into the metaslab rotor so that * once we online the device it's open for business. */ if (tvd->vdev_islog && mg != NULL) metaslab_group_activate(mg); } vd->vdev_tmpoffline = !!(flags & ZFS_OFFLINE_TEMPORARY); return (spa_vdev_state_exit(spa, vd, 0)); } int vdev_offline(spa_t *spa, uint64_t guid, uint64_t flags) { int error; mutex_enter(&spa->spa_vdev_top_lock); error = vdev_offline_locked(spa, guid, flags); mutex_exit(&spa->spa_vdev_top_lock); return (error); } /* * Clear the error counts associated with this vdev. Unlike vdev_online() and * vdev_offline(), we assume the spa config is locked. We also clear all * children. If 'vd' is NULL, then the user wants to clear all vdevs. */ void vdev_clear(spa_t *spa, vdev_t *vd) { vdev_t *rvd = spa->spa_root_vdev; ASSERT(spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); if (vd == NULL) vd = rvd; vd->vdev_stat.vs_read_errors = 0; vd->vdev_stat.vs_write_errors = 0; vd->vdev_stat.vs_checksum_errors = 0; vd->vdev_stat.vs_slow_ios = 0; for (int c = 0; c < vd->vdev_children; c++) vdev_clear(spa, vd->vdev_child[c]); /* * It makes no sense to "clear" an indirect vdev. */ if (!vdev_is_concrete(vd)) return; /* * If we're in the FAULTED state or have experienced failed I/O, then * clear the persistent state and attempt to reopen the device. We * also mark the vdev config dirty, so that the new faulted state is * written out to disk. */ if (vd->vdev_faulted || vd->vdev_degraded || !vdev_readable(vd) || !vdev_writeable(vd)) { /* * When reopening in response to a clear event, it may be due to * a fmadm repair request. In this case, if the device is * still broken, we want to still post the ereport again. */ vd->vdev_forcefault = B_TRUE; vd->vdev_faulted = vd->vdev_degraded = 0ULL; vd->vdev_cant_read = B_FALSE; vd->vdev_cant_write = B_FALSE; vd->vdev_stat.vs_aux = 0; vdev_reopen(vd == rvd ? rvd : vd->vdev_top); vd->vdev_forcefault = B_FALSE; if (vd != rvd && vdev_writeable(vd->vdev_top)) vdev_state_dirty(vd->vdev_top); /* If a resilver isn't required, check if vdevs can be culled */ if (vd->vdev_aux == NULL && !vdev_is_dead(vd) && !dsl_scan_resilvering(spa->spa_dsl_pool) && !dsl_scan_resilver_scheduled(spa->spa_dsl_pool)) spa_async_request(spa, SPA_ASYNC_RESILVER_DONE); spa_event_notify(spa, vd, NULL, ESC_ZFS_VDEV_CLEAR); } /* * When clearing a FMA-diagnosed fault, we always want to * unspare the device, as we assume that the original spare was * done in response to the FMA fault. */ if (!vdev_is_dead(vd) && vd->vdev_parent != NULL && vd->vdev_parent->vdev_ops == &vdev_spare_ops && vd->vdev_parent->vdev_child[0] == vd) vd->vdev_unspare = B_TRUE; } boolean_t vdev_is_dead(vdev_t *vd) { /* * Holes and missing devices are always considered "dead". * This simplifies the code since we don't have to check for * these types of devices in the various code paths. * Instead we rely on the fact that we skip over dead devices * before issuing I/O to them. */ return (vd->vdev_state < VDEV_STATE_DEGRADED || vd->vdev_ops == &vdev_hole_ops || vd->vdev_ops == &vdev_missing_ops); } boolean_t vdev_readable(vdev_t *vd) { return (!vdev_is_dead(vd) && !vd->vdev_cant_read); } boolean_t vdev_writeable(vdev_t *vd) { return (!vdev_is_dead(vd) && !vd->vdev_cant_write && vdev_is_concrete(vd)); } boolean_t vdev_allocatable(vdev_t *vd) { uint64_t state = vd->vdev_state; /* * We currently allow allocations from vdevs which may be in the * process of reopening (i.e. VDEV_STATE_CLOSED). If the device * fails to reopen then we'll catch it later when we're holding * the proper locks. Note that we have to get the vdev state * in a local variable because although it changes atomically, * we're asking two separate questions about it. */ return (!(state < VDEV_STATE_DEGRADED && state != VDEV_STATE_CLOSED) && !vd->vdev_cant_write && vdev_is_concrete(vd) && vd->vdev_mg->mg_initialized); } boolean_t vdev_accessible(vdev_t *vd, zio_t *zio) { ASSERT(zio->io_vd == vd); if (vdev_is_dead(vd) || vd->vdev_remove_wanted) return (B_FALSE); if (zio->io_type == ZIO_TYPE_READ) return (!vd->vdev_cant_read); if (zio->io_type == ZIO_TYPE_WRITE) return (!vd->vdev_cant_write); return (B_TRUE); } static void vdev_get_child_stat(vdev_t *cvd, vdev_stat_t *vs, vdev_stat_t *cvs) { for (int t = 0; t < VS_ZIO_TYPES; t++) { vs->vs_ops[t] += cvs->vs_ops[t]; vs->vs_bytes[t] += cvs->vs_bytes[t]; } cvs->vs_scan_removing = cvd->vdev_removing; } /* * Get extended stats */ static void vdev_get_child_stat_ex(vdev_t *cvd, vdev_stat_ex_t *vsx, vdev_stat_ex_t *cvsx) { int t, b; for (t = 0; t < ZIO_TYPES; t++) { for (b = 0; b < ARRAY_SIZE(vsx->vsx_disk_histo[0]); b++) vsx->vsx_disk_histo[t][b] += cvsx->vsx_disk_histo[t][b]; for (b = 0; b < ARRAY_SIZE(vsx->vsx_total_histo[0]); b++) { vsx->vsx_total_histo[t][b] += cvsx->vsx_total_histo[t][b]; } } for (t = 0; t < ZIO_PRIORITY_NUM_QUEUEABLE; t++) { for (b = 0; b < ARRAY_SIZE(vsx->vsx_queue_histo[0]); b++) { vsx->vsx_queue_histo[t][b] += cvsx->vsx_queue_histo[t][b]; } vsx->vsx_active_queue[t] += cvsx->vsx_active_queue[t]; vsx->vsx_pend_queue[t] += cvsx->vsx_pend_queue[t]; for (b = 0; b < ARRAY_SIZE(vsx->vsx_ind_histo[0]); b++) vsx->vsx_ind_histo[t][b] += cvsx->vsx_ind_histo[t][b]; for (b = 0; b < ARRAY_SIZE(vsx->vsx_agg_histo[0]); b++) vsx->vsx_agg_histo[t][b] += cvsx->vsx_agg_histo[t][b]; } } boolean_t vdev_is_spacemap_addressable(vdev_t *vd) { if (spa_feature_is_active(vd->vdev_spa, SPA_FEATURE_SPACEMAP_V2)) return (B_TRUE); /* * If double-word space map entries are not enabled we assume * 47 bits of the space map entry are dedicated to the entry's * offset (see SM_OFFSET_BITS in space_map.h). We then use that * to calculate the maximum address that can be described by a * space map entry for the given device. */ uint64_t shift = vd->vdev_ashift + SM_OFFSET_BITS; if (shift >= 63) /* detect potential overflow */ return (B_TRUE); return (vd->vdev_asize < (1ULL << shift)); } /* * Get statistics for the given vdev. */ static void vdev_get_stats_ex_impl(vdev_t *vd, vdev_stat_t *vs, vdev_stat_ex_t *vsx) { int t; /* * If we're getting stats on the root vdev, aggregate the I/O counts * over all top-level vdevs (i.e. the direct children of the root). */ if (!vd->vdev_ops->vdev_op_leaf) { if (vs) { memset(vs->vs_ops, 0, sizeof (vs->vs_ops)); memset(vs->vs_bytes, 0, sizeof (vs->vs_bytes)); } if (vsx) memset(vsx, 0, sizeof (*vsx)); for (int c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; vdev_stat_t *cvs = &cvd->vdev_stat; vdev_stat_ex_t *cvsx = &cvd->vdev_stat_ex; vdev_get_stats_ex_impl(cvd, cvs, cvsx); if (vs) vdev_get_child_stat(cvd, vs, cvs); if (vsx) vdev_get_child_stat_ex(cvd, vsx, cvsx); } } else { /* * We're a leaf. Just copy our ZIO active queue stats in. The * other leaf stats are updated in vdev_stat_update(). */ if (!vsx) return; memcpy(vsx, &vd->vdev_stat_ex, sizeof (vd->vdev_stat_ex)); for (t = 0; t < ARRAY_SIZE(vd->vdev_queue.vq_class); t++) { vsx->vsx_active_queue[t] = vd->vdev_queue.vq_class[t].vqc_active; vsx->vsx_pend_queue[t] = avl_numnodes( &vd->vdev_queue.vq_class[t].vqc_queued_tree); } } } void vdev_get_stats_ex(vdev_t *vd, vdev_stat_t *vs, vdev_stat_ex_t *vsx) { vdev_t *tvd = vd->vdev_top; mutex_enter(&vd->vdev_stat_lock); if (vs) { bcopy(&vd->vdev_stat, vs, sizeof (*vs)); vs->vs_timestamp = gethrtime() - vs->vs_timestamp; vs->vs_state = vd->vdev_state; vs->vs_rsize = vdev_get_min_asize(vd); if (vd->vdev_ops->vdev_op_leaf) { vs->vs_rsize += VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE; /* * Report initializing progress. Since we don't * have the initializing locks held, this is only * an estimate (although a fairly accurate one). */ vs->vs_initialize_bytes_done = vd->vdev_initialize_bytes_done; vs->vs_initialize_bytes_est = vd->vdev_initialize_bytes_est; vs->vs_initialize_state = vd->vdev_initialize_state; vs->vs_initialize_action_time = vd->vdev_initialize_action_time; /* * Report manual TRIM progress. Since we don't have * the manual TRIM locks held, this is only an * estimate (although fairly accurate one). */ vs->vs_trim_notsup = !vd->vdev_has_trim; vs->vs_trim_bytes_done = vd->vdev_trim_bytes_done; vs->vs_trim_bytes_est = vd->vdev_trim_bytes_est; vs->vs_trim_state = vd->vdev_trim_state; vs->vs_trim_action_time = vd->vdev_trim_action_time; /* Set when there is a deferred resilver. */ vs->vs_resilver_deferred = vd->vdev_resilver_deferred; } /* * Report expandable space on top-level, non-auxiliary devices * only. The expandable space is reported in terms of metaslab * sized units since that determines how much space the pool * can expand. */ if (vd->vdev_aux == NULL && tvd != NULL) { vs->vs_esize = P2ALIGN( vd->vdev_max_asize - vd->vdev_asize, 1ULL << tvd->vdev_ms_shift); } vs->vs_configured_ashift = vd->vdev_top != NULL ? vd->vdev_top->vdev_ashift : vd->vdev_ashift; vs->vs_logical_ashift = vd->vdev_logical_ashift; vs->vs_physical_ashift = vd->vdev_physical_ashift; /* * Report fragmentation and rebuild progress for top-level, * non-auxiliary, concrete devices. */ if (vd->vdev_aux == NULL && vd == vd->vdev_top && vdev_is_concrete(vd)) { vs->vs_fragmentation = (vd->vdev_mg != NULL) ? vd->vdev_mg->mg_fragmentation : 0; } } vdev_get_stats_ex_impl(vd, vs, vsx); mutex_exit(&vd->vdev_stat_lock); } void vdev_get_stats(vdev_t *vd, vdev_stat_t *vs) { return (vdev_get_stats_ex(vd, vs, NULL)); } void vdev_clear_stats(vdev_t *vd) { mutex_enter(&vd->vdev_stat_lock); vd->vdev_stat.vs_space = 0; vd->vdev_stat.vs_dspace = 0; vd->vdev_stat.vs_alloc = 0; mutex_exit(&vd->vdev_stat_lock); } void vdev_scan_stat_init(vdev_t *vd) { vdev_stat_t *vs = &vd->vdev_stat; for (int c = 0; c < vd->vdev_children; c++) vdev_scan_stat_init(vd->vdev_child[c]); mutex_enter(&vd->vdev_stat_lock); vs->vs_scan_processed = 0; mutex_exit(&vd->vdev_stat_lock); } void vdev_stat_update(zio_t *zio, uint64_t psize) { spa_t *spa = zio->io_spa; vdev_t *rvd = spa->spa_root_vdev; vdev_t *vd = zio->io_vd ? zio->io_vd : rvd; vdev_t *pvd; uint64_t txg = zio->io_txg; vdev_stat_t *vs = &vd->vdev_stat; vdev_stat_ex_t *vsx = &vd->vdev_stat_ex; zio_type_t type = zio->io_type; int flags = zio->io_flags; /* * If this i/o is a gang leader, it didn't do any actual work. */ if (zio->io_gang_tree) return; if (zio->io_error == 0) { /* * If this is a root i/o, don't count it -- we've already * counted the top-level vdevs, and vdev_get_stats() will * aggregate them when asked. This reduces contention on * the root vdev_stat_lock and implicitly handles blocks * that compress away to holes, for which there is no i/o. * (Holes never create vdev children, so all the counters * remain zero, which is what we want.) * * Note: this only applies to successful i/o (io_error == 0) * because unlike i/o counts, errors are not additive. * When reading a ditto block, for example, failure of * one top-level vdev does not imply a root-level error. */ if (vd == rvd) return; ASSERT(vd == zio->io_vd); if (flags & ZIO_FLAG_IO_BYPASS) return; mutex_enter(&vd->vdev_stat_lock); if (flags & ZIO_FLAG_IO_REPAIR) { /* * Repair is the result of a resilver issued by the * scan thread (spa_sync). */ if (flags & ZIO_FLAG_SCAN_THREAD) { dsl_scan_t *scn = spa->spa_dsl_pool->dp_scan; dsl_scan_phys_t *scn_phys = &scn->scn_phys; uint64_t *processed = &scn_phys->scn_processed; if (vd->vdev_ops->vdev_op_leaf) atomic_add_64(processed, psize); vs->vs_scan_processed += psize; } /* * Repair is the result of a rebuild issued by the * rebuild thread (vdev_rebuild_thread). */ if (zio->io_priority == ZIO_PRIORITY_REBUILD) { vdev_t *tvd = vd->vdev_top; vdev_rebuild_t *vr = &tvd->vdev_rebuild_config; vdev_rebuild_phys_t *vrp = &vr->vr_rebuild_phys; uint64_t *rebuilt = &vrp->vrp_bytes_rebuilt; if (vd->vdev_ops->vdev_op_leaf) atomic_add_64(rebuilt, psize); vs->vs_rebuild_processed += psize; } if (flags & ZIO_FLAG_SELF_HEAL) vs->vs_self_healed += psize; } /* * The bytes/ops/histograms are recorded at the leaf level and * aggregated into the higher level vdevs in vdev_get_stats(). */ if (vd->vdev_ops->vdev_op_leaf && (zio->io_priority < ZIO_PRIORITY_NUM_QUEUEABLE)) { zio_type_t vs_type = type; zio_priority_t priority = zio->io_priority; /* * TRIM ops and bytes are reported to user space as * ZIO_TYPE_IOCTL. This is done to preserve the * vdev_stat_t structure layout for user space. */ if (type == ZIO_TYPE_TRIM) vs_type = ZIO_TYPE_IOCTL; /* * Solely for the purposes of 'zpool iostat -lqrw' * reporting use the priority to catagorize the IO. * Only the following are reported to user space: * * ZIO_PRIORITY_SYNC_READ, * ZIO_PRIORITY_SYNC_WRITE, * ZIO_PRIORITY_ASYNC_READ, * ZIO_PRIORITY_ASYNC_WRITE, * ZIO_PRIORITY_SCRUB, * ZIO_PRIORITY_TRIM. */ if (priority == ZIO_PRIORITY_REBUILD) { priority = ((type == ZIO_TYPE_WRITE) ? ZIO_PRIORITY_ASYNC_WRITE : ZIO_PRIORITY_SCRUB); } else if (priority == ZIO_PRIORITY_INITIALIZING) { ASSERT3U(type, ==, ZIO_TYPE_WRITE); priority = ZIO_PRIORITY_ASYNC_WRITE; } else if (priority == ZIO_PRIORITY_REMOVAL) { priority = ((type == ZIO_TYPE_WRITE) ? ZIO_PRIORITY_ASYNC_WRITE : ZIO_PRIORITY_ASYNC_READ); } vs->vs_ops[vs_type]++; vs->vs_bytes[vs_type] += psize; if (flags & ZIO_FLAG_DELEGATED) { vsx->vsx_agg_histo[priority] [RQ_HISTO(zio->io_size)]++; } else { vsx->vsx_ind_histo[priority] [RQ_HISTO(zio->io_size)]++; } if (zio->io_delta && zio->io_delay) { vsx->vsx_queue_histo[priority] [L_HISTO(zio->io_delta - zio->io_delay)]++; vsx->vsx_disk_histo[type] [L_HISTO(zio->io_delay)]++; vsx->vsx_total_histo[type] [L_HISTO(zio->io_delta)]++; } } mutex_exit(&vd->vdev_stat_lock); return; } if (flags & ZIO_FLAG_SPECULATIVE) return; /* * If this is an I/O error that is going to be retried, then ignore the * error. Otherwise, the user may interpret B_FAILFAST I/O errors as * hard errors, when in reality they can happen for any number of * innocuous reasons (bus resets, MPxIO link failure, etc). */ if (zio->io_error == EIO && !(zio->io_flags & ZIO_FLAG_IO_RETRY)) return; /* * Intent logs writes won't propagate their error to the root * I/O so don't mark these types of failures as pool-level * errors. */ if (zio->io_vd == NULL && (zio->io_flags & ZIO_FLAG_DONT_PROPAGATE)) return; if (spa->spa_load_state == SPA_LOAD_NONE && type == ZIO_TYPE_WRITE && txg != 0 && (!(flags & ZIO_FLAG_IO_REPAIR) || (flags & ZIO_FLAG_SCAN_THREAD) || spa->spa_claiming)) { /* * This is either a normal write (not a repair), or it's * a repair induced by the scrub thread, or it's a repair * made by zil_claim() during spa_load() in the first txg. * In the normal case, we commit the DTL change in the same * txg as the block was born. In the scrub-induced repair * case, we know that scrubs run in first-pass syncing context, * so we commit the DTL change in spa_syncing_txg(spa). * In the zil_claim() case, we commit in spa_first_txg(spa). * * We currently do not make DTL entries for failed spontaneous * self-healing writes triggered by normal (non-scrubbing) * reads, because we have no transactional context in which to * do so -- and it's not clear that it'd be desirable anyway. */ if (vd->vdev_ops->vdev_op_leaf) { uint64_t commit_txg = txg; if (flags & ZIO_FLAG_SCAN_THREAD) { ASSERT(flags & ZIO_FLAG_IO_REPAIR); ASSERT(spa_sync_pass(spa) == 1); vdev_dtl_dirty(vd, DTL_SCRUB, txg, 1); commit_txg = spa_syncing_txg(spa); } else if (spa->spa_claiming) { ASSERT(flags & ZIO_FLAG_IO_REPAIR); commit_txg = spa_first_txg(spa); } ASSERT(commit_txg >= spa_syncing_txg(spa)); if (vdev_dtl_contains(vd, DTL_MISSING, txg, 1)) return; for (pvd = vd; pvd != rvd; pvd = pvd->vdev_parent) vdev_dtl_dirty(pvd, DTL_PARTIAL, txg, 1); vdev_dirty(vd->vdev_top, VDD_DTL, vd, commit_txg); } if (vd != rvd) vdev_dtl_dirty(vd, DTL_MISSING, txg, 1); } } int64_t vdev_deflated_space(vdev_t *vd, int64_t space) { ASSERT((space & (SPA_MINBLOCKSIZE-1)) == 0); ASSERT(vd->vdev_deflate_ratio != 0 || vd->vdev_isl2cache); return ((space >> SPA_MINBLOCKSHIFT) * vd->vdev_deflate_ratio); } /* * Update the in-core space usage stats for this vdev, its metaslab class, * and the root vdev. */ void vdev_space_update(vdev_t *vd, int64_t alloc_delta, int64_t defer_delta, int64_t space_delta) { int64_t dspace_delta; spa_t *spa = vd->vdev_spa; vdev_t *rvd = spa->spa_root_vdev; ASSERT(vd == vd->vdev_top); /* * Apply the inverse of the psize-to-asize (ie. RAID-Z) space-expansion * factor. We must calculate this here and not at the root vdev * because the root vdev's psize-to-asize is simply the max of its * children's, thus not accurate enough for us. */ dspace_delta = vdev_deflated_space(vd, space_delta); mutex_enter(&vd->vdev_stat_lock); /* ensure we won't underflow */ if (alloc_delta < 0) { ASSERT3U(vd->vdev_stat.vs_alloc, >=, -alloc_delta); } vd->vdev_stat.vs_alloc += alloc_delta; vd->vdev_stat.vs_space += space_delta; vd->vdev_stat.vs_dspace += dspace_delta; mutex_exit(&vd->vdev_stat_lock); /* every class but log contributes to root space stats */ if (vd->vdev_mg != NULL && !vd->vdev_islog) { ASSERT(!vd->vdev_isl2cache); mutex_enter(&rvd->vdev_stat_lock); rvd->vdev_stat.vs_alloc += alloc_delta; rvd->vdev_stat.vs_space += space_delta; rvd->vdev_stat.vs_dspace += dspace_delta; mutex_exit(&rvd->vdev_stat_lock); } /* Note: metaslab_class_space_update moved to metaslab_space_update */ } /* * Mark a top-level vdev's config as dirty, placing it on the dirty list * so that it will be written out next time the vdev configuration is synced. * If the root vdev is specified (vdev_top == NULL), dirty all top-level vdevs. */ void vdev_config_dirty(vdev_t *vd) { spa_t *spa = vd->vdev_spa; vdev_t *rvd = spa->spa_root_vdev; int c; ASSERT(spa_writeable(spa)); /* * If this is an aux vdev (as with l2cache and spare devices), then we * update the vdev config manually and set the sync flag. */ if (vd->vdev_aux != NULL) { spa_aux_vdev_t *sav = vd->vdev_aux; nvlist_t **aux; uint_t naux; for (c = 0; c < sav->sav_count; c++) { if (sav->sav_vdevs[c] == vd) break; } if (c == sav->sav_count) { /* * We're being removed. There's nothing more to do. */ ASSERT(sav->sav_sync == B_TRUE); return; } sav->sav_sync = B_TRUE; if (nvlist_lookup_nvlist_array(sav->sav_config, ZPOOL_CONFIG_L2CACHE, &aux, &naux) != 0) { VERIFY(nvlist_lookup_nvlist_array(sav->sav_config, ZPOOL_CONFIG_SPARES, &aux, &naux) == 0); } ASSERT(c < naux); /* * Setting the nvlist in the middle if the array is a little * sketchy, but it will work. */ nvlist_free(aux[c]); aux[c] = vdev_config_generate(spa, vd, B_TRUE, 0); return; } /* * The dirty list is protected by the SCL_CONFIG lock. The caller * must either hold SCL_CONFIG as writer, or must be the sync thread * (which holds SCL_CONFIG as reader). There's only one sync thread, * so this is sufficient to ensure mutual exclusion. */ ASSERT(spa_config_held(spa, SCL_CONFIG, RW_WRITER) || (dsl_pool_sync_context(spa_get_dsl(spa)) && spa_config_held(spa, SCL_CONFIG, RW_READER))); if (vd == rvd) { for (c = 0; c < rvd->vdev_children; c++) vdev_config_dirty(rvd->vdev_child[c]); } else { ASSERT(vd == vd->vdev_top); if (!list_link_active(&vd->vdev_config_dirty_node) && vdev_is_concrete(vd)) { list_insert_head(&spa->spa_config_dirty_list, vd); } } } void vdev_config_clean(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT(spa_config_held(spa, SCL_CONFIG, RW_WRITER) || (dsl_pool_sync_context(spa_get_dsl(spa)) && spa_config_held(spa, SCL_CONFIG, RW_READER))); ASSERT(list_link_active(&vd->vdev_config_dirty_node)); list_remove(&spa->spa_config_dirty_list, vd); } /* * Mark a top-level vdev's state as dirty, so that the next pass of * spa_sync() can convert this into vdev_config_dirty(). We distinguish * the state changes from larger config changes because they require * much less locking, and are often needed for administrative actions. */ void vdev_state_dirty(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT(spa_writeable(spa)); ASSERT(vd == vd->vdev_top); /* * The state list is protected by the SCL_STATE lock. The caller * must either hold SCL_STATE as writer, or must be the sync thread * (which holds SCL_STATE as reader). There's only one sync thread, * so this is sufficient to ensure mutual exclusion. */ ASSERT(spa_config_held(spa, SCL_STATE, RW_WRITER) || (dsl_pool_sync_context(spa_get_dsl(spa)) && spa_config_held(spa, SCL_STATE, RW_READER))); if (!list_link_active(&vd->vdev_state_dirty_node) && vdev_is_concrete(vd)) list_insert_head(&spa->spa_state_dirty_list, vd); } void vdev_state_clean(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT(spa_config_held(spa, SCL_STATE, RW_WRITER) || (dsl_pool_sync_context(spa_get_dsl(spa)) && spa_config_held(spa, SCL_STATE, RW_READER))); ASSERT(list_link_active(&vd->vdev_state_dirty_node)); list_remove(&spa->spa_state_dirty_list, vd); } /* * Propagate vdev state up from children to parent. */ void vdev_propagate_state(vdev_t *vd) { spa_t *spa = vd->vdev_spa; vdev_t *rvd = spa->spa_root_vdev; int degraded = 0, faulted = 0; int corrupted = 0; vdev_t *child; if (vd->vdev_children > 0) { for (int c = 0; c < vd->vdev_children; c++) { child = vd->vdev_child[c]; /* * Don't factor holes or indirect vdevs into the * decision. */ if (!vdev_is_concrete(child)) continue; if (!vdev_readable(child) || (!vdev_writeable(child) && spa_writeable(spa))) { /* * Root special: if there is a top-level log * device, treat the root vdev as if it were * degraded. */ if (child->vdev_islog && vd == rvd) degraded++; else faulted++; } else if (child->vdev_state <= VDEV_STATE_DEGRADED) { degraded++; } if (child->vdev_stat.vs_aux == VDEV_AUX_CORRUPT_DATA) corrupted++; } vd->vdev_ops->vdev_op_state_change(vd, faulted, degraded); /* * Root special: if there is a top-level vdev that cannot be * opened due to corrupted metadata, then propagate the root * vdev's aux state as 'corrupt' rather than 'insufficient * replicas'. */ if (corrupted && vd == rvd && rvd->vdev_state == VDEV_STATE_CANT_OPEN) vdev_set_state(rvd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); } if (vd->vdev_parent) vdev_propagate_state(vd->vdev_parent); } /* * Set a vdev's state. If this is during an open, we don't update the parent * state, because we're in the process of opening children depth-first. * Otherwise, we propagate the change to the parent. * * If this routine places a device in a faulted state, an appropriate ereport is * generated. */ void vdev_set_state(vdev_t *vd, boolean_t isopen, vdev_state_t state, vdev_aux_t aux) { uint64_t save_state; spa_t *spa = vd->vdev_spa; if (state == vd->vdev_state) { /* * Since vdev_offline() code path is already in an offline * state we can miss a statechange event to OFFLINE. Check * the previous state to catch this condition. */ if (vd->vdev_ops->vdev_op_leaf && (state == VDEV_STATE_OFFLINE) && (vd->vdev_prevstate >= VDEV_STATE_FAULTED)) { /* post an offline state change */ zfs_post_state_change(spa, vd, vd->vdev_prevstate); } vd->vdev_stat.vs_aux = aux; return; } save_state = vd->vdev_state; vd->vdev_state = state; vd->vdev_stat.vs_aux = aux; /* * If we are setting the vdev state to anything but an open state, then * always close the underlying device unless the device has requested * a delayed close (i.e. we're about to remove or fault the device). * Otherwise, we keep accessible but invalid devices open forever. * We don't call vdev_close() itself, because that implies some extra * checks (offline, etc) that we don't want here. This is limited to * leaf devices, because otherwise closing the device will affect other * children. */ if (!vd->vdev_delayed_close && vdev_is_dead(vd) && vd->vdev_ops->vdev_op_leaf) vd->vdev_ops->vdev_op_close(vd); if (vd->vdev_removed && state == VDEV_STATE_CANT_OPEN && (aux == VDEV_AUX_OPEN_FAILED || vd->vdev_checkremove)) { /* * If the previous state is set to VDEV_STATE_REMOVED, then this * device was previously marked removed and someone attempted to * reopen it. If this failed due to a nonexistent device, then * keep the device in the REMOVED state. We also let this be if * it is one of our special test online cases, which is only * attempting to online the device and shouldn't generate an FMA * fault. */ vd->vdev_state = VDEV_STATE_REMOVED; vd->vdev_stat.vs_aux = VDEV_AUX_NONE; } else if (state == VDEV_STATE_REMOVED) { vd->vdev_removed = B_TRUE; } else if (state == VDEV_STATE_CANT_OPEN) { /* * If we fail to open a vdev during an import or recovery, we * mark it as "not available", which signifies that it was * never there to begin with. Failure to open such a device * is not considered an error. */ if ((spa_load_state(spa) == SPA_LOAD_IMPORT || spa_load_state(spa) == SPA_LOAD_RECOVER) && vd->vdev_ops->vdev_op_leaf) vd->vdev_not_present = 1; /* * Post the appropriate ereport. If the 'prevstate' field is * set to something other than VDEV_STATE_UNKNOWN, it indicates * that this is part of a vdev_reopen(). In this case, we don't * want to post the ereport if the device was already in the * CANT_OPEN state beforehand. * * If the 'checkremove' flag is set, then this is an attempt to * online the device in response to an insertion event. If we * hit this case, then we have detected an insertion event for a * faulted or offline device that wasn't in the removed state. * In this scenario, we don't post an ereport because we are * about to replace the device, or attempt an online with * vdev_forcefault, which will generate the fault for us. */ if ((vd->vdev_prevstate != state || vd->vdev_forcefault) && !vd->vdev_not_present && !vd->vdev_checkremove && vd != spa->spa_root_vdev) { const char *class; switch (aux) { case VDEV_AUX_OPEN_FAILED: class = FM_EREPORT_ZFS_DEVICE_OPEN_FAILED; break; case VDEV_AUX_CORRUPT_DATA: class = FM_EREPORT_ZFS_DEVICE_CORRUPT_DATA; break; case VDEV_AUX_NO_REPLICAS: class = FM_EREPORT_ZFS_DEVICE_NO_REPLICAS; break; case VDEV_AUX_BAD_GUID_SUM: class = FM_EREPORT_ZFS_DEVICE_BAD_GUID_SUM; break; case VDEV_AUX_TOO_SMALL: class = FM_EREPORT_ZFS_DEVICE_TOO_SMALL; break; case VDEV_AUX_BAD_LABEL: class = FM_EREPORT_ZFS_DEVICE_BAD_LABEL; break; case VDEV_AUX_BAD_ASHIFT: class = FM_EREPORT_ZFS_DEVICE_BAD_ASHIFT; break; default: class = FM_EREPORT_ZFS_DEVICE_UNKNOWN; } (void) zfs_ereport_post(class, spa, vd, NULL, NULL, save_state); } /* Erase any notion of persistent removed state */ vd->vdev_removed = B_FALSE; } else { vd->vdev_removed = B_FALSE; } /* * Notify ZED of any significant state-change on a leaf vdev. * */ if (vd->vdev_ops->vdev_op_leaf) { /* preserve original state from a vdev_reopen() */ if ((vd->vdev_prevstate != VDEV_STATE_UNKNOWN) && (vd->vdev_prevstate != vd->vdev_state) && (save_state <= VDEV_STATE_CLOSED)) save_state = vd->vdev_prevstate; /* filter out state change due to initial vdev_open */ if (save_state > VDEV_STATE_CLOSED) zfs_post_state_change(spa, vd, save_state); } if (!isopen && vd->vdev_parent) vdev_propagate_state(vd->vdev_parent); } boolean_t vdev_children_are_offline(vdev_t *vd) { ASSERT(!vd->vdev_ops->vdev_op_leaf); for (uint64_t i = 0; i < vd->vdev_children; i++) { if (vd->vdev_child[i]->vdev_state != VDEV_STATE_OFFLINE) return (B_FALSE); } return (B_TRUE); } /* * Check the vdev configuration to ensure that it's capable of supporting * a root pool. We do not support partial configuration. */ boolean_t vdev_is_bootable(vdev_t *vd) { if (!vd->vdev_ops->vdev_op_leaf) { const char *vdev_type = vd->vdev_ops->vdev_op_type; if (strcmp(vdev_type, VDEV_TYPE_MISSING) == 0 || strcmp(vdev_type, VDEV_TYPE_INDIRECT) == 0) { return (B_FALSE); } } for (int c = 0; c < vd->vdev_children; c++) { if (!vdev_is_bootable(vd->vdev_child[c])) return (B_FALSE); } return (B_TRUE); } boolean_t vdev_is_concrete(vdev_t *vd) { vdev_ops_t *ops = vd->vdev_ops; if (ops == &vdev_indirect_ops || ops == &vdev_hole_ops || ops == &vdev_missing_ops || ops == &vdev_root_ops) { return (B_FALSE); } else { return (B_TRUE); } } /* * Determine if a log device has valid content. If the vdev was * removed or faulted in the MOS config then we know that * the content on the log device has already been written to the pool. */ boolean_t vdev_log_state_valid(vdev_t *vd) { if (vd->vdev_ops->vdev_op_leaf && !vd->vdev_faulted && !vd->vdev_removed) return (B_TRUE); for (int c = 0; c < vd->vdev_children; c++) if (vdev_log_state_valid(vd->vdev_child[c])) return (B_TRUE); return (B_FALSE); } /* * Expand a vdev if possible. */ void vdev_expand(vdev_t *vd, uint64_t txg) { ASSERT(vd->vdev_top == vd); ASSERT(spa_config_held(vd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL); ASSERT(vdev_is_concrete(vd)); vdev_set_deflate_ratio(vd); if ((vd->vdev_asize >> vd->vdev_ms_shift) > vd->vdev_ms_count && vdev_is_concrete(vd)) { vdev_metaslab_group_create(vd); VERIFY(vdev_metaslab_init(vd, txg) == 0); vdev_config_dirty(vd); } } /* * Split a vdev. */ void vdev_split(vdev_t *vd) { vdev_t *cvd, *pvd = vd->vdev_parent; vdev_remove_child(pvd, vd); vdev_compact_children(pvd); cvd = pvd->vdev_child[0]; if (pvd->vdev_children == 1) { vdev_remove_parent(cvd); cvd->vdev_splitting = B_TRUE; } vdev_propagate_state(cvd); } void vdev_deadman(vdev_t *vd, char *tag) { for (int c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; vdev_deadman(cvd, tag); } if (vd->vdev_ops->vdev_op_leaf) { vdev_queue_t *vq = &vd->vdev_queue; mutex_enter(&vq->vq_lock); if (avl_numnodes(&vq->vq_active_tree) > 0) { spa_t *spa = vd->vdev_spa; zio_t *fio; uint64_t delta; zfs_dbgmsg("slow vdev: %s has %d active IOs", vd->vdev_path, avl_numnodes(&vq->vq_active_tree)); /* * Look at the head of all the pending queues, * if any I/O has been outstanding for longer than * the spa_deadman_synctime invoke the deadman logic. */ fio = avl_first(&vq->vq_active_tree); delta = gethrtime() - fio->io_timestamp; if (delta > spa_deadman_synctime(spa)) zio_deadman(fio, tag); } mutex_exit(&vq->vq_lock); } } void vdev_defer_resilver(vdev_t *vd) { ASSERT(vd->vdev_ops->vdev_op_leaf); vd->vdev_resilver_deferred = B_TRUE; vd->vdev_spa->spa_resilver_deferred = B_TRUE; } /* * Clears the resilver deferred flag on all leaf devs under vd. Returns * B_TRUE if we have devices that need to be resilvered and are available to * accept resilver I/Os. */ boolean_t vdev_clear_resilver_deferred(vdev_t *vd, dmu_tx_t *tx) { boolean_t resilver_needed = B_FALSE; spa_t *spa = vd->vdev_spa; for (int c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; resilver_needed |= vdev_clear_resilver_deferred(cvd, tx); } if (vd == spa->spa_root_vdev && spa_feature_is_active(spa, SPA_FEATURE_RESILVER_DEFER)) { spa_feature_decr(spa, SPA_FEATURE_RESILVER_DEFER, tx); vdev_config_dirty(vd); spa->spa_resilver_deferred = B_FALSE; return (resilver_needed); } if (!vdev_is_concrete(vd) || vd->vdev_aux || !vd->vdev_ops->vdev_op_leaf) return (resilver_needed); vd->vdev_resilver_deferred = B_FALSE; return (!vdev_is_dead(vd) && !vd->vdev_offline && vdev_resilver_needed(vd, NULL, NULL)); } /* * Translate a logical range to the physical range for the specified vdev_t. * This function is initially called with a leaf vdev and will walk each * parent vdev until it reaches a top-level vdev. Once the top-level is * reached the physical range is initialized and the recursive function * begins to unwind. As it unwinds it calls the parent's vdev specific * translation function to do the real conversion. */ void vdev_xlate(vdev_t *vd, const range_seg64_t *logical_rs, range_seg64_t *physical_rs) { /* * Walk up the vdev tree */ if (vd != vd->vdev_top) { vdev_xlate(vd->vdev_parent, logical_rs, physical_rs); } else { /* * We've reached the top-level vdev, initialize the * physical range to the logical range and start to * unwind. */ physical_rs->rs_start = logical_rs->rs_start; physical_rs->rs_end = logical_rs->rs_end; return; } vdev_t *pvd = vd->vdev_parent; ASSERT3P(pvd, !=, NULL); ASSERT3P(pvd->vdev_ops->vdev_op_xlate, !=, NULL); /* * As this recursive function unwinds, translate the logical * range into its physical components by calling the * vdev specific translate function. */ range_seg64_t intermediate = { 0 }; pvd->vdev_ops->vdev_op_xlate(vd, physical_rs, &intermediate); physical_rs->rs_start = intermediate.rs_start; physical_rs->rs_end = intermediate.rs_end; } /* * Look at the vdev tree and determine whether any devices are currently being * replaced. */ boolean_t vdev_replace_in_progress(vdev_t *vdev) { ASSERT(spa_config_held(vdev->vdev_spa, SCL_ALL, RW_READER) != 0); if (vdev->vdev_ops == &vdev_replacing_ops) return (B_TRUE); /* * A 'spare' vdev indicates that we have a replace in progress, unless * it has exactly two children, and the second, the hot spare, has * finished being resilvered. */ if (vdev->vdev_ops == &vdev_spare_ops && (vdev->vdev_children > 2 || !vdev_dtl_empty(vdev->vdev_child[1], DTL_MISSING))) return (B_TRUE); for (int i = 0; i < vdev->vdev_children; i++) { if (vdev_replace_in_progress(vdev->vdev_child[i])) return (B_TRUE); } return (B_FALSE); } EXPORT_SYMBOL(vdev_fault); EXPORT_SYMBOL(vdev_degrade); EXPORT_SYMBOL(vdev_online); EXPORT_SYMBOL(vdev_offline); EXPORT_SYMBOL(vdev_clear); /* BEGIN CSTYLED */ ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, default_ms_count, INT, ZMOD_RW, "Target number of metaslabs per top-level vdev"); ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, default_ms_shift, INT, ZMOD_RW, "Default limit for metaslab size"); ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, min_ms_count, INT, ZMOD_RW, "Minimum number of metaslabs per top-level vdev"); ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, ms_count_limit, INT, ZMOD_RW, "Practical upper limit of total metaslabs per top-level vdev"); ZFS_MODULE_PARAM(zfs, zfs_, slow_io_events_per_second, UINT, ZMOD_RW, "Rate limit slow IO (delay) events to this many per second"); ZFS_MODULE_PARAM(zfs, zfs_, checksum_events_per_second, UINT, ZMOD_RW, "Rate limit checksum events to this many checksum errors per second " "(do not set below zed threshold)."); ZFS_MODULE_PARAM(zfs, zfs_, scan_ignore_errors, INT, ZMOD_RW, "Ignore errors during resilver/scrub"); ZFS_MODULE_PARAM(zfs_vdev, vdev_, validate_skip, INT, ZMOD_RW, "Bypass vdev_validate()"); ZFS_MODULE_PARAM(zfs, zfs_, nocacheflush, INT, ZMOD_RW, "Disable cache flushes"); ZFS_MODULE_PARAM_CALL(zfs_vdev, zfs_vdev_, min_auto_ashift, param_set_min_auto_ashift, param_get_ulong, ZMOD_RW, "Minimum ashift used when creating new top-level vdevs"); ZFS_MODULE_PARAM_CALL(zfs_vdev, zfs_vdev_, max_auto_ashift, param_set_max_auto_ashift, param_get_ulong, ZMOD_RW, "Maximum ashift used when optimizing for logical -> physical sector " "size on new top-level vdevs"); /* END CSTYLED */ diff --git a/module/zfs/zio.c b/module/zfs/zio.c index 8a8fbccd7d63..260e88b0be2c 100644 --- a/module/zfs/zio.c +++ b/module/zfs/zio.c @@ -1,4971 +1,4971 @@ /* * 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) 2011, 2020 by Delphix. All rights reserved. * Copyright (c) 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2019, Klara Inc. * Copyright (c) 2019, Allan Jude */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * ========================================================================== * I/O type descriptions * ========================================================================== */ const char *zio_type_name[ZIO_TYPES] = { /* * Note: Linux kernel thread name length is limited * so these names will differ from upstream open zfs. */ "z_null", "z_rd", "z_wr", "z_fr", "z_cl", "z_ioctl", "z_trim" }; int zio_dva_throttle_enabled = B_TRUE; int zio_deadman_log_all = B_FALSE; /* * ========================================================================== * I/O kmem caches * ========================================================================== */ kmem_cache_t *zio_cache; kmem_cache_t *zio_link_cache; kmem_cache_t *zio_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; kmem_cache_t *zio_data_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; #if defined(ZFS_DEBUG) && !defined(_KERNEL) uint64_t zio_buf_cache_allocs[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; uint64_t zio_buf_cache_frees[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; #endif /* Mark IOs as "slow" if they take longer than 30 seconds */ int zio_slow_io_ms = (30 * MILLISEC); #define BP_SPANB(indblkshift, level) \ (((uint64_t)1) << ((level) * ((indblkshift) - SPA_BLKPTRSHIFT))) #define COMPARE_META_LEVEL 0x80000000ul /* * The following actions directly effect the spa's sync-to-convergence logic. * The values below define the sync pass when we start performing the action. * Care should be taken when changing these values as they directly impact * spa_sync() performance. Tuning these values may introduce subtle performance * pathologies and should only be done in the context of performance analysis. * These tunables will eventually be removed and replaced with #defines once * enough analysis has been done to determine optimal values. * * The 'zfs_sync_pass_deferred_free' pass must be greater than 1 to ensure that * regular blocks are not deferred. * * Starting in sync pass 8 (zfs_sync_pass_dont_compress), we disable * compression (including of metadata). In practice, we don't have this * many sync passes, so this has no effect. * * The original intent was that disabling compression would help the sync * passes to converge. However, in practice disabling compression increases * the average number of sync passes, because when we turn compression off, a * lot of block's size will change and thus we have to re-allocate (not * overwrite) them. It also increases the number of 128KB allocations (e.g. * for indirect blocks and spacemaps) because these will not be compressed. * The 128K allocations are especially detrimental to performance on highly * fragmented systems, which may have very few free segments of this size, * and may need to load new metaslabs to satisfy 128K allocations. */ int zfs_sync_pass_deferred_free = 2; /* defer frees starting in this pass */ int zfs_sync_pass_dont_compress = 8; /* don't compress starting in this pass */ int zfs_sync_pass_rewrite = 2; /* rewrite new bps starting in this pass */ /* * An allocating zio is one that either currently has the DVA allocate * stage set or will have it later in its lifetime. */ #define IO_IS_ALLOCATING(zio) ((zio)->io_orig_pipeline & ZIO_STAGE_DVA_ALLOCATE) /* * Enable smaller cores by excluding metadata * allocations as well. */ int zio_exclude_metadata = 0; int zio_requeue_io_start_cut_in_line = 1; #ifdef ZFS_DEBUG int zio_buf_debug_limit = 16384; #else int zio_buf_debug_limit = 0; #endif static inline void __zio_execute(zio_t *zio); static void zio_taskq_dispatch(zio_t *, zio_taskq_type_t, boolean_t); void zio_init(void) { size_t c; vmem_t *data_alloc_arena = NULL; zio_cache = kmem_cache_create("zio_cache", sizeof (zio_t), 0, NULL, NULL, NULL, NULL, NULL, 0); zio_link_cache = kmem_cache_create("zio_link_cache", sizeof (zio_link_t), 0, NULL, NULL, NULL, NULL, NULL, 0); /* * For small buffers, we want a cache for each multiple of * SPA_MINBLOCKSIZE. For larger buffers, we want a cache * for each quarter-power of 2. */ for (c = 0; c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; c++) { size_t size = (c + 1) << SPA_MINBLOCKSHIFT; size_t p2 = size; size_t align = 0; size_t data_cflags, cflags; data_cflags = KMC_NODEBUG; cflags = (zio_exclude_metadata || size > zio_buf_debug_limit) ? KMC_NODEBUG : 0; #if defined(_ILP32) && defined(_KERNEL) /* * Cache size limited to 1M on 32-bit platforms until ARC * buffers no longer require virtual address space. */ if (size > zfs_max_recordsize) break; #endif while (!ISP2(p2)) p2 &= p2 - 1; #ifndef _KERNEL /* * If we are using watchpoints, put each buffer on its own page, * to eliminate the performance overhead of trapping to the * kernel when modifying a non-watched buffer that shares the * page with a watched buffer. */ if (arc_watch && !IS_P2ALIGNED(size, PAGESIZE)) continue; /* * Here's the problem - on 4K native devices in userland on * Linux using O_DIRECT, buffers must be 4K aligned or I/O * will fail with EINVAL, causing zdb (and others) to coredump. * Since userland probably doesn't need optimized buffer caches, * we just force 4K alignment on everything. */ align = 8 * SPA_MINBLOCKSIZE; #else if (size < PAGESIZE) { align = SPA_MINBLOCKSIZE; } else if (IS_P2ALIGNED(size, p2 >> 2)) { align = PAGESIZE; } #endif if (align != 0) { char name[36]; (void) snprintf(name, sizeof (name), "zio_buf_%lu", (ulong_t)size); zio_buf_cache[c] = kmem_cache_create(name, size, align, NULL, NULL, NULL, NULL, NULL, cflags); (void) snprintf(name, sizeof (name), "zio_data_buf_%lu", (ulong_t)size); zio_data_buf_cache[c] = kmem_cache_create(name, size, align, NULL, NULL, NULL, NULL, data_alloc_arena, data_cflags); } } while (--c != 0) { ASSERT(zio_buf_cache[c] != NULL); if (zio_buf_cache[c - 1] == NULL) zio_buf_cache[c - 1] = zio_buf_cache[c]; ASSERT(zio_data_buf_cache[c] != NULL); if (zio_data_buf_cache[c - 1] == NULL) zio_data_buf_cache[c - 1] = zio_data_buf_cache[c]; } zio_inject_init(); lz4_init(); } void zio_fini(void) { size_t c; kmem_cache_t *last_cache = NULL; kmem_cache_t *last_data_cache = NULL; for (c = 0; c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; c++) { #ifdef _ILP32 /* * Cache size limited to 1M on 32-bit platforms until ARC * buffers no longer require virtual address space. */ if (((c + 1) << SPA_MINBLOCKSHIFT) > zfs_max_recordsize) break; #endif #if defined(ZFS_DEBUG) && !defined(_KERNEL) if (zio_buf_cache_allocs[c] != zio_buf_cache_frees[c]) (void) printf("zio_fini: [%d] %llu != %llu\n", (int)((c + 1) << SPA_MINBLOCKSHIFT), (long long unsigned)zio_buf_cache_allocs[c], (long long unsigned)zio_buf_cache_frees[c]); #endif if (zio_buf_cache[c] != last_cache) { last_cache = zio_buf_cache[c]; kmem_cache_destroy(zio_buf_cache[c]); } zio_buf_cache[c] = NULL; if (zio_data_buf_cache[c] != last_data_cache) { last_data_cache = zio_data_buf_cache[c]; kmem_cache_destroy(zio_data_buf_cache[c]); } zio_data_buf_cache[c] = NULL; } kmem_cache_destroy(zio_link_cache); kmem_cache_destroy(zio_cache); zio_inject_fini(); lz4_fini(); } /* * ========================================================================== * Allocate and free I/O buffers * ========================================================================== */ /* * Use zio_buf_alloc to allocate ZFS metadata. This data will appear in a * crashdump if the kernel panics, so use it judiciously. Obviously, it's * useful to inspect ZFS metadata, but if possible, we should avoid keeping * excess / transient data in-core during a crashdump. */ void * zio_buf_alloc(size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); #if defined(ZFS_DEBUG) && !defined(_KERNEL) atomic_add_64(&zio_buf_cache_allocs[c], 1); #endif return (kmem_cache_alloc(zio_buf_cache[c], KM_PUSHPAGE)); } /* * Use zio_data_buf_alloc to allocate data. The data will not appear in a * crashdump if the kernel panics. This exists so that we will limit the amount * of ZFS data that shows up in a kernel crashdump. (Thus reducing the amount * of kernel heap dumped to disk when the kernel panics) */ void * zio_data_buf_alloc(size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); return (kmem_cache_alloc(zio_data_buf_cache[c], KM_PUSHPAGE)); } void zio_buf_free(void *buf, size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); #if defined(ZFS_DEBUG) && !defined(_KERNEL) atomic_add_64(&zio_buf_cache_frees[c], 1); #endif kmem_cache_free(zio_buf_cache[c], buf); } void zio_data_buf_free(void *buf, size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); kmem_cache_free(zio_data_buf_cache[c], buf); } static void zio_abd_free(void *abd, size_t size) { abd_free((abd_t *)abd); } /* * ========================================================================== * Push and pop I/O transform buffers * ========================================================================== */ void zio_push_transform(zio_t *zio, abd_t *data, uint64_t size, uint64_t bufsize, zio_transform_func_t *transform) { zio_transform_t *zt = kmem_alloc(sizeof (zio_transform_t), KM_SLEEP); zt->zt_orig_abd = zio->io_abd; zt->zt_orig_size = zio->io_size; zt->zt_bufsize = bufsize; zt->zt_transform = transform; zt->zt_next = zio->io_transform_stack; zio->io_transform_stack = zt; zio->io_abd = data; zio->io_size = size; } void zio_pop_transforms(zio_t *zio) { zio_transform_t *zt; while ((zt = zio->io_transform_stack) != NULL) { if (zt->zt_transform != NULL) zt->zt_transform(zio, zt->zt_orig_abd, zt->zt_orig_size); if (zt->zt_bufsize != 0) abd_free(zio->io_abd); zio->io_abd = zt->zt_orig_abd; zio->io_size = zt->zt_orig_size; zio->io_transform_stack = zt->zt_next; kmem_free(zt, sizeof (zio_transform_t)); } } /* * ========================================================================== * I/O transform callbacks for subblocks, decompression, and decryption * ========================================================================== */ static void zio_subblock(zio_t *zio, abd_t *data, uint64_t size) { ASSERT(zio->io_size > size); if (zio->io_type == ZIO_TYPE_READ) abd_copy(data, zio->io_abd, size); } static void zio_decompress(zio_t *zio, abd_t *data, uint64_t size) { if (zio->io_error == 0) { void *tmp = abd_borrow_buf(data, size); int ret = zio_decompress_data(BP_GET_COMPRESS(zio->io_bp), zio->io_abd, tmp, zio->io_size, size, &zio->io_prop.zp_complevel); abd_return_buf_copy(data, tmp, size); if (zio_injection_enabled && ret == 0) ret = zio_handle_fault_injection(zio, EINVAL); if (ret != 0) zio->io_error = SET_ERROR(EIO); } } static void zio_decrypt(zio_t *zio, abd_t *data, uint64_t size) { int ret; void *tmp; blkptr_t *bp = zio->io_bp; spa_t *spa = zio->io_spa; uint64_t dsobj = zio->io_bookmark.zb_objset; uint64_t lsize = BP_GET_LSIZE(bp); dmu_object_type_t ot = BP_GET_TYPE(bp); uint8_t salt[ZIO_DATA_SALT_LEN]; uint8_t iv[ZIO_DATA_IV_LEN]; uint8_t mac[ZIO_DATA_MAC_LEN]; boolean_t no_crypt = B_FALSE; ASSERT(BP_USES_CRYPT(bp)); ASSERT3U(size, !=, 0); if (zio->io_error != 0) return; /* * Verify the cksum of MACs stored in an indirect bp. It will always * be possible to verify this since it does not require an encryption * key. */ if (BP_HAS_INDIRECT_MAC_CKSUM(bp)) { zio_crypt_decode_mac_bp(bp, mac); if (BP_GET_COMPRESS(bp) != ZIO_COMPRESS_OFF) { /* * We haven't decompressed the data yet, but * zio_crypt_do_indirect_mac_checksum() requires * decompressed data to be able to parse out the MACs * from the indirect block. We decompress it now and * throw away the result after we are finished. */ tmp = zio_buf_alloc(lsize); ret = zio_decompress_data(BP_GET_COMPRESS(bp), zio->io_abd, tmp, zio->io_size, lsize, &zio->io_prop.zp_complevel); if (ret != 0) { ret = SET_ERROR(EIO); goto error; } ret = zio_crypt_do_indirect_mac_checksum(B_FALSE, tmp, lsize, BP_SHOULD_BYTESWAP(bp), mac); zio_buf_free(tmp, lsize); } else { ret = zio_crypt_do_indirect_mac_checksum_abd(B_FALSE, zio->io_abd, size, BP_SHOULD_BYTESWAP(bp), mac); } abd_copy(data, zio->io_abd, size); if (zio_injection_enabled && ot != DMU_OT_DNODE && ret == 0) { ret = zio_handle_decrypt_injection(spa, &zio->io_bookmark, ot, ECKSUM); } if (ret != 0) goto error; return; } /* * If this is an authenticated block, just check the MAC. It would be * nice to separate this out into its own flag, but for the moment * enum zio_flag is out of bits. */ if (BP_IS_AUTHENTICATED(bp)) { if (ot == DMU_OT_OBJSET) { ret = spa_do_crypt_objset_mac_abd(B_FALSE, spa, dsobj, zio->io_abd, size, BP_SHOULD_BYTESWAP(bp)); } else { zio_crypt_decode_mac_bp(bp, mac); ret = spa_do_crypt_mac_abd(B_FALSE, spa, dsobj, zio->io_abd, size, mac); if (zio_injection_enabled && ret == 0) { ret = zio_handle_decrypt_injection(spa, &zio->io_bookmark, ot, ECKSUM); } } abd_copy(data, zio->io_abd, size); if (ret != 0) goto error; return; } zio_crypt_decode_params_bp(bp, salt, iv); if (ot == DMU_OT_INTENT_LOG) { tmp = abd_borrow_buf_copy(zio->io_abd, sizeof (zil_chain_t)); zio_crypt_decode_mac_zil(tmp, mac); abd_return_buf(zio->io_abd, tmp, sizeof (zil_chain_t)); } else { zio_crypt_decode_mac_bp(bp, mac); } ret = spa_do_crypt_abd(B_FALSE, spa, &zio->io_bookmark, BP_GET_TYPE(bp), BP_GET_DEDUP(bp), BP_SHOULD_BYTESWAP(bp), salt, iv, mac, size, data, zio->io_abd, &no_crypt); if (no_crypt) abd_copy(data, zio->io_abd, size); if (ret != 0) goto error; return; error: /* assert that the key was found unless this was speculative */ ASSERT(ret != EACCES || (zio->io_flags & ZIO_FLAG_SPECULATIVE)); /* * If there was a decryption / authentication error return EIO as * the io_error. If this was not a speculative zio, create an ereport. */ if (ret == ECKSUM) { zio->io_error = SET_ERROR(EIO); if ((zio->io_flags & ZIO_FLAG_SPECULATIVE) == 0) { spa_log_error(spa, &zio->io_bookmark); (void) zfs_ereport_post(FM_EREPORT_ZFS_AUTHENTICATION, spa, NULL, &zio->io_bookmark, zio, 0); } } else { zio->io_error = ret; } } /* * ========================================================================== * I/O parent/child relationships and pipeline interlocks * ========================================================================== */ zio_t * zio_walk_parents(zio_t *cio, zio_link_t **zl) { list_t *pl = &cio->io_parent_list; *zl = (*zl == NULL) ? list_head(pl) : list_next(pl, *zl); if (*zl == NULL) return (NULL); ASSERT((*zl)->zl_child == cio); return ((*zl)->zl_parent); } zio_t * zio_walk_children(zio_t *pio, zio_link_t **zl) { list_t *cl = &pio->io_child_list; ASSERT(MUTEX_HELD(&pio->io_lock)); *zl = (*zl == NULL) ? list_head(cl) : list_next(cl, *zl); if (*zl == NULL) return (NULL); ASSERT((*zl)->zl_parent == pio); return ((*zl)->zl_child); } zio_t * zio_unique_parent(zio_t *cio) { zio_link_t *zl = NULL; zio_t *pio = zio_walk_parents(cio, &zl); VERIFY3P(zio_walk_parents(cio, &zl), ==, NULL); return (pio); } void zio_add_child(zio_t *pio, zio_t *cio) { zio_link_t *zl = kmem_cache_alloc(zio_link_cache, KM_SLEEP); /* * Logical I/Os can have logical, gang, or vdev children. * Gang I/Os can have gang or vdev children. * Vdev I/Os can only have vdev children. * The following ASSERT captures all of these constraints. */ ASSERT3S(cio->io_child_type, <=, pio->io_child_type); zl->zl_parent = pio; zl->zl_child = cio; mutex_enter(&pio->io_lock); mutex_enter(&cio->io_lock); ASSERT(pio->io_state[ZIO_WAIT_DONE] == 0); for (int w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_children[cio->io_child_type][w] += !cio->io_state[w]; list_insert_head(&pio->io_child_list, zl); list_insert_head(&cio->io_parent_list, zl); pio->io_child_count++; cio->io_parent_count++; mutex_exit(&cio->io_lock); mutex_exit(&pio->io_lock); } static void zio_remove_child(zio_t *pio, zio_t *cio, zio_link_t *zl) { ASSERT(zl->zl_parent == pio); ASSERT(zl->zl_child == cio); mutex_enter(&pio->io_lock); mutex_enter(&cio->io_lock); list_remove(&pio->io_child_list, zl); list_remove(&cio->io_parent_list, zl); pio->io_child_count--; cio->io_parent_count--; mutex_exit(&cio->io_lock); mutex_exit(&pio->io_lock); kmem_cache_free(zio_link_cache, zl); } static boolean_t zio_wait_for_children(zio_t *zio, uint8_t childbits, enum zio_wait_type wait) { boolean_t waiting = B_FALSE; mutex_enter(&zio->io_lock); ASSERT(zio->io_stall == NULL); for (int c = 0; c < ZIO_CHILD_TYPES; c++) { if (!(ZIO_CHILD_BIT_IS_SET(childbits, c))) continue; uint64_t *countp = &zio->io_children[c][wait]; if (*countp != 0) { zio->io_stage >>= 1; ASSERT3U(zio->io_stage, !=, ZIO_STAGE_OPEN); zio->io_stall = countp; waiting = B_TRUE; break; } } mutex_exit(&zio->io_lock); return (waiting); } __attribute__((always_inline)) static inline void zio_notify_parent(zio_t *pio, zio_t *zio, enum zio_wait_type wait, zio_t **next_to_executep) { uint64_t *countp = &pio->io_children[zio->io_child_type][wait]; int *errorp = &pio->io_child_error[zio->io_child_type]; mutex_enter(&pio->io_lock); if (zio->io_error && !(zio->io_flags & ZIO_FLAG_DONT_PROPAGATE)) *errorp = zio_worst_error(*errorp, zio->io_error); pio->io_reexecute |= zio->io_reexecute; ASSERT3U(*countp, >, 0); (*countp)--; if (*countp == 0 && pio->io_stall == countp) { zio_taskq_type_t type = pio->io_stage < ZIO_STAGE_VDEV_IO_START ? ZIO_TASKQ_ISSUE : ZIO_TASKQ_INTERRUPT; pio->io_stall = NULL; mutex_exit(&pio->io_lock); /* * If we can tell the caller to execute this parent next, do * so. Otherwise dispatch the parent zio as its own task. * * Having the caller execute the parent when possible reduces * locking on the zio taskq's, reduces context switch * overhead, and has no recursion penalty. Note that one * read from disk typically causes at least 3 zio's: a * zio_null(), the logical zio_read(), and then a physical * zio. When the physical ZIO completes, we are able to call * zio_done() on all 3 of these zio's from one invocation of * zio_execute() by returning the parent back to * zio_execute(). Since the parent isn't executed until this * thread returns back to zio_execute(), the caller should do * so promptly. * * In other cases, dispatching the parent prevents * overflowing the stack when we have deeply nested * parent-child relationships, as we do with the "mega zio" * of writes for spa_sync(), and the chain of ZIL blocks. */ if (next_to_executep != NULL && *next_to_executep == NULL) { *next_to_executep = pio; } else { zio_taskq_dispatch(pio, type, B_FALSE); } } else { mutex_exit(&pio->io_lock); } } static void zio_inherit_child_errors(zio_t *zio, enum zio_child c) { if (zio->io_child_error[c] != 0 && zio->io_error == 0) zio->io_error = zio->io_child_error[c]; } int zio_bookmark_compare(const void *x1, const void *x2) { const zio_t *z1 = x1; const zio_t *z2 = x2; if (z1->io_bookmark.zb_objset < z2->io_bookmark.zb_objset) return (-1); if (z1->io_bookmark.zb_objset > z2->io_bookmark.zb_objset) return (1); if (z1->io_bookmark.zb_object < z2->io_bookmark.zb_object) return (-1); if (z1->io_bookmark.zb_object > z2->io_bookmark.zb_object) return (1); if (z1->io_bookmark.zb_level < z2->io_bookmark.zb_level) return (-1); if (z1->io_bookmark.zb_level > z2->io_bookmark.zb_level) return (1); if (z1->io_bookmark.zb_blkid < z2->io_bookmark.zb_blkid) return (-1); if (z1->io_bookmark.zb_blkid > z2->io_bookmark.zb_blkid) return (1); if (z1 < z2) return (-1); if (z1 > z2) return (1); return (0); } /* * ========================================================================== * Create the various types of I/O (read, write, free, etc) * ========================================================================== */ static zio_t * zio_create(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, abd_t *data, uint64_t lsize, uint64_t psize, zio_done_func_t *done, void *private, zio_type_t type, zio_priority_t priority, enum zio_flag flags, vdev_t *vd, uint64_t offset, const zbookmark_phys_t *zb, enum zio_stage stage, enum zio_stage pipeline) { zio_t *zio; IMPLY(type != ZIO_TYPE_TRIM, psize <= SPA_MAXBLOCKSIZE); ASSERT(P2PHASE(psize, SPA_MINBLOCKSIZE) == 0); ASSERT(P2PHASE(offset, SPA_MINBLOCKSIZE) == 0); ASSERT(!vd || spa_config_held(spa, SCL_STATE_ALL, RW_READER)); ASSERT(!bp || !(flags & ZIO_FLAG_CONFIG_WRITER)); ASSERT(vd || stage == ZIO_STAGE_OPEN); IMPLY(lsize != psize, (flags & ZIO_FLAG_RAW_COMPRESS) != 0); zio = kmem_cache_alloc(zio_cache, KM_SLEEP); bzero(zio, sizeof (zio_t)); mutex_init(&zio->io_lock, NULL, MUTEX_NOLOCKDEP, NULL); cv_init(&zio->io_cv, NULL, CV_DEFAULT, NULL); list_create(&zio->io_parent_list, sizeof (zio_link_t), offsetof(zio_link_t, zl_parent_node)); list_create(&zio->io_child_list, sizeof (zio_link_t), offsetof(zio_link_t, zl_child_node)); metaslab_trace_init(&zio->io_alloc_list); if (vd != NULL) zio->io_child_type = ZIO_CHILD_VDEV; else if (flags & ZIO_FLAG_GANG_CHILD) zio->io_child_type = ZIO_CHILD_GANG; else if (flags & ZIO_FLAG_DDT_CHILD) zio->io_child_type = ZIO_CHILD_DDT; else zio->io_child_type = ZIO_CHILD_LOGICAL; if (bp != NULL) { zio->io_bp = (blkptr_t *)bp; zio->io_bp_copy = *bp; zio->io_bp_orig = *bp; if (type != ZIO_TYPE_WRITE || zio->io_child_type == ZIO_CHILD_DDT) zio->io_bp = &zio->io_bp_copy; /* so caller can free */ if (zio->io_child_type == ZIO_CHILD_LOGICAL) zio->io_logical = zio; if (zio->io_child_type > ZIO_CHILD_GANG && BP_IS_GANG(bp)) pipeline |= ZIO_GANG_STAGES; } zio->io_spa = spa; zio->io_txg = txg; zio->io_done = done; zio->io_private = private; zio->io_type = type; zio->io_priority = priority; zio->io_vd = vd; zio->io_offset = offset; zio->io_orig_abd = zio->io_abd = data; zio->io_orig_size = zio->io_size = psize; zio->io_lsize = lsize; zio->io_orig_flags = zio->io_flags = flags; zio->io_orig_stage = zio->io_stage = stage; zio->io_orig_pipeline = zio->io_pipeline = pipeline; zio->io_pipeline_trace = ZIO_STAGE_OPEN; zio->io_state[ZIO_WAIT_READY] = (stage >= ZIO_STAGE_READY); zio->io_state[ZIO_WAIT_DONE] = (stage >= ZIO_STAGE_DONE); if (zb != NULL) zio->io_bookmark = *zb; if (pio != NULL) { if (zio->io_metaslab_class == NULL) zio->io_metaslab_class = pio->io_metaslab_class; if (zio->io_logical == NULL) zio->io_logical = pio->io_logical; if (zio->io_child_type == ZIO_CHILD_GANG) zio->io_gang_leader = pio->io_gang_leader; zio_add_child(pio, zio); } taskq_init_ent(&zio->io_tqent); return (zio); } static void zio_destroy(zio_t *zio) { metaslab_trace_fini(&zio->io_alloc_list); list_destroy(&zio->io_parent_list); list_destroy(&zio->io_child_list); mutex_destroy(&zio->io_lock); cv_destroy(&zio->io_cv); kmem_cache_free(zio_cache, zio); } zio_t * zio_null(zio_t *pio, spa_t *spa, vdev_t *vd, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; zio = zio_create(pio, spa, 0, NULL, NULL, 0, 0, done, private, ZIO_TYPE_NULL, ZIO_PRIORITY_NOW, flags, vd, 0, NULL, ZIO_STAGE_OPEN, ZIO_INTERLOCK_PIPELINE); return (zio); } zio_t * zio_root(spa_t *spa, zio_done_func_t *done, void *private, enum zio_flag flags) { return (zio_null(NULL, spa, NULL, done, private, flags)); } static int zfs_blkptr_verify_log(spa_t *spa, const blkptr_t *bp, enum blk_verify_flag blk_verify, const char *fmt, ...) { va_list adx; char buf[256]; va_start(adx, fmt); (void) vsnprintf(buf, sizeof (buf), fmt, adx); va_end(adx); switch (blk_verify) { case BLK_VERIFY_HALT: dprintf_bp(bp, "blkptr at %p dprintf_bp():", bp); zfs_panic_recover("%s: %s", spa_name(spa), buf); break; case BLK_VERIFY_LOG: zfs_dbgmsg("%s: %s", spa_name(spa), buf); break; case BLK_VERIFY_ONLY: break; } return (1); } /* * Verify the block pointer fields contain reasonable values. This means * it only contains known object types, checksum/compression identifiers, * block sizes within the maximum allowed limits, valid DVAs, etc. * * If everything checks out B_TRUE is returned. The zfs_blkptr_verify * argument controls the behavior when an invalid field is detected. * * Modes for zfs_blkptr_verify: * 1) BLK_VERIFY_ONLY (evaluate the block) * 2) BLK_VERIFY_LOG (evaluate the block and log problems) * 3) BLK_VERIFY_HALT (call zfs_panic_recover on error) */ boolean_t zfs_blkptr_verify(spa_t *spa, const blkptr_t *bp, boolean_t config_held, enum blk_verify_flag blk_verify) { int errors = 0; if (!DMU_OT_IS_VALID(BP_GET_TYPE(bp))) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p has invalid TYPE %llu", bp, (longlong_t)BP_GET_TYPE(bp)); } if (BP_GET_CHECKSUM(bp) >= ZIO_CHECKSUM_FUNCTIONS || BP_GET_CHECKSUM(bp) <= ZIO_CHECKSUM_ON) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p has invalid CHECKSUM %llu", bp, (longlong_t)BP_GET_CHECKSUM(bp)); } if (BP_GET_COMPRESS(bp) >= ZIO_COMPRESS_FUNCTIONS || BP_GET_COMPRESS(bp) <= ZIO_COMPRESS_ON) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p has invalid COMPRESS %llu", bp, (longlong_t)BP_GET_COMPRESS(bp)); } if (BP_GET_LSIZE(bp) > SPA_MAXBLOCKSIZE) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p has invalid LSIZE %llu", bp, (longlong_t)BP_GET_LSIZE(bp)); } if (BP_GET_PSIZE(bp) > SPA_MAXBLOCKSIZE) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p has invalid PSIZE %llu", bp, (longlong_t)BP_GET_PSIZE(bp)); } if (BP_IS_EMBEDDED(bp)) { if (BPE_GET_ETYPE(bp) >= NUM_BP_EMBEDDED_TYPES) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p has invalid ETYPE %llu", bp, (longlong_t)BPE_GET_ETYPE(bp)); } } /* * Do not verify individual DVAs if the config is not trusted. This * will be done once the zio is executed in vdev_mirror_map_alloc. */ if (!spa->spa_trust_config) return (B_TRUE); if (!config_held) spa_config_enter(spa, SCL_VDEV, bp, RW_READER); else ASSERT(spa_config_held(spa, SCL_VDEV, RW_WRITER)); /* * Pool-specific checks. * * Note: it would be nice to verify that the blk_birth and * BP_PHYSICAL_BIRTH() are not too large. However, spa_freeze() * allows the birth time of log blocks (and dmu_sync()-ed blocks * that are in the log) to be arbitrarily large. */ for (int i = 0; i < BP_GET_NDVAS(bp); i++) { uint64_t vdevid = DVA_GET_VDEV(&bp->blk_dva[i]); if (vdevid >= spa->spa_root_vdev->vdev_children) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p DVA %u has invalid VDEV %llu", bp, i, (longlong_t)vdevid); continue; } vdev_t *vd = spa->spa_root_vdev->vdev_child[vdevid]; if (vd == NULL) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p DVA %u has invalid VDEV %llu", bp, i, (longlong_t)vdevid); continue; } if (vd->vdev_ops == &vdev_hole_ops) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p DVA %u has hole VDEV %llu", bp, i, (longlong_t)vdevid); continue; } if (vd->vdev_ops == &vdev_missing_ops) { /* * "missing" vdevs are valid during import, but we * don't have their detailed info (e.g. asize), so * we can't perform any more checks on them. */ continue; } uint64_t offset = DVA_GET_OFFSET(&bp->blk_dva[i]); uint64_t asize = DVA_GET_ASIZE(&bp->blk_dva[i]); if (BP_IS_GANG(bp)) asize = vdev_psize_to_asize(vd, SPA_GANGBLOCKSIZE); if (offset + asize > vd->vdev_asize) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p DVA %u has invalid OFFSET %llu", bp, i, (longlong_t)offset); } } if (errors > 0) dprintf_bp(bp, "blkptr at %p dprintf_bp():", bp); if (!config_held) spa_config_exit(spa, SCL_VDEV, bp); return (errors == 0); } boolean_t zfs_dva_valid(spa_t *spa, const dva_t *dva, const blkptr_t *bp) { uint64_t vdevid = DVA_GET_VDEV(dva); if (vdevid >= spa->spa_root_vdev->vdev_children) return (B_FALSE); vdev_t *vd = spa->spa_root_vdev->vdev_child[vdevid]; if (vd == NULL) return (B_FALSE); if (vd->vdev_ops == &vdev_hole_ops) return (B_FALSE); if (vd->vdev_ops == &vdev_missing_ops) { return (B_FALSE); } uint64_t offset = DVA_GET_OFFSET(dva); uint64_t asize = DVA_GET_ASIZE(dva); if (BP_IS_GANG(bp)) asize = vdev_psize_to_asize(vd, SPA_GANGBLOCKSIZE); if (offset + asize > vd->vdev_asize) return (B_FALSE); return (B_TRUE); } zio_t * zio_read(zio_t *pio, spa_t *spa, const blkptr_t *bp, abd_t *data, uint64_t size, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, const zbookmark_phys_t *zb) { zio_t *zio; (void) zfs_blkptr_verify(spa, bp, flags & ZIO_FLAG_CONFIG_WRITER, BLK_VERIFY_HALT); zio = zio_create(pio, spa, BP_PHYSICAL_BIRTH(bp), bp, data, size, size, done, private, ZIO_TYPE_READ, priority, flags, NULL, 0, zb, ZIO_STAGE_OPEN, (flags & ZIO_FLAG_DDT_CHILD) ? ZIO_DDT_CHILD_READ_PIPELINE : ZIO_READ_PIPELINE); return (zio); } zio_t * zio_write(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, abd_t *data, uint64_t lsize, uint64_t psize, const zio_prop_t *zp, zio_done_func_t *ready, zio_done_func_t *children_ready, zio_done_func_t *physdone, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, const zbookmark_phys_t *zb) { zio_t *zio; ASSERT(zp->zp_checksum >= ZIO_CHECKSUM_OFF && zp->zp_checksum < ZIO_CHECKSUM_FUNCTIONS && zp->zp_compress >= ZIO_COMPRESS_OFF && zp->zp_compress < ZIO_COMPRESS_FUNCTIONS && DMU_OT_IS_VALID(zp->zp_type) && zp->zp_level < 32 && zp->zp_copies > 0 && zp->zp_copies <= spa_max_replication(spa)); zio = zio_create(pio, spa, txg, bp, data, lsize, psize, done, private, ZIO_TYPE_WRITE, priority, flags, NULL, 0, zb, ZIO_STAGE_OPEN, (flags & ZIO_FLAG_DDT_CHILD) ? ZIO_DDT_CHILD_WRITE_PIPELINE : ZIO_WRITE_PIPELINE); zio->io_ready = ready; zio->io_children_ready = children_ready; zio->io_physdone = physdone; zio->io_prop = *zp; /* * Data can be NULL if we are going to call zio_write_override() to * provide the already-allocated BP. But we may need the data to * verify a dedup hit (if requested). In this case, don't try to * dedup (just take the already-allocated BP verbatim). Encrypted * dedup blocks need data as well so we also disable dedup in this * case. */ if (data == NULL && (zio->io_prop.zp_dedup_verify || zio->io_prop.zp_encrypt)) { zio->io_prop.zp_dedup = zio->io_prop.zp_dedup_verify = B_FALSE; } return (zio); } zio_t * zio_rewrite(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, abd_t *data, uint64_t size, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, zbookmark_phys_t *zb) { zio_t *zio; zio = zio_create(pio, spa, txg, bp, data, size, size, done, private, ZIO_TYPE_WRITE, priority, flags | ZIO_FLAG_IO_REWRITE, NULL, 0, zb, ZIO_STAGE_OPEN, ZIO_REWRITE_PIPELINE); return (zio); } void zio_write_override(zio_t *zio, blkptr_t *bp, int copies, boolean_t nopwrite) { ASSERT(zio->io_type == ZIO_TYPE_WRITE); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(zio->io_stage == ZIO_STAGE_OPEN); ASSERT(zio->io_txg == spa_syncing_txg(zio->io_spa)); /* * We must reset the io_prop to match the values that existed * when the bp was first written by dmu_sync() keeping in mind * that nopwrite and dedup are mutually exclusive. */ zio->io_prop.zp_dedup = nopwrite ? B_FALSE : zio->io_prop.zp_dedup; zio->io_prop.zp_nopwrite = nopwrite; zio->io_prop.zp_copies = copies; zio->io_bp_override = bp; } void zio_free(spa_t *spa, uint64_t txg, const blkptr_t *bp) { (void) zfs_blkptr_verify(spa, bp, B_FALSE, BLK_VERIFY_HALT); /* * The check for EMBEDDED is a performance optimization. We * process the free here (by ignoring it) rather than * putting it on the list and then processing it in zio_free_sync(). */ if (BP_IS_EMBEDDED(bp)) return; metaslab_check_free(spa, bp); /* * Frees that are for the currently-syncing txg, are not going to be * deferred, and which will not need to do a read (i.e. not GANG or * DEDUP), can be processed immediately. Otherwise, put them on the * in-memory list for later processing. * * Note that we only defer frees after zfs_sync_pass_deferred_free * when the log space map feature is disabled. [see relevant comment * in spa_sync_iterate_to_convergence()] */ if (BP_IS_GANG(bp) || BP_GET_DEDUP(bp) || txg != spa->spa_syncing_txg || (spa_sync_pass(spa) >= zfs_sync_pass_deferred_free && !spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP))) { bplist_append(&spa->spa_free_bplist[txg & TXG_MASK], bp); } else { VERIFY3P(zio_free_sync(NULL, spa, txg, bp, 0), ==, NULL); } } /* * To improve performance, this function may return NULL if we were able * to do the free immediately. This avoids the cost of creating a zio * (and linking it to the parent, etc). */ zio_t * zio_free_sync(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, enum zio_flag flags) { ASSERT(!BP_IS_HOLE(bp)); ASSERT(spa_syncing_txg(spa) == txg); if (BP_IS_EMBEDDED(bp)) return (NULL); metaslab_check_free(spa, bp); arc_freed(spa, bp); dsl_scan_freed(spa, bp); if (BP_IS_GANG(bp) || BP_GET_DEDUP(bp)) { /* * GANG and DEDUP blocks can induce a read (for the gang block * header, or the DDT), so issue them asynchronously so that * this thread is not tied up. */ enum zio_stage stage = ZIO_FREE_PIPELINE | ZIO_STAGE_ISSUE_ASYNC; return (zio_create(pio, spa, txg, bp, NULL, BP_GET_PSIZE(bp), BP_GET_PSIZE(bp), NULL, NULL, ZIO_TYPE_FREE, ZIO_PRIORITY_NOW, flags, NULL, 0, NULL, ZIO_STAGE_OPEN, stage)); } else { metaslab_free(spa, bp, txg, B_FALSE); return (NULL); } } zio_t * zio_claim(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; (void) zfs_blkptr_verify(spa, bp, flags & ZIO_FLAG_CONFIG_WRITER, BLK_VERIFY_HALT); if (BP_IS_EMBEDDED(bp)) return (zio_null(pio, spa, NULL, NULL, NULL, 0)); /* * A claim is an allocation of a specific block. Claims are needed * to support immediate writes in the intent log. The issue is that * immediate writes contain committed data, but in a txg that was * *not* committed. Upon opening the pool after an unclean shutdown, * the intent log claims all blocks that contain immediate write data * so that the SPA knows they're in use. * * All claims *must* be resolved in the first txg -- before the SPA * starts allocating blocks -- so that nothing is allocated twice. * If txg == 0 we just verify that the block is claimable. */ ASSERT3U(spa->spa_uberblock.ub_rootbp.blk_birth, <, spa_min_claim_txg(spa)); ASSERT(txg == spa_min_claim_txg(spa) || txg == 0); - ASSERT(!BP_GET_DEDUP(bp) || !spa_writeable(spa)); /* zdb(1M) */ + ASSERT(!BP_GET_DEDUP(bp) || !spa_writeable(spa)); /* zdb(8) */ zio = zio_create(pio, spa, txg, bp, NULL, BP_GET_PSIZE(bp), BP_GET_PSIZE(bp), done, private, ZIO_TYPE_CLAIM, ZIO_PRIORITY_NOW, flags, NULL, 0, NULL, ZIO_STAGE_OPEN, ZIO_CLAIM_PIPELINE); ASSERT0(zio->io_queued_timestamp); return (zio); } zio_t * zio_ioctl(zio_t *pio, spa_t *spa, vdev_t *vd, int cmd, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; int c; if (vd->vdev_children == 0) { zio = zio_create(pio, spa, 0, NULL, NULL, 0, 0, done, private, ZIO_TYPE_IOCTL, ZIO_PRIORITY_NOW, flags, vd, 0, NULL, ZIO_STAGE_OPEN, ZIO_IOCTL_PIPELINE); zio->io_cmd = cmd; } else { zio = zio_null(pio, spa, NULL, NULL, NULL, flags); for (c = 0; c < vd->vdev_children; c++) zio_nowait(zio_ioctl(zio, spa, vd->vdev_child[c], cmd, done, private, flags)); } return (zio); } zio_t * zio_trim(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, enum trim_flag trim_flags) { zio_t *zio; ASSERT0(vd->vdev_children); ASSERT0(P2PHASE(offset, 1ULL << vd->vdev_ashift)); ASSERT0(P2PHASE(size, 1ULL << vd->vdev_ashift)); ASSERT3U(size, !=, 0); zio = zio_create(pio, vd->vdev_spa, 0, NULL, NULL, size, size, done, private, ZIO_TYPE_TRIM, priority, flags | ZIO_FLAG_PHYSICAL, vd, offset, NULL, ZIO_STAGE_OPEN, ZIO_TRIM_PIPELINE); zio->io_trim_flags = trim_flags; return (zio); } zio_t * zio_read_phys(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, abd_t *data, int checksum, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, boolean_t labels) { zio_t *zio; ASSERT(vd->vdev_children == 0); ASSERT(!labels || offset + size <= VDEV_LABEL_START_SIZE || offset >= vd->vdev_psize - VDEV_LABEL_END_SIZE); ASSERT3U(offset + size, <=, vd->vdev_psize); zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, size, done, private, ZIO_TYPE_READ, priority, flags | ZIO_FLAG_PHYSICAL, vd, offset, NULL, ZIO_STAGE_OPEN, ZIO_READ_PHYS_PIPELINE); zio->io_prop.zp_checksum = checksum; return (zio); } zio_t * zio_write_phys(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, abd_t *data, int checksum, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, boolean_t labels) { zio_t *zio; ASSERT(vd->vdev_children == 0); ASSERT(!labels || offset + size <= VDEV_LABEL_START_SIZE || offset >= vd->vdev_psize - VDEV_LABEL_END_SIZE); ASSERT3U(offset + size, <=, vd->vdev_psize); zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, size, done, private, ZIO_TYPE_WRITE, priority, flags | ZIO_FLAG_PHYSICAL, vd, offset, NULL, ZIO_STAGE_OPEN, ZIO_WRITE_PHYS_PIPELINE); zio->io_prop.zp_checksum = checksum; if (zio_checksum_table[checksum].ci_flags & ZCHECKSUM_FLAG_EMBEDDED) { /* * zec checksums are necessarily destructive -- they modify * the end of the write buffer to hold the verifier/checksum. * Therefore, we must make a local copy in case the data is * being written to multiple places in parallel. */ abd_t *wbuf = abd_alloc_sametype(data, size); abd_copy(wbuf, data, size); zio_push_transform(zio, wbuf, size, size, NULL); } return (zio); } /* * Create a child I/O to do some work for us. */ zio_t * zio_vdev_child_io(zio_t *pio, blkptr_t *bp, vdev_t *vd, uint64_t offset, abd_t *data, uint64_t size, int type, zio_priority_t priority, enum zio_flag flags, zio_done_func_t *done, void *private) { enum zio_stage pipeline = ZIO_VDEV_CHILD_PIPELINE; zio_t *zio; /* * vdev child I/Os do not propagate their error to the parent. * Therefore, for correct operation the caller *must* check for * and handle the error in the child i/o's done callback. * The only exceptions are i/os that we don't care about * (OPTIONAL or REPAIR). */ ASSERT((flags & ZIO_FLAG_OPTIONAL) || (flags & ZIO_FLAG_IO_REPAIR) || done != NULL); if (type == ZIO_TYPE_READ && bp != NULL) { /* * If we have the bp, then the child should perform the * checksum and the parent need not. This pushes error * detection as close to the leaves as possible and * eliminates redundant checksums in the interior nodes. */ pipeline |= ZIO_STAGE_CHECKSUM_VERIFY; pio->io_pipeline &= ~ZIO_STAGE_CHECKSUM_VERIFY; } if (vd->vdev_ops->vdev_op_leaf) { ASSERT0(vd->vdev_children); offset += VDEV_LABEL_START_SIZE; } flags |= ZIO_VDEV_CHILD_FLAGS(pio); /* * If we've decided to do a repair, the write is not speculative -- * even if the original read was. */ if (flags & ZIO_FLAG_IO_REPAIR) flags &= ~ZIO_FLAG_SPECULATIVE; /* * If we're creating a child I/O that is not associated with a * top-level vdev, then the child zio is not an allocating I/O. * If this is a retried I/O then we ignore it since we will * have already processed the original allocating I/O. */ if (flags & ZIO_FLAG_IO_ALLOCATING && (vd != vd->vdev_top || (flags & ZIO_FLAG_IO_RETRY))) { ASSERT(pio->io_metaslab_class != NULL); ASSERT(pio->io_metaslab_class->mc_alloc_throttle_enabled); ASSERT(type == ZIO_TYPE_WRITE); ASSERT(priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(!(flags & ZIO_FLAG_IO_REPAIR)); ASSERT(!(pio->io_flags & ZIO_FLAG_IO_REWRITE) || pio->io_child_type == ZIO_CHILD_GANG); flags &= ~ZIO_FLAG_IO_ALLOCATING; } zio = zio_create(pio, pio->io_spa, pio->io_txg, bp, data, size, size, done, private, type, priority, flags, vd, offset, &pio->io_bookmark, ZIO_STAGE_VDEV_IO_START >> 1, pipeline); ASSERT3U(zio->io_child_type, ==, ZIO_CHILD_VDEV); zio->io_physdone = pio->io_physdone; if (vd->vdev_ops->vdev_op_leaf && zio->io_logical != NULL) zio->io_logical->io_phys_children++; return (zio); } zio_t * zio_vdev_delegated_io(vdev_t *vd, uint64_t offset, abd_t *data, uint64_t size, zio_type_t type, zio_priority_t priority, enum zio_flag flags, zio_done_func_t *done, void *private) { zio_t *zio; ASSERT(vd->vdev_ops->vdev_op_leaf); zio = zio_create(NULL, vd->vdev_spa, 0, NULL, data, size, size, done, private, type, priority, flags | ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_RETRY | ZIO_FLAG_DELEGATED, vd, offset, NULL, ZIO_STAGE_VDEV_IO_START >> 1, ZIO_VDEV_CHILD_PIPELINE); return (zio); } void zio_flush(zio_t *zio, vdev_t *vd) { zio_nowait(zio_ioctl(zio, zio->io_spa, vd, DKIOCFLUSHWRITECACHE, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_DONT_RETRY)); } void zio_shrink(zio_t *zio, uint64_t size) { ASSERT3P(zio->io_executor, ==, NULL); ASSERT3U(zio->io_orig_size, ==, zio->io_size); ASSERT3U(size, <=, zio->io_size); /* * We don't shrink for raidz because of problems with the * reconstruction when reading back less than the block size. * Note, BP_IS_RAIDZ() assumes no compression. */ ASSERT(BP_GET_COMPRESS(zio->io_bp) == ZIO_COMPRESS_OFF); if (!BP_IS_RAIDZ(zio->io_bp)) { /* we are not doing a raw write */ ASSERT3U(zio->io_size, ==, zio->io_lsize); zio->io_orig_size = zio->io_size = zio->io_lsize = size; } } /* * ========================================================================== * Prepare to read and write logical blocks * ========================================================================== */ static zio_t * zio_read_bp_init(zio_t *zio) { blkptr_t *bp = zio->io_bp; uint64_t psize = BP_IS_EMBEDDED(bp) ? BPE_GET_PSIZE(bp) : BP_GET_PSIZE(bp); ASSERT3P(zio->io_bp, ==, &zio->io_bp_copy); if (BP_GET_COMPRESS(bp) != ZIO_COMPRESS_OFF && zio->io_child_type == ZIO_CHILD_LOGICAL && !(zio->io_flags & ZIO_FLAG_RAW_COMPRESS)) { zio_push_transform(zio, abd_alloc_sametype(zio->io_abd, psize), psize, psize, zio_decompress); } if (((BP_IS_PROTECTED(bp) && !(zio->io_flags & ZIO_FLAG_RAW_ENCRYPT)) || BP_HAS_INDIRECT_MAC_CKSUM(bp)) && zio->io_child_type == ZIO_CHILD_LOGICAL) { zio_push_transform(zio, abd_alloc_sametype(zio->io_abd, psize), psize, psize, zio_decrypt); } if (BP_IS_EMBEDDED(bp) && BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_DATA) { int psize = BPE_GET_PSIZE(bp); void *data = abd_borrow_buf(zio->io_abd, psize); zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; decode_embedded_bp_compressed(bp, data); abd_return_buf_copy(zio->io_abd, data, psize); } else { ASSERT(!BP_IS_EMBEDDED(bp)); ASSERT3P(zio->io_bp, ==, &zio->io_bp_copy); } if (!DMU_OT_IS_METADATA(BP_GET_TYPE(bp)) && BP_GET_LEVEL(bp) == 0) zio->io_flags |= ZIO_FLAG_DONT_CACHE; if (BP_GET_TYPE(bp) == DMU_OT_DDT_ZAP) zio->io_flags |= ZIO_FLAG_DONT_CACHE; if (BP_GET_DEDUP(bp) && zio->io_child_type == ZIO_CHILD_LOGICAL) zio->io_pipeline = ZIO_DDT_READ_PIPELINE; return (zio); } static zio_t * zio_write_bp_init(zio_t *zio) { if (!IO_IS_ALLOCATING(zio)) return (zio); ASSERT(zio->io_child_type != ZIO_CHILD_DDT); if (zio->io_bp_override) { blkptr_t *bp = zio->io_bp; zio_prop_t *zp = &zio->io_prop; ASSERT(bp->blk_birth != zio->io_txg); ASSERT(BP_GET_DEDUP(zio->io_bp_override) == 0); *bp = *zio->io_bp_override; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (BP_IS_EMBEDDED(bp)) return (zio); /* * If we've been overridden and nopwrite is set then * set the flag accordingly to indicate that a nopwrite * has already occurred. */ if (!BP_IS_HOLE(bp) && zp->zp_nopwrite) { ASSERT(!zp->zp_dedup); ASSERT3U(BP_GET_CHECKSUM(bp), ==, zp->zp_checksum); zio->io_flags |= ZIO_FLAG_NOPWRITE; return (zio); } ASSERT(!zp->zp_nopwrite); if (BP_IS_HOLE(bp) || !zp->zp_dedup) return (zio); ASSERT((zio_checksum_table[zp->zp_checksum].ci_flags & ZCHECKSUM_FLAG_DEDUP) || zp->zp_dedup_verify); if (BP_GET_CHECKSUM(bp) == zp->zp_checksum && !zp->zp_encrypt) { BP_SET_DEDUP(bp, 1); zio->io_pipeline |= ZIO_STAGE_DDT_WRITE; return (zio); } /* * We were unable to handle this as an override bp, treat * it as a regular write I/O. */ zio->io_bp_override = NULL; *bp = zio->io_bp_orig; zio->io_pipeline = zio->io_orig_pipeline; } return (zio); } static zio_t * zio_write_compress(zio_t *zio) { spa_t *spa = zio->io_spa; zio_prop_t *zp = &zio->io_prop; enum zio_compress compress = zp->zp_compress; blkptr_t *bp = zio->io_bp; uint64_t lsize = zio->io_lsize; uint64_t psize = zio->io_size; int pass = 1; /* * If our children haven't all reached the ready stage, * wait for them and then repeat this pipeline stage. */ if (zio_wait_for_children(zio, ZIO_CHILD_LOGICAL_BIT | ZIO_CHILD_GANG_BIT, ZIO_WAIT_READY)) { return (NULL); } if (!IO_IS_ALLOCATING(zio)) return (zio); if (zio->io_children_ready != NULL) { /* * Now that all our children are ready, run the callback * associated with this zio in case it wants to modify the * data to be written. */ ASSERT3U(zp->zp_level, >, 0); zio->io_children_ready(zio); } ASSERT(zio->io_child_type != ZIO_CHILD_DDT); ASSERT(zio->io_bp_override == NULL); if (!BP_IS_HOLE(bp) && bp->blk_birth == zio->io_txg) { /* * We're rewriting an existing block, which means we're * working on behalf of spa_sync(). For spa_sync() to * converge, it must eventually be the case that we don't * have to allocate new blocks. But compression changes * the blocksize, which forces a reallocate, and makes * convergence take longer. Therefore, after the first * few passes, stop compressing to ensure convergence. */ pass = spa_sync_pass(spa); ASSERT(zio->io_txg == spa_syncing_txg(spa)); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!BP_GET_DEDUP(bp)); if (pass >= zfs_sync_pass_dont_compress) compress = ZIO_COMPRESS_OFF; /* Make sure someone doesn't change their mind on overwrites */ ASSERT(BP_IS_EMBEDDED(bp) || MIN(zp->zp_copies + BP_IS_GANG(bp), spa_max_replication(spa)) == BP_GET_NDVAS(bp)); } /* If it's a compressed write that is not raw, compress the buffer. */ if (compress != ZIO_COMPRESS_OFF && !(zio->io_flags & ZIO_FLAG_RAW_COMPRESS)) { void *cbuf = zio_buf_alloc(lsize); psize = zio_compress_data(compress, zio->io_abd, cbuf, lsize, zp->zp_complevel); if (psize == 0 || psize >= lsize) { compress = ZIO_COMPRESS_OFF; zio_buf_free(cbuf, lsize); } else if (!zp->zp_dedup && !zp->zp_encrypt && psize <= BPE_PAYLOAD_SIZE && zp->zp_level == 0 && !DMU_OT_HAS_FILL(zp->zp_type) && spa_feature_is_enabled(spa, SPA_FEATURE_EMBEDDED_DATA)) { encode_embedded_bp_compressed(bp, cbuf, compress, lsize, psize); BPE_SET_ETYPE(bp, BP_EMBEDDED_TYPE_DATA); BP_SET_TYPE(bp, zio->io_prop.zp_type); BP_SET_LEVEL(bp, zio->io_prop.zp_level); zio_buf_free(cbuf, lsize); bp->blk_birth = zio->io_txg; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; ASSERT(spa_feature_is_active(spa, SPA_FEATURE_EMBEDDED_DATA)); return (zio); } else { /* * Round up compressed size up to the ashift * of the smallest-ashift device, and zero the tail. * This ensures that the compressed size of the BP * (and thus compressratio property) are correct, * in that we charge for the padding used to fill out * the last sector. */ ASSERT3U(spa->spa_min_ashift, >=, SPA_MINBLOCKSHIFT); size_t rounded = (size_t)P2ROUNDUP(psize, 1ULL << spa->spa_min_ashift); if (rounded >= lsize) { compress = ZIO_COMPRESS_OFF; zio_buf_free(cbuf, lsize); psize = lsize; } else { abd_t *cdata = abd_get_from_buf(cbuf, lsize); abd_take_ownership_of_buf(cdata, B_TRUE); abd_zero_off(cdata, psize, rounded - psize); psize = rounded; zio_push_transform(zio, cdata, psize, lsize, NULL); } } /* * We were unable to handle this as an override bp, treat * it as a regular write I/O. */ zio->io_bp_override = NULL; *bp = zio->io_bp_orig; zio->io_pipeline = zio->io_orig_pipeline; } else if ((zio->io_flags & ZIO_FLAG_RAW_ENCRYPT) != 0 && zp->zp_type == DMU_OT_DNODE) { /* * The DMU actually relies on the zio layer's compression * to free metadnode blocks that have had all contained * dnodes freed. As a result, even when doing a raw * receive, we must check whether the block can be compressed * to a hole. */ psize = zio_compress_data(ZIO_COMPRESS_EMPTY, zio->io_abd, NULL, lsize, zp->zp_complevel); if (psize == 0 || psize >= lsize) compress = ZIO_COMPRESS_OFF; } else { ASSERT3U(psize, !=, 0); } /* * The final pass of spa_sync() must be all rewrites, but the first * few passes offer a trade-off: allocating blocks defers convergence, * but newly allocated blocks are sequential, so they can be written * to disk faster. Therefore, we allow the first few passes of * spa_sync() to allocate new blocks, but force rewrites after that. * There should only be a handful of blocks after pass 1 in any case. */ if (!BP_IS_HOLE(bp) && bp->blk_birth == zio->io_txg && BP_GET_PSIZE(bp) == psize && pass >= zfs_sync_pass_rewrite) { VERIFY3U(psize, !=, 0); enum zio_stage gang_stages = zio->io_pipeline & ZIO_GANG_STAGES; zio->io_pipeline = ZIO_REWRITE_PIPELINE | gang_stages; zio->io_flags |= ZIO_FLAG_IO_REWRITE; } else { BP_ZERO(bp); zio->io_pipeline = ZIO_WRITE_PIPELINE; } if (psize == 0) { if (zio->io_bp_orig.blk_birth != 0 && spa_feature_is_active(spa, SPA_FEATURE_HOLE_BIRTH)) { BP_SET_LSIZE(bp, lsize); BP_SET_TYPE(bp, zp->zp_type); BP_SET_LEVEL(bp, zp->zp_level); BP_SET_BIRTH(bp, zio->io_txg, 0); } zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; } else { ASSERT(zp->zp_checksum != ZIO_CHECKSUM_GANG_HEADER); BP_SET_LSIZE(bp, lsize); BP_SET_TYPE(bp, zp->zp_type); BP_SET_LEVEL(bp, zp->zp_level); BP_SET_PSIZE(bp, psize); BP_SET_COMPRESS(bp, compress); BP_SET_CHECKSUM(bp, zp->zp_checksum); BP_SET_DEDUP(bp, zp->zp_dedup); BP_SET_BYTEORDER(bp, ZFS_HOST_BYTEORDER); if (zp->zp_dedup) { ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); ASSERT(!zp->zp_encrypt || DMU_OT_IS_ENCRYPTED(zp->zp_type)); zio->io_pipeline = ZIO_DDT_WRITE_PIPELINE; } if (zp->zp_nopwrite) { ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); zio->io_pipeline |= ZIO_STAGE_NOP_WRITE; } } return (zio); } static zio_t * zio_free_bp_init(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (zio->io_child_type == ZIO_CHILD_LOGICAL) { if (BP_GET_DEDUP(bp)) zio->io_pipeline = ZIO_DDT_FREE_PIPELINE; } ASSERT3P(zio->io_bp, ==, &zio->io_bp_copy); return (zio); } /* * ========================================================================== * Execute the I/O pipeline * ========================================================================== */ static void zio_taskq_dispatch(zio_t *zio, zio_taskq_type_t q, boolean_t cutinline) { spa_t *spa = zio->io_spa; zio_type_t t = zio->io_type; int flags = (cutinline ? TQ_FRONT : 0); /* * If we're a config writer or a probe, the normal issue and * interrupt threads may all be blocked waiting for the config lock. * In this case, select the otherwise-unused taskq for ZIO_TYPE_NULL. */ if (zio->io_flags & (ZIO_FLAG_CONFIG_WRITER | ZIO_FLAG_PROBE)) t = ZIO_TYPE_NULL; /* * A similar issue exists for the L2ARC write thread until L2ARC 2.0. */ if (t == ZIO_TYPE_WRITE && zio->io_vd && zio->io_vd->vdev_aux) t = ZIO_TYPE_NULL; /* * If this is a high priority I/O, then use the high priority taskq if * available. */ if ((zio->io_priority == ZIO_PRIORITY_NOW || zio->io_priority == ZIO_PRIORITY_SYNC_WRITE) && spa->spa_zio_taskq[t][q + 1].stqs_count != 0) q++; ASSERT3U(q, <, ZIO_TASKQ_TYPES); /* * NB: We are assuming that the zio can only be dispatched * to a single taskq at a time. It would be a grievous error * to dispatch the zio to another taskq at the same time. */ ASSERT(taskq_empty_ent(&zio->io_tqent)); spa_taskq_dispatch_ent(spa, t, q, (task_func_t *)zio_execute, zio, flags, &zio->io_tqent); } static boolean_t zio_taskq_member(zio_t *zio, zio_taskq_type_t q) { spa_t *spa = zio->io_spa; taskq_t *tq = taskq_of_curthread(); for (zio_type_t t = 0; t < ZIO_TYPES; t++) { spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; uint_t i; for (i = 0; i < tqs->stqs_count; i++) { if (tqs->stqs_taskq[i] == tq) return (B_TRUE); } } return (B_FALSE); } static zio_t * zio_issue_async(zio_t *zio) { zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_FALSE); return (NULL); } void zio_interrupt(zio_t *zio) { zio_taskq_dispatch(zio, ZIO_TASKQ_INTERRUPT, B_FALSE); } void zio_delay_interrupt(zio_t *zio) { /* * The timeout_generic() function isn't defined in userspace, so * rather than trying to implement the function, the zio delay * functionality has been disabled for userspace builds. */ #ifdef _KERNEL /* * If io_target_timestamp is zero, then no delay has been registered * for this IO, thus jump to the end of this function and "skip" the * delay; issuing it directly to the zio layer. */ if (zio->io_target_timestamp != 0) { hrtime_t now = gethrtime(); if (now >= zio->io_target_timestamp) { /* * This IO has already taken longer than the target * delay to complete, so we don't want to delay it * any longer; we "miss" the delay and issue it * directly to the zio layer. This is likely due to * the target latency being set to a value less than * the underlying hardware can satisfy (e.g. delay * set to 1ms, but the disks take 10ms to complete an * IO request). */ DTRACE_PROBE2(zio__delay__miss, zio_t *, zio, hrtime_t, now); zio_interrupt(zio); } else { taskqid_t tid; hrtime_t diff = zio->io_target_timestamp - now; clock_t expire_at_tick = ddi_get_lbolt() + NSEC_TO_TICK(diff); DTRACE_PROBE3(zio__delay__hit, zio_t *, zio, hrtime_t, now, hrtime_t, diff); if (NSEC_TO_TICK(diff) == 0) { /* Our delay is less than a jiffy - just spin */ zfs_sleep_until(zio->io_target_timestamp); zio_interrupt(zio); } else { /* * Use taskq_dispatch_delay() in the place of * OpenZFS's timeout_generic(). */ tid = taskq_dispatch_delay(system_taskq, (task_func_t *)zio_interrupt, zio, TQ_NOSLEEP, expire_at_tick); if (tid == TASKQID_INVALID) { /* * Couldn't allocate a task. Just * finish the zio without a delay. */ zio_interrupt(zio); } } } return; } #endif DTRACE_PROBE1(zio__delay__skip, zio_t *, zio); zio_interrupt(zio); } static void zio_deadman_impl(zio_t *pio, int ziodepth) { zio_t *cio, *cio_next; zio_link_t *zl = NULL; vdev_t *vd = pio->io_vd; if (zio_deadman_log_all || (vd != NULL && vd->vdev_ops->vdev_op_leaf)) { vdev_queue_t *vq = vd ? &vd->vdev_queue : NULL; zbookmark_phys_t *zb = &pio->io_bookmark; uint64_t delta = gethrtime() - pio->io_timestamp; uint64_t failmode = spa_get_deadman_failmode(pio->io_spa); zfs_dbgmsg("slow zio[%d]: zio=%px timestamp=%llu " "delta=%llu queued=%llu io=%llu " "path=%s last=%llu " "type=%d priority=%d flags=0x%x " "stage=0x%x pipeline=0x%x pipeline-trace=0x%x " "objset=%llu object=%llu level=%llu blkid=%llu " "offset=%llu size=%llu error=%d", ziodepth, pio, pio->io_timestamp, delta, pio->io_delta, pio->io_delay, vd ? vd->vdev_path : "NULL", vq ? vq->vq_io_complete_ts : 0, pio->io_type, pio->io_priority, pio->io_flags, pio->io_stage, pio->io_pipeline, pio->io_pipeline_trace, zb->zb_objset, zb->zb_object, zb->zb_level, zb->zb_blkid, pio->io_offset, pio->io_size, pio->io_error); (void) zfs_ereport_post(FM_EREPORT_ZFS_DEADMAN, pio->io_spa, vd, zb, pio, 0); if (failmode == ZIO_FAILURE_MODE_CONTINUE && taskq_empty_ent(&pio->io_tqent)) { zio_interrupt(pio); } } mutex_enter(&pio->io_lock); for (cio = zio_walk_children(pio, &zl); cio != NULL; cio = cio_next) { cio_next = zio_walk_children(pio, &zl); zio_deadman_impl(cio, ziodepth + 1); } mutex_exit(&pio->io_lock); } /* * Log the critical information describing this zio and all of its children * using the zfs_dbgmsg() interface then post deadman event for the ZED. */ void zio_deadman(zio_t *pio, char *tag) { spa_t *spa = pio->io_spa; char *name = spa_name(spa); if (!zfs_deadman_enabled || spa_suspended(spa)) return; zio_deadman_impl(pio, 0); switch (spa_get_deadman_failmode(spa)) { case ZIO_FAILURE_MODE_WAIT: zfs_dbgmsg("%s waiting for hung I/O to pool '%s'", tag, name); break; case ZIO_FAILURE_MODE_CONTINUE: zfs_dbgmsg("%s restarting hung I/O for pool '%s'", tag, name); break; case ZIO_FAILURE_MODE_PANIC: fm_panic("%s determined I/O to pool '%s' is hung.", tag, name); break; } } /* * Execute the I/O pipeline until one of the following occurs: * (1) the I/O completes; (2) the pipeline stalls waiting for * dependent child I/Os; (3) the I/O issues, so we're waiting * for an I/O completion interrupt; (4) the I/O is delegated by * vdev-level caching or aggregation; (5) the I/O is deferred * due to vdev-level queueing; (6) the I/O is handed off to * another thread. In all cases, the pipeline stops whenever * there's no CPU work; it never burns a thread in cv_wait_io(). * * There's no locking on io_stage because there's no legitimate way * for multiple threads to be attempting to process the same I/O. */ static zio_pipe_stage_t *zio_pipeline[]; /* * zio_execute() is a wrapper around the static function * __zio_execute() so that we can force __zio_execute() to be * inlined. This reduces stack overhead which is important * because __zio_execute() is called recursively in several zio * code paths. zio_execute() itself cannot be inlined because * it is externally visible. */ void zio_execute(zio_t *zio) { fstrans_cookie_t cookie; cookie = spl_fstrans_mark(); __zio_execute(zio); spl_fstrans_unmark(cookie); } /* * Used to determine if in the current context the stack is sized large * enough to allow zio_execute() to be called recursively. A minimum * stack size of 16K is required to avoid needing to re-dispatch the zio. */ static boolean_t zio_execute_stack_check(zio_t *zio) { #if !defined(HAVE_LARGE_STACKS) dsl_pool_t *dp = spa_get_dsl(zio->io_spa); /* Executing in txg_sync_thread() context. */ if (dp && curthread == dp->dp_tx.tx_sync_thread) return (B_TRUE); /* Pool initialization outside of zio_taskq context. */ if (dp && spa_is_initializing(dp->dp_spa) && !zio_taskq_member(zio, ZIO_TASKQ_ISSUE) && !zio_taskq_member(zio, ZIO_TASKQ_ISSUE_HIGH)) return (B_TRUE); #endif /* HAVE_LARGE_STACKS */ return (B_FALSE); } __attribute__((always_inline)) static inline void __zio_execute(zio_t *zio) { ASSERT3U(zio->io_queued_timestamp, >, 0); while (zio->io_stage < ZIO_STAGE_DONE) { enum zio_stage pipeline = zio->io_pipeline; enum zio_stage stage = zio->io_stage; zio->io_executor = curthread; ASSERT(!MUTEX_HELD(&zio->io_lock)); ASSERT(ISP2(stage)); ASSERT(zio->io_stall == NULL); do { stage <<= 1; } while ((stage & pipeline) == 0); ASSERT(stage <= ZIO_STAGE_DONE); /* * If we are in interrupt context and this pipeline stage * will grab a config lock that is held across I/O, * or may wait for an I/O that needs an interrupt thread * to complete, issue async to avoid deadlock. * * For VDEV_IO_START, we cut in line so that the io will * be sent to disk promptly. */ if ((stage & ZIO_BLOCKING_STAGES) && zio->io_vd == NULL && zio_taskq_member(zio, ZIO_TASKQ_INTERRUPT)) { boolean_t cut = (stage == ZIO_STAGE_VDEV_IO_START) ? zio_requeue_io_start_cut_in_line : B_FALSE; zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, cut); return; } /* * If the current context doesn't have large enough stacks * the zio must be issued asynchronously to prevent overflow. */ if (zio_execute_stack_check(zio)) { boolean_t cut = (stage == ZIO_STAGE_VDEV_IO_START) ? zio_requeue_io_start_cut_in_line : B_FALSE; zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, cut); return; } zio->io_stage = stage; zio->io_pipeline_trace |= zio->io_stage; /* * The zio pipeline stage returns the next zio to execute * (typically the same as this one), or NULL if we should * stop. */ zio = zio_pipeline[highbit64(stage) - 1](zio); if (zio == NULL) return; } } /* * ========================================================================== * Initiate I/O, either sync or async * ========================================================================== */ int zio_wait(zio_t *zio) { /* * Some routines, like zio_free_sync(), may return a NULL zio * to avoid the performance overhead of creating and then destroying * an unneeded zio. For the callers' simplicity, we accept a NULL * zio and ignore it. */ if (zio == NULL) return (0); long timeout = MSEC_TO_TICK(zfs_deadman_ziotime_ms); int error; ASSERT3S(zio->io_stage, ==, ZIO_STAGE_OPEN); ASSERT3P(zio->io_executor, ==, NULL); zio->io_waiter = curthread; ASSERT0(zio->io_queued_timestamp); zio->io_queued_timestamp = gethrtime(); __zio_execute(zio); mutex_enter(&zio->io_lock); while (zio->io_executor != NULL) { error = cv_timedwait_io(&zio->io_cv, &zio->io_lock, ddi_get_lbolt() + timeout); if (zfs_deadman_enabled && error == -1 && gethrtime() - zio->io_queued_timestamp > spa_deadman_ziotime(zio->io_spa)) { mutex_exit(&zio->io_lock); timeout = MSEC_TO_TICK(zfs_deadman_checktime_ms); zio_deadman(zio, FTAG); mutex_enter(&zio->io_lock); } } mutex_exit(&zio->io_lock); error = zio->io_error; zio_destroy(zio); return (error); } void zio_nowait(zio_t *zio) { /* * See comment in zio_wait(). */ if (zio == NULL) return; ASSERT3P(zio->io_executor, ==, NULL); if (zio->io_child_type == ZIO_CHILD_LOGICAL && zio_unique_parent(zio) == NULL) { zio_t *pio; /* * This is a logical async I/O with no parent to wait for it. * We add it to the spa_async_root_zio "Godfather" I/O which * will ensure they complete prior to unloading the pool. */ spa_t *spa = zio->io_spa; kpreempt_disable(); pio = spa->spa_async_zio_root[CPU_SEQID]; kpreempt_enable(); zio_add_child(pio, zio); } ASSERT0(zio->io_queued_timestamp); zio->io_queued_timestamp = gethrtime(); __zio_execute(zio); } /* * ========================================================================== * Reexecute, cancel, or suspend/resume failed I/O * ========================================================================== */ static void zio_reexecute(zio_t *pio) { zio_t *cio, *cio_next; ASSERT(pio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(pio->io_orig_stage == ZIO_STAGE_OPEN); ASSERT(pio->io_gang_leader == NULL); ASSERT(pio->io_gang_tree == NULL); pio->io_flags = pio->io_orig_flags; pio->io_stage = pio->io_orig_stage; pio->io_pipeline = pio->io_orig_pipeline; pio->io_reexecute = 0; pio->io_flags |= ZIO_FLAG_REEXECUTED; pio->io_pipeline_trace = 0; pio->io_error = 0; for (int w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_state[w] = 0; for (int c = 0; c < ZIO_CHILD_TYPES; c++) pio->io_child_error[c] = 0; if (IO_IS_ALLOCATING(pio)) BP_ZERO(pio->io_bp); /* * As we reexecute pio's children, new children could be created. * New children go to the head of pio's io_child_list, however, * so we will (correctly) not reexecute them. The key is that * the remainder of pio's io_child_list, from 'cio_next' onward, * cannot be affected by any side effects of reexecuting 'cio'. */ zio_link_t *zl = NULL; mutex_enter(&pio->io_lock); for (cio = zio_walk_children(pio, &zl); cio != NULL; cio = cio_next) { cio_next = zio_walk_children(pio, &zl); for (int w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_children[cio->io_child_type][w]++; mutex_exit(&pio->io_lock); zio_reexecute(cio); mutex_enter(&pio->io_lock); } mutex_exit(&pio->io_lock); /* * Now that all children have been reexecuted, execute the parent. * We don't reexecute "The Godfather" I/O here as it's the * responsibility of the caller to wait on it. */ if (!(pio->io_flags & ZIO_FLAG_GODFATHER)) { pio->io_queued_timestamp = gethrtime(); __zio_execute(pio); } } void zio_suspend(spa_t *spa, zio_t *zio, zio_suspend_reason_t reason) { if (spa_get_failmode(spa) == ZIO_FAILURE_MODE_PANIC) fm_panic("Pool '%s' has encountered an uncorrectable I/O " "failure and the failure mode property for this pool " "is set to panic.", spa_name(spa)); cmn_err(CE_WARN, "Pool '%s' has encountered an uncorrectable I/O " "failure and has been suspended.\n", spa_name(spa)); (void) zfs_ereport_post(FM_EREPORT_ZFS_IO_FAILURE, spa, NULL, NULL, NULL, 0); mutex_enter(&spa->spa_suspend_lock); if (spa->spa_suspend_zio_root == NULL) spa->spa_suspend_zio_root = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_GODFATHER); spa->spa_suspended = reason; if (zio != NULL) { ASSERT(!(zio->io_flags & ZIO_FLAG_GODFATHER)); ASSERT(zio != spa->spa_suspend_zio_root); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(zio_unique_parent(zio) == NULL); ASSERT(zio->io_stage == ZIO_STAGE_DONE); zio_add_child(spa->spa_suspend_zio_root, zio); } mutex_exit(&spa->spa_suspend_lock); } int zio_resume(spa_t *spa) { zio_t *pio; /* * Reexecute all previously suspended i/o. */ mutex_enter(&spa->spa_suspend_lock); spa->spa_suspended = ZIO_SUSPEND_NONE; cv_broadcast(&spa->spa_suspend_cv); pio = spa->spa_suspend_zio_root; spa->spa_suspend_zio_root = NULL; mutex_exit(&spa->spa_suspend_lock); if (pio == NULL) return (0); zio_reexecute(pio); return (zio_wait(pio)); } void zio_resume_wait(spa_t *spa) { mutex_enter(&spa->spa_suspend_lock); while (spa_suspended(spa)) cv_wait(&spa->spa_suspend_cv, &spa->spa_suspend_lock); mutex_exit(&spa->spa_suspend_lock); } /* * ========================================================================== * Gang blocks. * * A gang block is a collection of small blocks that looks to the DMU * like one large block. When zio_dva_allocate() cannot find a block * of the requested size, due to either severe fragmentation or the pool * being nearly full, it calls zio_write_gang_block() to construct the * block from smaller fragments. * * A gang block consists of a gang header (zio_gbh_phys_t) and up to * three (SPA_GBH_NBLKPTRS) gang members. The gang header is just like * an indirect block: it's an array of block pointers. It consumes * only one sector and hence is allocatable regardless of fragmentation. * The gang header's bps point to its gang members, which hold the data. * * Gang blocks are self-checksumming, using the bp's * as the verifier to ensure uniqueness of the SHA256 checksum. * Critically, the gang block bp's blk_cksum is the checksum of the data, * not the gang header. This ensures that data block signatures (needed for * deduplication) are independent of how the block is physically stored. * * Gang blocks can be nested: a gang member may itself be a gang block. * Thus every gang block is a tree in which root and all interior nodes are * gang headers, and the leaves are normal blocks that contain user data. * The root of the gang tree is called the gang leader. * * To perform any operation (read, rewrite, free, claim) on a gang block, * zio_gang_assemble() first assembles the gang tree (minus data leaves) * in the io_gang_tree field of the original logical i/o by recursively * reading the gang leader and all gang headers below it. This yields * an in-core tree containing the contents of every gang header and the * bps for every constituent of the gang block. * * With the gang tree now assembled, zio_gang_issue() just walks the gang tree * and invokes a callback on each bp. To free a gang block, zio_gang_issue() * calls zio_free_gang() -- a trivial wrapper around zio_free() -- for each bp. * zio_claim_gang() provides a similarly trivial wrapper for zio_claim(). * zio_read_gang() is a wrapper around zio_read() that omits reading gang * headers, since we already have those in io_gang_tree. zio_rewrite_gang() * performs a zio_rewrite() of the data or, for gang headers, a zio_rewrite() * of the gang header plus zio_checksum_compute() of the data to update the * gang header's blk_cksum as described above. * * The two-phase assemble/issue model solves the problem of partial failure -- * what if you'd freed part of a gang block but then couldn't read the * gang header for another part? Assembling the entire gang tree first * ensures that all the necessary gang header I/O has succeeded before * starting the actual work of free, claim, or write. Once the gang tree * is assembled, free and claim are in-memory operations that cannot fail. * * In the event that a gang write fails, zio_dva_unallocate() walks the * gang tree to immediately free (i.e. insert back into the space map) * everything we've allocated. This ensures that we don't get ENOSPC * errors during repeated suspend/resume cycles due to a flaky device. * * Gang rewrites only happen during sync-to-convergence. If we can't assemble * the gang tree, we won't modify the block, so we can safely defer the free * (knowing that the block is still intact). If we *can* assemble the gang * tree, then even if some of the rewrites fail, zio_dva_unallocate() will free * each constituent bp and we can allocate a new block on the next sync pass. * * In all cases, the gang tree allows complete recovery from partial failure. * ========================================================================== */ static void zio_gang_issue_func_done(zio_t *zio) { abd_put(zio->io_abd); } static zio_t * zio_read_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, abd_t *data, uint64_t offset) { if (gn != NULL) return (pio); return (zio_read(pio, pio->io_spa, bp, abd_get_offset(data, offset), BP_GET_PSIZE(bp), zio_gang_issue_func_done, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark)); } static zio_t * zio_rewrite_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, abd_t *data, uint64_t offset) { zio_t *zio; if (gn != NULL) { abd_t *gbh_abd = abd_get_from_buf(gn->gn_gbh, SPA_GANGBLOCKSIZE); zio = zio_rewrite(pio, pio->io_spa, pio->io_txg, bp, gbh_abd, SPA_GANGBLOCKSIZE, zio_gang_issue_func_done, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); /* * As we rewrite each gang header, the pipeline will compute * a new gang block header checksum for it; but no one will * compute a new data checksum, so we do that here. The one * exception is the gang leader: the pipeline already computed * its data checksum because that stage precedes gang assembly. * (Presently, nothing actually uses interior data checksums; * this is just good hygiene.) */ if (gn != pio->io_gang_leader->io_gang_tree) { abd_t *buf = abd_get_offset(data, offset); zio_checksum_compute(zio, BP_GET_CHECKSUM(bp), buf, BP_GET_PSIZE(bp)); abd_put(buf); } /* * If we are here to damage data for testing purposes, * leave the GBH alone so that we can detect the damage. */ if (pio->io_gang_leader->io_flags & ZIO_FLAG_INDUCE_DAMAGE) zio->io_pipeline &= ~ZIO_VDEV_IO_STAGES; } else { zio = zio_rewrite(pio, pio->io_spa, pio->io_txg, bp, abd_get_offset(data, offset), BP_GET_PSIZE(bp), zio_gang_issue_func_done, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); } return (zio); } /* ARGSUSED */ static zio_t * zio_free_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, abd_t *data, uint64_t offset) { zio_t *zio = zio_free_sync(pio, pio->io_spa, pio->io_txg, bp, ZIO_GANG_CHILD_FLAGS(pio)); if (zio == NULL) { zio = zio_null(pio, pio->io_spa, NULL, NULL, NULL, ZIO_GANG_CHILD_FLAGS(pio)); } return (zio); } /* ARGSUSED */ static zio_t * zio_claim_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, abd_t *data, uint64_t offset) { return (zio_claim(pio, pio->io_spa, pio->io_txg, bp, NULL, NULL, ZIO_GANG_CHILD_FLAGS(pio))); } static zio_gang_issue_func_t *zio_gang_issue_func[ZIO_TYPES] = { NULL, zio_read_gang, zio_rewrite_gang, zio_free_gang, zio_claim_gang, NULL }; static void zio_gang_tree_assemble_done(zio_t *zio); static zio_gang_node_t * zio_gang_node_alloc(zio_gang_node_t **gnpp) { zio_gang_node_t *gn; ASSERT(*gnpp == NULL); gn = kmem_zalloc(sizeof (*gn), KM_SLEEP); gn->gn_gbh = zio_buf_alloc(SPA_GANGBLOCKSIZE); *gnpp = gn; return (gn); } static void zio_gang_node_free(zio_gang_node_t **gnpp) { zio_gang_node_t *gn = *gnpp; for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) ASSERT(gn->gn_child[g] == NULL); zio_buf_free(gn->gn_gbh, SPA_GANGBLOCKSIZE); kmem_free(gn, sizeof (*gn)); *gnpp = NULL; } static void zio_gang_tree_free(zio_gang_node_t **gnpp) { zio_gang_node_t *gn = *gnpp; if (gn == NULL) return; for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) zio_gang_tree_free(&gn->gn_child[g]); zio_gang_node_free(gnpp); } static void zio_gang_tree_assemble(zio_t *gio, blkptr_t *bp, zio_gang_node_t **gnpp) { zio_gang_node_t *gn = zio_gang_node_alloc(gnpp); abd_t *gbh_abd = abd_get_from_buf(gn->gn_gbh, SPA_GANGBLOCKSIZE); ASSERT(gio->io_gang_leader == gio); ASSERT(BP_IS_GANG(bp)); zio_nowait(zio_read(gio, gio->io_spa, bp, gbh_abd, SPA_GANGBLOCKSIZE, zio_gang_tree_assemble_done, gn, gio->io_priority, ZIO_GANG_CHILD_FLAGS(gio), &gio->io_bookmark)); } static void zio_gang_tree_assemble_done(zio_t *zio) { zio_t *gio = zio->io_gang_leader; zio_gang_node_t *gn = zio->io_private; blkptr_t *bp = zio->io_bp; ASSERT(gio == zio_unique_parent(zio)); ASSERT(zio->io_child_count == 0); if (zio->io_error) return; /* this ABD was created from a linear buf in zio_gang_tree_assemble */ if (BP_SHOULD_BYTESWAP(bp)) byteswap_uint64_array(abd_to_buf(zio->io_abd), zio->io_size); ASSERT3P(abd_to_buf(zio->io_abd), ==, gn->gn_gbh); ASSERT(zio->io_size == SPA_GANGBLOCKSIZE); ASSERT(gn->gn_gbh->zg_tail.zec_magic == ZEC_MAGIC); abd_put(zio->io_abd); for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) { blkptr_t *gbp = &gn->gn_gbh->zg_blkptr[g]; if (!BP_IS_GANG(gbp)) continue; zio_gang_tree_assemble(gio, gbp, &gn->gn_child[g]); } } static void zio_gang_tree_issue(zio_t *pio, zio_gang_node_t *gn, blkptr_t *bp, abd_t *data, uint64_t offset) { zio_t *gio = pio->io_gang_leader; zio_t *zio; ASSERT(BP_IS_GANG(bp) == !!gn); ASSERT(BP_GET_CHECKSUM(bp) == BP_GET_CHECKSUM(gio->io_bp)); ASSERT(BP_GET_LSIZE(bp) == BP_GET_PSIZE(bp) || gn == gio->io_gang_tree); /* * If you're a gang header, your data is in gn->gn_gbh. * If you're a gang member, your data is in 'data' and gn == NULL. */ zio = zio_gang_issue_func[gio->io_type](pio, bp, gn, data, offset); if (gn != NULL) { ASSERT(gn->gn_gbh->zg_tail.zec_magic == ZEC_MAGIC); for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) { blkptr_t *gbp = &gn->gn_gbh->zg_blkptr[g]; if (BP_IS_HOLE(gbp)) continue; zio_gang_tree_issue(zio, gn->gn_child[g], gbp, data, offset); offset += BP_GET_PSIZE(gbp); } } if (gn == gio->io_gang_tree) ASSERT3U(gio->io_size, ==, offset); if (zio != pio) zio_nowait(zio); } static zio_t * zio_gang_assemble(zio_t *zio) { blkptr_t *bp = zio->io_bp; ASSERT(BP_IS_GANG(bp) && zio->io_gang_leader == NULL); ASSERT(zio->io_child_type > ZIO_CHILD_GANG); zio->io_gang_leader = zio; zio_gang_tree_assemble(zio, bp, &zio->io_gang_tree); return (zio); } static zio_t * zio_gang_issue(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (zio_wait_for_children(zio, ZIO_CHILD_GANG_BIT, ZIO_WAIT_DONE)) { return (NULL); } ASSERT(BP_IS_GANG(bp) && zio->io_gang_leader == zio); ASSERT(zio->io_child_type > ZIO_CHILD_GANG); if (zio->io_child_error[ZIO_CHILD_GANG] == 0) zio_gang_tree_issue(zio, zio->io_gang_tree, bp, zio->io_abd, 0); else zio_gang_tree_free(&zio->io_gang_tree); zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; return (zio); } static void zio_write_gang_member_ready(zio_t *zio) { zio_t *pio = zio_unique_parent(zio); dva_t *cdva = zio->io_bp->blk_dva; dva_t *pdva = pio->io_bp->blk_dva; uint64_t asize; zio_t *gio __maybe_unused = zio->io_gang_leader; if (BP_IS_HOLE(zio->io_bp)) return; ASSERT(BP_IS_HOLE(&zio->io_bp_orig)); ASSERT(zio->io_child_type == ZIO_CHILD_GANG); ASSERT3U(zio->io_prop.zp_copies, ==, gio->io_prop.zp_copies); ASSERT3U(zio->io_prop.zp_copies, <=, BP_GET_NDVAS(zio->io_bp)); ASSERT3U(pio->io_prop.zp_copies, <=, BP_GET_NDVAS(pio->io_bp)); ASSERT3U(BP_GET_NDVAS(zio->io_bp), <=, BP_GET_NDVAS(pio->io_bp)); mutex_enter(&pio->io_lock); for (int d = 0; d < BP_GET_NDVAS(zio->io_bp); d++) { ASSERT(DVA_GET_GANG(&pdva[d])); asize = DVA_GET_ASIZE(&pdva[d]); asize += DVA_GET_ASIZE(&cdva[d]); DVA_SET_ASIZE(&pdva[d], asize); } mutex_exit(&pio->io_lock); } static void zio_write_gang_done(zio_t *zio) { /* * The io_abd field will be NULL for a zio with no data. The io_flags * will initially have the ZIO_FLAG_NODATA bit flag set, but we can't * check for it here as it is cleared in zio_ready. */ if (zio->io_abd != NULL) abd_put(zio->io_abd); } static zio_t * zio_write_gang_block(zio_t *pio) { spa_t *spa = pio->io_spa; metaslab_class_t *mc = spa_normal_class(spa); blkptr_t *bp = pio->io_bp; zio_t *gio = pio->io_gang_leader; zio_t *zio; zio_gang_node_t *gn, **gnpp; zio_gbh_phys_t *gbh; abd_t *gbh_abd; uint64_t txg = pio->io_txg; uint64_t resid = pio->io_size; uint64_t lsize; int copies = gio->io_prop.zp_copies; int gbh_copies; zio_prop_t zp; int error; boolean_t has_data = !(pio->io_flags & ZIO_FLAG_NODATA); /* * encrypted blocks need DVA[2] free so encrypted gang headers can't * have a third copy. */ gbh_copies = MIN(copies + 1, spa_max_replication(spa)); if (gio->io_prop.zp_encrypt && gbh_copies >= SPA_DVAS_PER_BP) gbh_copies = SPA_DVAS_PER_BP - 1; int flags = METASLAB_HINTBP_FAVOR | METASLAB_GANG_HEADER; if (pio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(pio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(has_data); flags |= METASLAB_ASYNC_ALLOC; VERIFY(zfs_refcount_held(&mc->mc_alloc_slots[pio->io_allocator], pio)); /* * The logical zio has already placed a reservation for * 'copies' allocation slots but gang blocks may require * additional copies. These additional copies * (i.e. gbh_copies - copies) are guaranteed to succeed * since metaslab_class_throttle_reserve() always allows * additional reservations for gang blocks. */ VERIFY(metaslab_class_throttle_reserve(mc, gbh_copies - copies, pio->io_allocator, pio, flags)); } error = metaslab_alloc(spa, mc, SPA_GANGBLOCKSIZE, bp, gbh_copies, txg, pio == gio ? NULL : gio->io_bp, flags, &pio->io_alloc_list, pio, pio->io_allocator); if (error) { if (pio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(pio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(has_data); /* * If we failed to allocate the gang block header then * we remove any additional allocation reservations that * we placed here. The original reservation will * be removed when the logical I/O goes to the ready * stage. */ metaslab_class_throttle_unreserve(mc, gbh_copies - copies, pio->io_allocator, pio); } pio->io_error = error; return (pio); } if (pio == gio) { gnpp = &gio->io_gang_tree; } else { gnpp = pio->io_private; ASSERT(pio->io_ready == zio_write_gang_member_ready); } gn = zio_gang_node_alloc(gnpp); gbh = gn->gn_gbh; bzero(gbh, SPA_GANGBLOCKSIZE); gbh_abd = abd_get_from_buf(gbh, SPA_GANGBLOCKSIZE); /* * Create the gang header. */ zio = zio_rewrite(pio, spa, txg, bp, gbh_abd, SPA_GANGBLOCKSIZE, zio_write_gang_done, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); /* * Create and nowait the gang children. */ for (int g = 0; resid != 0; resid -= lsize, g++) { lsize = P2ROUNDUP(resid / (SPA_GBH_NBLKPTRS - g), SPA_MINBLOCKSIZE); ASSERT(lsize >= SPA_MINBLOCKSIZE && lsize <= resid); zp.zp_checksum = gio->io_prop.zp_checksum; zp.zp_compress = ZIO_COMPRESS_OFF; zp.zp_complevel = gio->io_prop.zp_complevel; zp.zp_type = DMU_OT_NONE; zp.zp_level = 0; zp.zp_copies = gio->io_prop.zp_copies; zp.zp_dedup = B_FALSE; zp.zp_dedup_verify = B_FALSE; zp.zp_nopwrite = B_FALSE; zp.zp_encrypt = gio->io_prop.zp_encrypt; zp.zp_byteorder = gio->io_prop.zp_byteorder; bzero(zp.zp_salt, ZIO_DATA_SALT_LEN); bzero(zp.zp_iv, ZIO_DATA_IV_LEN); bzero(zp.zp_mac, ZIO_DATA_MAC_LEN); zio_t *cio = zio_write(zio, spa, txg, &gbh->zg_blkptr[g], has_data ? abd_get_offset(pio->io_abd, pio->io_size - resid) : NULL, lsize, lsize, &zp, zio_write_gang_member_ready, NULL, NULL, zio_write_gang_done, &gn->gn_child[g], pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); if (pio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(pio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(has_data); /* * Gang children won't throttle but we should * account for their work, so reserve an allocation * slot for them here. */ VERIFY(metaslab_class_throttle_reserve(mc, zp.zp_copies, cio->io_allocator, cio, flags)); } zio_nowait(cio); } /* * Set pio's pipeline to just wait for zio to finish. */ pio->io_pipeline = ZIO_INTERLOCK_PIPELINE; /* * We didn't allocate this bp, so make sure it doesn't get unmarked. */ pio->io_flags &= ~ZIO_FLAG_FASTWRITE; zio_nowait(zio); return (pio); } /* * The zio_nop_write stage in the pipeline determines if allocating a * new bp is necessary. The nopwrite feature can handle writes in * either syncing or open context (i.e. zil writes) and as a result is * mutually exclusive with dedup. * * By leveraging a cryptographically secure checksum, such as SHA256, we * can compare the checksums of the new data and the old to determine if * allocating a new block is required. Note that our requirements for * cryptographic strength are fairly weak: there can't be any accidental * hash collisions, but we don't need to be secure against intentional * (malicious) collisions. To trigger a nopwrite, you have to be able * to write the file to begin with, and triggering an incorrect (hash * collision) nopwrite is no worse than simply writing to the file. * That said, there are no known attacks against the checksum algorithms * used for nopwrite, assuming that the salt and the checksums * themselves remain secret. */ static zio_t * zio_nop_write(zio_t *zio) { blkptr_t *bp = zio->io_bp; blkptr_t *bp_orig = &zio->io_bp_orig; zio_prop_t *zp = &zio->io_prop; ASSERT(BP_GET_LEVEL(bp) == 0); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); ASSERT(zp->zp_nopwrite); ASSERT(!zp->zp_dedup); ASSERT(zio->io_bp_override == NULL); ASSERT(IO_IS_ALLOCATING(zio)); /* * Check to see if the original bp and the new bp have matching * characteristics (i.e. same checksum, compression algorithms, etc). * If they don't then just continue with the pipeline which will * allocate a new bp. */ if (BP_IS_HOLE(bp_orig) || !(zio_checksum_table[BP_GET_CHECKSUM(bp)].ci_flags & ZCHECKSUM_FLAG_NOPWRITE) || BP_IS_ENCRYPTED(bp) || BP_IS_ENCRYPTED(bp_orig) || BP_GET_CHECKSUM(bp) != BP_GET_CHECKSUM(bp_orig) || BP_GET_COMPRESS(bp) != BP_GET_COMPRESS(bp_orig) || BP_GET_DEDUP(bp) != BP_GET_DEDUP(bp_orig) || zp->zp_copies != BP_GET_NDVAS(bp_orig)) return (zio); /* * If the checksums match then reset the pipeline so that we * avoid allocating a new bp and issuing any I/O. */ if (ZIO_CHECKSUM_EQUAL(bp->blk_cksum, bp_orig->blk_cksum)) { ASSERT(zio_checksum_table[zp->zp_checksum].ci_flags & ZCHECKSUM_FLAG_NOPWRITE); ASSERT3U(BP_GET_PSIZE(bp), ==, BP_GET_PSIZE(bp_orig)); ASSERT3U(BP_GET_LSIZE(bp), ==, BP_GET_LSIZE(bp_orig)); ASSERT(zp->zp_compress != ZIO_COMPRESS_OFF); ASSERT(bcmp(&bp->blk_prop, &bp_orig->blk_prop, sizeof (uint64_t)) == 0); /* * If we're overwriting a block that is currently on an * indirect vdev, then ignore the nopwrite request and * allow a new block to be allocated on a concrete vdev. */ spa_config_enter(zio->io_spa, SCL_VDEV, FTAG, RW_READER); vdev_t *tvd = vdev_lookup_top(zio->io_spa, DVA_GET_VDEV(&bp->blk_dva[0])); if (tvd->vdev_ops == &vdev_indirect_ops) { spa_config_exit(zio->io_spa, SCL_VDEV, FTAG); return (zio); } spa_config_exit(zio->io_spa, SCL_VDEV, FTAG); *bp = *bp_orig; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; zio->io_flags |= ZIO_FLAG_NOPWRITE; } return (zio); } /* * ========================================================================== * Dedup * ========================================================================== */ static void zio_ddt_child_read_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp; zio_t *pio = zio_unique_parent(zio); mutex_enter(&pio->io_lock); ddp = ddt_phys_select(dde, bp); if (zio->io_error == 0) ddt_phys_clear(ddp); /* this ddp doesn't need repair */ if (zio->io_error == 0 && dde->dde_repair_abd == NULL) dde->dde_repair_abd = zio->io_abd; else abd_free(zio->io_abd); mutex_exit(&pio->io_lock); } static zio_t * zio_ddt_read_start(zio_t *zio) { blkptr_t *bp = zio->io_bp; ASSERT(BP_GET_DEDUP(bp)); ASSERT(BP_GET_PSIZE(bp) == zio->io_size); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); if (zio->io_child_error[ZIO_CHILD_DDT]) { ddt_t *ddt = ddt_select(zio->io_spa, bp); ddt_entry_t *dde = ddt_repair_start(ddt, bp); ddt_phys_t *ddp = dde->dde_phys; ddt_phys_t *ddp_self = ddt_phys_select(dde, bp); blkptr_t blk; ASSERT(zio->io_vsd == NULL); zio->io_vsd = dde; if (ddp_self == NULL) return (zio); for (int p = 0; p < DDT_PHYS_TYPES; p++, ddp++) { if (ddp->ddp_phys_birth == 0 || ddp == ddp_self) continue; ddt_bp_create(ddt->ddt_checksum, &dde->dde_key, ddp, &blk); zio_nowait(zio_read(zio, zio->io_spa, &blk, abd_alloc_for_io(zio->io_size, B_TRUE), zio->io_size, zio_ddt_child_read_done, dde, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio) | ZIO_FLAG_DONT_PROPAGATE, &zio->io_bookmark)); } return (zio); } zio_nowait(zio_read(zio, zio->io_spa, bp, zio->io_abd, zio->io_size, NULL, NULL, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark)); return (zio); } static zio_t * zio_ddt_read_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (zio_wait_for_children(zio, ZIO_CHILD_DDT_BIT, ZIO_WAIT_DONE)) { return (NULL); } ASSERT(BP_GET_DEDUP(bp)); ASSERT(BP_GET_PSIZE(bp) == zio->io_size); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); if (zio->io_child_error[ZIO_CHILD_DDT]) { ddt_t *ddt = ddt_select(zio->io_spa, bp); ddt_entry_t *dde = zio->io_vsd; if (ddt == NULL) { ASSERT(spa_load_state(zio->io_spa) != SPA_LOAD_NONE); return (zio); } if (dde == NULL) { zio->io_stage = ZIO_STAGE_DDT_READ_START >> 1; zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_FALSE); return (NULL); } if (dde->dde_repair_abd != NULL) { abd_copy(zio->io_abd, dde->dde_repair_abd, zio->io_size); zio->io_child_error[ZIO_CHILD_DDT] = 0; } ddt_repair_done(ddt, dde); zio->io_vsd = NULL; } ASSERT(zio->io_vsd == NULL); return (zio); } static boolean_t zio_ddt_collision(zio_t *zio, ddt_t *ddt, ddt_entry_t *dde) { spa_t *spa = zio->io_spa; boolean_t do_raw = !!(zio->io_flags & ZIO_FLAG_RAW); ASSERT(!(zio->io_bp_override && do_raw)); /* * Note: we compare the original data, not the transformed data, * because when zio->io_bp is an override bp, we will not have * pushed the I/O transforms. That's an important optimization * because otherwise we'd compress/encrypt all dmu_sync() data twice. * However, we should never get a raw, override zio so in these * cases we can compare the io_abd directly. This is useful because * it allows us to do dedup verification even if we don't have access * to the original data (for instance, if the encryption keys aren't * loaded). */ for (int p = DDT_PHYS_SINGLE; p <= DDT_PHYS_TRIPLE; p++) { zio_t *lio = dde->dde_lead_zio[p]; if (lio != NULL && do_raw) { return (lio->io_size != zio->io_size || abd_cmp(zio->io_abd, lio->io_abd) != 0); } else if (lio != NULL) { return (lio->io_orig_size != zio->io_orig_size || abd_cmp(zio->io_orig_abd, lio->io_orig_abd) != 0); } } for (int p = DDT_PHYS_SINGLE; p <= DDT_PHYS_TRIPLE; p++) { ddt_phys_t *ddp = &dde->dde_phys[p]; if (ddp->ddp_phys_birth != 0 && do_raw) { blkptr_t blk = *zio->io_bp; uint64_t psize; abd_t *tmpabd; int error; ddt_bp_fill(ddp, &blk, ddp->ddp_phys_birth); psize = BP_GET_PSIZE(&blk); if (psize != zio->io_size) return (B_TRUE); ddt_exit(ddt); tmpabd = abd_alloc_for_io(psize, B_TRUE); error = zio_wait(zio_read(NULL, spa, &blk, tmpabd, psize, NULL, NULL, ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_RAW, &zio->io_bookmark)); if (error == 0) { if (abd_cmp(tmpabd, zio->io_abd) != 0) error = SET_ERROR(ENOENT); } abd_free(tmpabd); ddt_enter(ddt); return (error != 0); } else if (ddp->ddp_phys_birth != 0) { arc_buf_t *abuf = NULL; arc_flags_t aflags = ARC_FLAG_WAIT; blkptr_t blk = *zio->io_bp; int error; ddt_bp_fill(ddp, &blk, ddp->ddp_phys_birth); if (BP_GET_LSIZE(&blk) != zio->io_orig_size) return (B_TRUE); ddt_exit(ddt); error = arc_read(NULL, spa, &blk, arc_getbuf_func, &abuf, ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &aflags, &zio->io_bookmark); if (error == 0) { if (abd_cmp_buf(zio->io_orig_abd, abuf->b_data, zio->io_orig_size) != 0) error = SET_ERROR(ENOENT); arc_buf_destroy(abuf, &abuf); } ddt_enter(ddt); return (error != 0); } } return (B_FALSE); } static void zio_ddt_child_write_ready(zio_t *zio) { int p = zio->io_prop.zp_copies; ddt_t *ddt = ddt_select(zio->io_spa, zio->io_bp); ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp = &dde->dde_phys[p]; zio_t *pio; if (zio->io_error) return; ddt_enter(ddt); ASSERT(dde->dde_lead_zio[p] == zio); ddt_phys_fill(ddp, zio->io_bp); zio_link_t *zl = NULL; while ((pio = zio_walk_parents(zio, &zl)) != NULL) ddt_bp_fill(ddp, pio->io_bp, zio->io_txg); ddt_exit(ddt); } static void zio_ddt_child_write_done(zio_t *zio) { int p = zio->io_prop.zp_copies; ddt_t *ddt = ddt_select(zio->io_spa, zio->io_bp); ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp = &dde->dde_phys[p]; ddt_enter(ddt); ASSERT(ddp->ddp_refcnt == 0); ASSERT(dde->dde_lead_zio[p] == zio); dde->dde_lead_zio[p] = NULL; if (zio->io_error == 0) { zio_link_t *zl = NULL; while (zio_walk_parents(zio, &zl) != NULL) ddt_phys_addref(ddp); } else { ddt_phys_clear(ddp); } ddt_exit(ddt); } static zio_t * zio_ddt_write(zio_t *zio) { spa_t *spa = zio->io_spa; blkptr_t *bp = zio->io_bp; uint64_t txg = zio->io_txg; zio_prop_t *zp = &zio->io_prop; int p = zp->zp_copies; zio_t *cio = NULL; ddt_t *ddt = ddt_select(spa, bp); ddt_entry_t *dde; ddt_phys_t *ddp; ASSERT(BP_GET_DEDUP(bp)); ASSERT(BP_GET_CHECKSUM(bp) == zp->zp_checksum); ASSERT(BP_IS_HOLE(bp) || zio->io_bp_override); ASSERT(!(zio->io_bp_override && (zio->io_flags & ZIO_FLAG_RAW))); ddt_enter(ddt); dde = ddt_lookup(ddt, bp, B_TRUE); ddp = &dde->dde_phys[p]; if (zp->zp_dedup_verify && zio_ddt_collision(zio, ddt, dde)) { /* * If we're using a weak checksum, upgrade to a strong checksum * and try again. If we're already using a strong checksum, * we can't resolve it, so just convert to an ordinary write. * (And automatically e-mail a paper to Nature?) */ if (!(zio_checksum_table[zp->zp_checksum].ci_flags & ZCHECKSUM_FLAG_DEDUP)) { zp->zp_checksum = spa_dedup_checksum(spa); zio_pop_transforms(zio); zio->io_stage = ZIO_STAGE_OPEN; BP_ZERO(bp); } else { zp->zp_dedup = B_FALSE; BP_SET_DEDUP(bp, B_FALSE); } ASSERT(!BP_GET_DEDUP(bp)); zio->io_pipeline = ZIO_WRITE_PIPELINE; ddt_exit(ddt); return (zio); } if (ddp->ddp_phys_birth != 0 || dde->dde_lead_zio[p] != NULL) { if (ddp->ddp_phys_birth != 0) ddt_bp_fill(ddp, bp, txg); if (dde->dde_lead_zio[p] != NULL) zio_add_child(zio, dde->dde_lead_zio[p]); else ddt_phys_addref(ddp); } else if (zio->io_bp_override) { ASSERT(bp->blk_birth == txg); ASSERT(BP_EQUAL(bp, zio->io_bp_override)); ddt_phys_fill(ddp, bp); ddt_phys_addref(ddp); } else { cio = zio_write(zio, spa, txg, bp, zio->io_orig_abd, zio->io_orig_size, zio->io_orig_size, zp, zio_ddt_child_write_ready, NULL, NULL, zio_ddt_child_write_done, dde, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark); zio_push_transform(cio, zio->io_abd, zio->io_size, 0, NULL); dde->dde_lead_zio[p] = cio; } ddt_exit(ddt); zio_nowait(cio); return (zio); } ddt_entry_t *freedde; /* for debugging */ static zio_t * zio_ddt_free(zio_t *zio) { spa_t *spa = zio->io_spa; blkptr_t *bp = zio->io_bp; ddt_t *ddt = ddt_select(spa, bp); ddt_entry_t *dde; ddt_phys_t *ddp; ASSERT(BP_GET_DEDUP(bp)); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ddt_enter(ddt); freedde = dde = ddt_lookup(ddt, bp, B_TRUE); if (dde) { ddp = ddt_phys_select(dde, bp); if (ddp) ddt_phys_decref(ddp); } ddt_exit(ddt); return (zio); } /* * ========================================================================== * Allocate and free blocks * ========================================================================== */ static zio_t * zio_io_to_allocate(spa_t *spa, int allocator) { zio_t *zio; ASSERT(MUTEX_HELD(&spa->spa_alloc_locks[allocator])); zio = avl_first(&spa->spa_alloc_trees[allocator]); if (zio == NULL) return (NULL); ASSERT(IO_IS_ALLOCATING(zio)); /* * Try to place a reservation for this zio. If we're unable to * reserve then we throttle. */ ASSERT3U(zio->io_allocator, ==, allocator); if (!metaslab_class_throttle_reserve(zio->io_metaslab_class, zio->io_prop.zp_copies, zio->io_allocator, zio, 0)) { return (NULL); } avl_remove(&spa->spa_alloc_trees[allocator], zio); ASSERT3U(zio->io_stage, <, ZIO_STAGE_DVA_ALLOCATE); return (zio); } static zio_t * zio_dva_throttle(zio_t *zio) { spa_t *spa = zio->io_spa; zio_t *nio; metaslab_class_t *mc; /* locate an appropriate allocation class */ mc = spa_preferred_class(spa, zio->io_size, zio->io_prop.zp_type, zio->io_prop.zp_level, zio->io_prop.zp_zpl_smallblk); if (zio->io_priority == ZIO_PRIORITY_SYNC_WRITE || !mc->mc_alloc_throttle_enabled || zio->io_child_type == ZIO_CHILD_GANG || zio->io_flags & ZIO_FLAG_NODATA) { return (zio); } ASSERT(zio->io_child_type > ZIO_CHILD_GANG); ASSERT3U(zio->io_queued_timestamp, >, 0); ASSERT(zio->io_stage == ZIO_STAGE_DVA_THROTTLE); zbookmark_phys_t *bm = &zio->io_bookmark; /* * We want to try to use as many allocators as possible to help improve * performance, but we also want logically adjacent IOs to be physically * adjacent to improve sequential read performance. We chunk each object * into 2^20 block regions, and then hash based on the objset, object, * level, and region to accomplish both of these goals. */ zio->io_allocator = cityhash4(bm->zb_objset, bm->zb_object, bm->zb_level, bm->zb_blkid >> 20) % spa->spa_alloc_count; mutex_enter(&spa->spa_alloc_locks[zio->io_allocator]); ASSERT(zio->io_type == ZIO_TYPE_WRITE); zio->io_metaslab_class = mc; avl_add(&spa->spa_alloc_trees[zio->io_allocator], zio); nio = zio_io_to_allocate(spa, zio->io_allocator); mutex_exit(&spa->spa_alloc_locks[zio->io_allocator]); return (nio); } static void zio_allocate_dispatch(spa_t *spa, int allocator) { zio_t *zio; mutex_enter(&spa->spa_alloc_locks[allocator]); zio = zio_io_to_allocate(spa, allocator); mutex_exit(&spa->spa_alloc_locks[allocator]); if (zio == NULL) return; ASSERT3U(zio->io_stage, ==, ZIO_STAGE_DVA_THROTTLE); ASSERT0(zio->io_error); zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_TRUE); } static zio_t * zio_dva_allocate(zio_t *zio) { spa_t *spa = zio->io_spa; metaslab_class_t *mc; blkptr_t *bp = zio->io_bp; int error; int flags = 0; if (zio->io_gang_leader == NULL) { ASSERT(zio->io_child_type > ZIO_CHILD_GANG); zio->io_gang_leader = zio; } ASSERT(BP_IS_HOLE(bp)); ASSERT0(BP_GET_NDVAS(bp)); ASSERT3U(zio->io_prop.zp_copies, >, 0); ASSERT3U(zio->io_prop.zp_copies, <=, spa_max_replication(spa)); ASSERT3U(zio->io_size, ==, BP_GET_PSIZE(bp)); flags |= (zio->io_flags & ZIO_FLAG_FASTWRITE) ? METASLAB_FASTWRITE : 0; if (zio->io_flags & ZIO_FLAG_NODATA) flags |= METASLAB_DONT_THROTTLE; if (zio->io_flags & ZIO_FLAG_GANG_CHILD) flags |= METASLAB_GANG_CHILD; if (zio->io_priority == ZIO_PRIORITY_ASYNC_WRITE) flags |= METASLAB_ASYNC_ALLOC; /* * if not already chosen, locate an appropriate allocation class */ mc = zio->io_metaslab_class; if (mc == NULL) { mc = spa_preferred_class(spa, zio->io_size, zio->io_prop.zp_type, zio->io_prop.zp_level, zio->io_prop.zp_zpl_smallblk); zio->io_metaslab_class = mc; } error = metaslab_alloc(spa, mc, zio->io_size, bp, zio->io_prop.zp_copies, zio->io_txg, NULL, flags, &zio->io_alloc_list, zio, zio->io_allocator); /* * Fallback to normal class when an alloc class is full */ if (error == ENOSPC && mc != spa_normal_class(spa)) { /* * If throttling, transfer reservation over to normal class. * The io_allocator slot can remain the same even though we * are switching classes. */ if (mc->mc_alloc_throttle_enabled && (zio->io_flags & ZIO_FLAG_IO_ALLOCATING)) { metaslab_class_throttle_unreserve(mc, zio->io_prop.zp_copies, zio->io_allocator, zio); zio->io_flags &= ~ZIO_FLAG_IO_ALLOCATING; mc = spa_normal_class(spa); VERIFY(metaslab_class_throttle_reserve(mc, zio->io_prop.zp_copies, zio->io_allocator, zio, flags | METASLAB_MUST_RESERVE)); } else { mc = spa_normal_class(spa); } zio->io_metaslab_class = mc; error = metaslab_alloc(spa, mc, zio->io_size, bp, zio->io_prop.zp_copies, zio->io_txg, NULL, flags, &zio->io_alloc_list, zio, zio->io_allocator); } if (error != 0) { zfs_dbgmsg("%s: metaslab allocation failure: zio %px, " "size %llu, error %d", spa_name(spa), zio, zio->io_size, error); if (error == ENOSPC && zio->io_size > SPA_MINBLOCKSIZE) return (zio_write_gang_block(zio)); zio->io_error = error; } return (zio); } static zio_t * zio_dva_free(zio_t *zio) { metaslab_free(zio->io_spa, zio->io_bp, zio->io_txg, B_FALSE); return (zio); } static zio_t * zio_dva_claim(zio_t *zio) { int error; error = metaslab_claim(zio->io_spa, zio->io_bp, zio->io_txg); if (error) zio->io_error = error; return (zio); } /* * Undo an allocation. This is used by zio_done() when an I/O fails * and we want to give back the block we just allocated. * This handles both normal blocks and gang blocks. */ static void zio_dva_unallocate(zio_t *zio, zio_gang_node_t *gn, blkptr_t *bp) { ASSERT(bp->blk_birth == zio->io_txg || BP_IS_HOLE(bp)); ASSERT(zio->io_bp_override == NULL); if (!BP_IS_HOLE(bp)) metaslab_free(zio->io_spa, bp, bp->blk_birth, B_TRUE); if (gn != NULL) { for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) { zio_dva_unallocate(zio, gn->gn_child[g], &gn->gn_gbh->zg_blkptr[g]); } } } /* * Try to allocate an intent log block. Return 0 on success, errno on failure. */ int zio_alloc_zil(spa_t *spa, objset_t *os, uint64_t txg, blkptr_t *new_bp, uint64_t size, boolean_t *slog) { int error = 1; zio_alloc_list_t io_alloc_list; ASSERT(txg > spa_syncing_txg(spa)); metaslab_trace_init(&io_alloc_list); /* * Block pointer fields are useful to metaslabs for stats and debugging. * Fill in the obvious ones before calling into metaslab_alloc(). */ BP_SET_TYPE(new_bp, DMU_OT_INTENT_LOG); BP_SET_PSIZE(new_bp, size); BP_SET_LEVEL(new_bp, 0); /* * When allocating a zil block, we don't have information about * the final destination of the block except the objset it's part * of, so we just hash the objset ID to pick the allocator to get * some parallelism. */ error = metaslab_alloc(spa, spa_log_class(spa), size, new_bp, 1, txg, NULL, METASLAB_FASTWRITE, &io_alloc_list, NULL, cityhash4(0, 0, 0, os->os_dsl_dataset->ds_object) % spa->spa_alloc_count); if (error == 0) { *slog = TRUE; } else { error = metaslab_alloc(spa, spa_normal_class(spa), size, new_bp, 1, txg, NULL, METASLAB_FASTWRITE, &io_alloc_list, NULL, cityhash4(0, 0, 0, os->os_dsl_dataset->ds_object) % spa->spa_alloc_count); if (error == 0) *slog = FALSE; } metaslab_trace_fini(&io_alloc_list); if (error == 0) { BP_SET_LSIZE(new_bp, size); BP_SET_PSIZE(new_bp, size); BP_SET_COMPRESS(new_bp, ZIO_COMPRESS_OFF); BP_SET_CHECKSUM(new_bp, spa_version(spa) >= SPA_VERSION_SLIM_ZIL ? ZIO_CHECKSUM_ZILOG2 : ZIO_CHECKSUM_ZILOG); BP_SET_TYPE(new_bp, DMU_OT_INTENT_LOG); BP_SET_LEVEL(new_bp, 0); BP_SET_DEDUP(new_bp, 0); BP_SET_BYTEORDER(new_bp, ZFS_HOST_BYTEORDER); /* * encrypted blocks will require an IV and salt. We generate * these now since we will not be rewriting the bp at * rewrite time. */ if (os->os_encrypted) { uint8_t iv[ZIO_DATA_IV_LEN]; uint8_t salt[ZIO_DATA_SALT_LEN]; BP_SET_CRYPT(new_bp, B_TRUE); VERIFY0(spa_crypt_get_salt(spa, dmu_objset_id(os), salt)); VERIFY0(zio_crypt_generate_iv(iv)); zio_crypt_encode_params_bp(new_bp, salt, iv); } } else { zfs_dbgmsg("%s: zil block allocation failure: " "size %llu, error %d", spa_name(spa), size, error); } return (error); } /* * ========================================================================== * Read and write to physical devices * ========================================================================== */ /* * Issue an I/O to the underlying vdev. Typically the issue pipeline * stops after this stage and will resume upon I/O completion. * However, there are instances where the vdev layer may need to * continue the pipeline when an I/O was not issued. Since the I/O * that was sent to the vdev layer might be different than the one * currently active in the pipeline (see vdev_queue_io()), we explicitly * force the underlying vdev layers to call either zio_execute() or * zio_interrupt() to ensure that the pipeline continues with the correct I/O. */ static zio_t * zio_vdev_io_start(zio_t *zio) { vdev_t *vd = zio->io_vd; uint64_t align; spa_t *spa = zio->io_spa; zio->io_delay = 0; ASSERT(zio->io_error == 0); ASSERT(zio->io_child_error[ZIO_CHILD_VDEV] == 0); if (vd == NULL) { if (!(zio->io_flags & ZIO_FLAG_CONFIG_WRITER)) spa_config_enter(spa, SCL_ZIO, zio, RW_READER); /* * The mirror_ops handle multiple DVAs in a single BP. */ vdev_mirror_ops.vdev_op_io_start(zio); return (NULL); } ASSERT3P(zio->io_logical, !=, zio); if (zio->io_type == ZIO_TYPE_WRITE) { ASSERT(spa->spa_trust_config); /* * Note: the code can handle other kinds of writes, * but we don't expect them. */ if (zio->io_vd->vdev_removing) { ASSERT(zio->io_flags & (ZIO_FLAG_PHYSICAL | ZIO_FLAG_SELF_HEAL | ZIO_FLAG_RESILVER | ZIO_FLAG_INDUCE_DAMAGE)); } } align = 1ULL << vd->vdev_top->vdev_ashift; if (!(zio->io_flags & ZIO_FLAG_PHYSICAL) && P2PHASE(zio->io_size, align) != 0) { /* Transform logical writes to be a full physical block size. */ uint64_t asize = P2ROUNDUP(zio->io_size, align); abd_t *abuf = abd_alloc_sametype(zio->io_abd, asize); ASSERT(vd == vd->vdev_top); if (zio->io_type == ZIO_TYPE_WRITE) { abd_copy(abuf, zio->io_abd, zio->io_size); abd_zero_off(abuf, zio->io_size, asize - zio->io_size); } zio_push_transform(zio, abuf, asize, asize, zio_subblock); } /* * If this is not a physical io, make sure that it is properly aligned * before proceeding. */ if (!(zio->io_flags & ZIO_FLAG_PHYSICAL)) { ASSERT0(P2PHASE(zio->io_offset, align)); ASSERT0(P2PHASE(zio->io_size, align)); } else { /* * For physical writes, we allow 512b aligned writes and assume * the device will perform a read-modify-write as necessary. */ ASSERT0(P2PHASE(zio->io_offset, SPA_MINBLOCKSIZE)); ASSERT0(P2PHASE(zio->io_size, SPA_MINBLOCKSIZE)); } VERIFY(zio->io_type != ZIO_TYPE_WRITE || spa_writeable(spa)); /* * If this is a repair I/O, and there's no self-healing involved -- * that is, we're just resilvering what we expect to resilver -- * then don't do the I/O unless zio's txg is actually in vd's DTL. * This prevents spurious resilvering. * * There are a few ways that we can end up creating these spurious * resilver i/os: * * 1. A resilver i/o will be issued if any DVA in the BP has a * dirty DTL. The mirror code will issue resilver writes to * each DVA, including the one(s) that are not on vdevs with dirty * DTLs. * * 2. With nested replication, which happens when we have a * "replacing" or "spare" vdev that's a child of a mirror or raidz. * For example, given mirror(replacing(A+B), C), it's likely that * only A is out of date (it's the new device). In this case, we'll * read from C, then use the data to resilver A+B -- but we don't * actually want to resilver B, just A. The top-level mirror has no * way to know this, so instead we just discard unnecessary repairs * as we work our way down the vdev tree. * * 3. ZTEST also creates mirrors of mirrors, mirrors of raidz, etc. * The same logic applies to any form of nested replication: ditto * + mirror, RAID-Z + replacing, etc. * * However, indirect vdevs point off to other vdevs which may have * DTL's, so we never bypass them. The child i/os on concrete vdevs * will be properly bypassed instead. */ if ((zio->io_flags & ZIO_FLAG_IO_REPAIR) && !(zio->io_flags & ZIO_FLAG_SELF_HEAL) && zio->io_txg != 0 && /* not a delegated i/o */ vd->vdev_ops != &vdev_indirect_ops && !vdev_dtl_contains(vd, DTL_PARTIAL, zio->io_txg, 1)) { ASSERT(zio->io_type == ZIO_TYPE_WRITE); zio_vdev_io_bypass(zio); return (zio); } if (vd->vdev_ops->vdev_op_leaf && (zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE || zio->io_type == ZIO_TYPE_TRIM)) { if (zio->io_type == ZIO_TYPE_READ && vdev_cache_read(zio)) return (zio); if ((zio = vdev_queue_io(zio)) == NULL) return (NULL); if (!vdev_accessible(vd, zio)) { zio->io_error = SET_ERROR(ENXIO); zio_interrupt(zio); return (NULL); } zio->io_delay = gethrtime(); } vd->vdev_ops->vdev_op_io_start(zio); return (NULL); } static zio_t * zio_vdev_io_done(zio_t *zio) { vdev_t *vd = zio->io_vd; vdev_ops_t *ops = vd ? vd->vdev_ops : &vdev_mirror_ops; boolean_t unexpected_error = B_FALSE; if (zio_wait_for_children(zio, ZIO_CHILD_VDEV_BIT, ZIO_WAIT_DONE)) { return (NULL); } ASSERT(zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE || zio->io_type == ZIO_TYPE_TRIM); if (zio->io_delay) zio->io_delay = gethrtime() - zio->io_delay; if (vd != NULL && vd->vdev_ops->vdev_op_leaf) { vdev_queue_io_done(zio); if (zio->io_type == ZIO_TYPE_WRITE) vdev_cache_write(zio); if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_device_injections(vd, zio, EIO, EILSEQ); if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_label_injection(zio, EIO); if (zio->io_error && zio->io_type != ZIO_TYPE_TRIM) { if (!vdev_accessible(vd, zio)) { zio->io_error = SET_ERROR(ENXIO); } else { unexpected_error = B_TRUE; } } } ops->vdev_op_io_done(zio); if (unexpected_error) VERIFY(vdev_probe(vd, zio) == NULL); return (zio); } /* * This function is used to change the priority of an existing zio that is * currently in-flight. This is used by the arc to upgrade priority in the * event that a demand read is made for a block that is currently queued * as a scrub or async read IO. Otherwise, the high priority read request * would end up having to wait for the lower priority IO. */ void zio_change_priority(zio_t *pio, zio_priority_t priority) { zio_t *cio, *cio_next; zio_link_t *zl = NULL; ASSERT3U(priority, <, ZIO_PRIORITY_NUM_QUEUEABLE); if (pio->io_vd != NULL && pio->io_vd->vdev_ops->vdev_op_leaf) { vdev_queue_change_io_priority(pio, priority); } else { pio->io_priority = priority; } mutex_enter(&pio->io_lock); for (cio = zio_walk_children(pio, &zl); cio != NULL; cio = cio_next) { cio_next = zio_walk_children(pio, &zl); zio_change_priority(cio, priority); } mutex_exit(&pio->io_lock); } /* * For non-raidz ZIOs, we can just copy aside the bad data read from the * disk, and use that to finish the checksum ereport later. */ static void zio_vsd_default_cksum_finish(zio_cksum_report_t *zcr, const abd_t *good_buf) { /* no processing needed */ zfs_ereport_finish_checksum(zcr, good_buf, zcr->zcr_cbdata, B_FALSE); } /*ARGSUSED*/ void zio_vsd_default_cksum_report(zio_t *zio, zio_cksum_report_t *zcr, void *ignored) { void *abd = abd_alloc_sametype(zio->io_abd, zio->io_size); abd_copy(abd, zio->io_abd, zio->io_size); zcr->zcr_cbinfo = zio->io_size; zcr->zcr_cbdata = abd; zcr->zcr_finish = zio_vsd_default_cksum_finish; zcr->zcr_free = zio_abd_free; } static zio_t * zio_vdev_io_assess(zio_t *zio) { vdev_t *vd = zio->io_vd; if (zio_wait_for_children(zio, ZIO_CHILD_VDEV_BIT, ZIO_WAIT_DONE)) { return (NULL); } if (vd == NULL && !(zio->io_flags & ZIO_FLAG_CONFIG_WRITER)) spa_config_exit(zio->io_spa, SCL_ZIO, zio); if (zio->io_vsd != NULL) { zio->io_vsd_ops->vsd_free(zio); zio->io_vsd = NULL; } if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_fault_injection(zio, EIO); /* * If the I/O failed, determine whether we should attempt to retry it. * * On retry, we cut in line in the issue queue, since we don't want * compression/checksumming/etc. work to prevent our (cheap) IO reissue. */ if (zio->io_error && vd == NULL && !(zio->io_flags & (ZIO_FLAG_DONT_RETRY | ZIO_FLAG_IO_RETRY))) { ASSERT(!(zio->io_flags & ZIO_FLAG_DONT_QUEUE)); /* not a leaf */ ASSERT(!(zio->io_flags & ZIO_FLAG_IO_BYPASS)); /* not a leaf */ zio->io_error = 0; zio->io_flags |= ZIO_FLAG_IO_RETRY | ZIO_FLAG_DONT_CACHE | ZIO_FLAG_DONT_AGGREGATE; zio->io_stage = ZIO_STAGE_VDEV_IO_START >> 1; zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, zio_requeue_io_start_cut_in_line); return (NULL); } /* * If we got an error on a leaf device, convert it to ENXIO * if the device is not accessible at all. */ if (zio->io_error && vd != NULL && vd->vdev_ops->vdev_op_leaf && !vdev_accessible(vd, zio)) zio->io_error = SET_ERROR(ENXIO); /* * If we can't write to an interior vdev (mirror or RAID-Z), * set vdev_cant_write so that we stop trying to allocate from it. */ if (zio->io_error == ENXIO && zio->io_type == ZIO_TYPE_WRITE && vd != NULL && !vd->vdev_ops->vdev_op_leaf) { vd->vdev_cant_write = B_TRUE; } /* * If a cache flush returns ENOTSUP or ENOTTY, we know that no future * attempts will ever succeed. In this case we set a persistent * boolean flag so that we don't bother with it in the future. */ if ((zio->io_error == ENOTSUP || zio->io_error == ENOTTY) && zio->io_type == ZIO_TYPE_IOCTL && zio->io_cmd == DKIOCFLUSHWRITECACHE && vd != NULL) vd->vdev_nowritecache = B_TRUE; if (zio->io_error) zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (vd != NULL && vd->vdev_ops->vdev_op_leaf && zio->io_physdone != NULL) { ASSERT(!(zio->io_flags & ZIO_FLAG_DELEGATED)); ASSERT(zio->io_child_type == ZIO_CHILD_VDEV); zio->io_physdone(zio->io_logical); } return (zio); } void zio_vdev_io_reissue(zio_t *zio) { ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_START); ASSERT(zio->io_error == 0); zio->io_stage >>= 1; } void zio_vdev_io_redone(zio_t *zio) { ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_DONE); zio->io_stage >>= 1; } void zio_vdev_io_bypass(zio_t *zio) { ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_START); ASSERT(zio->io_error == 0); zio->io_flags |= ZIO_FLAG_IO_BYPASS; zio->io_stage = ZIO_STAGE_VDEV_IO_ASSESS >> 1; } /* * ========================================================================== * Encrypt and store encryption parameters * ========================================================================== */ /* * This function is used for ZIO_STAGE_ENCRYPT. It is responsible for * managing the storage of encryption parameters and passing them to the * lower-level encryption functions. */ static zio_t * zio_encrypt(zio_t *zio) { zio_prop_t *zp = &zio->io_prop; spa_t *spa = zio->io_spa; blkptr_t *bp = zio->io_bp; uint64_t psize = BP_GET_PSIZE(bp); uint64_t dsobj = zio->io_bookmark.zb_objset; dmu_object_type_t ot = BP_GET_TYPE(bp); void *enc_buf = NULL; abd_t *eabd = NULL; uint8_t salt[ZIO_DATA_SALT_LEN]; uint8_t iv[ZIO_DATA_IV_LEN]; uint8_t mac[ZIO_DATA_MAC_LEN]; boolean_t no_crypt = B_FALSE; /* the root zio already encrypted the data */ if (zio->io_child_type == ZIO_CHILD_GANG) return (zio); /* only ZIL blocks are re-encrypted on rewrite */ if (!IO_IS_ALLOCATING(zio) && ot != DMU_OT_INTENT_LOG) return (zio); if (!(zp->zp_encrypt || BP_IS_ENCRYPTED(bp))) { BP_SET_CRYPT(bp, B_FALSE); return (zio); } /* if we are doing raw encryption set the provided encryption params */ if (zio->io_flags & ZIO_FLAG_RAW_ENCRYPT) { ASSERT0(BP_GET_LEVEL(bp)); BP_SET_CRYPT(bp, B_TRUE); BP_SET_BYTEORDER(bp, zp->zp_byteorder); if (ot != DMU_OT_OBJSET) zio_crypt_encode_mac_bp(bp, zp->zp_mac); /* dnode blocks must be written out in the provided byteorder */ if (zp->zp_byteorder != ZFS_HOST_BYTEORDER && ot == DMU_OT_DNODE) { void *bswap_buf = zio_buf_alloc(psize); abd_t *babd = abd_get_from_buf(bswap_buf, psize); ASSERT3U(BP_GET_COMPRESS(bp), ==, ZIO_COMPRESS_OFF); abd_copy_to_buf(bswap_buf, zio->io_abd, psize); dmu_ot_byteswap[DMU_OT_BYTESWAP(ot)].ob_func(bswap_buf, psize); abd_take_ownership_of_buf(babd, B_TRUE); zio_push_transform(zio, babd, psize, psize, NULL); } if (DMU_OT_IS_ENCRYPTED(ot)) zio_crypt_encode_params_bp(bp, zp->zp_salt, zp->zp_iv); return (zio); } /* indirect blocks only maintain a cksum of the lower level MACs */ if (BP_GET_LEVEL(bp) > 0) { BP_SET_CRYPT(bp, B_TRUE); VERIFY0(zio_crypt_do_indirect_mac_checksum_abd(B_TRUE, zio->io_orig_abd, BP_GET_LSIZE(bp), BP_SHOULD_BYTESWAP(bp), mac)); zio_crypt_encode_mac_bp(bp, mac); return (zio); } /* * Objset blocks are a special case since they have 2 256-bit MACs * embedded within them. */ if (ot == DMU_OT_OBJSET) { ASSERT0(DMU_OT_IS_ENCRYPTED(ot)); ASSERT3U(BP_GET_COMPRESS(bp), ==, ZIO_COMPRESS_OFF); BP_SET_CRYPT(bp, B_TRUE); VERIFY0(spa_do_crypt_objset_mac_abd(B_TRUE, spa, dsobj, zio->io_abd, psize, BP_SHOULD_BYTESWAP(bp))); return (zio); } /* unencrypted object types are only authenticated with a MAC */ if (!DMU_OT_IS_ENCRYPTED(ot)) { BP_SET_CRYPT(bp, B_TRUE); VERIFY0(spa_do_crypt_mac_abd(B_TRUE, spa, dsobj, zio->io_abd, psize, mac)); zio_crypt_encode_mac_bp(bp, mac); return (zio); } /* * Later passes of sync-to-convergence may decide to rewrite data * in place to avoid more disk reallocations. This presents a problem * for encryption because this constitutes rewriting the new data with * the same encryption key and IV. However, this only applies to blocks * in the MOS (particularly the spacemaps) and we do not encrypt the * MOS. We assert that the zio is allocating or an intent log write * to enforce this. */ ASSERT(IO_IS_ALLOCATING(zio) || ot == DMU_OT_INTENT_LOG); ASSERT(BP_GET_LEVEL(bp) == 0 || ot == DMU_OT_INTENT_LOG); ASSERT(spa_feature_is_active(spa, SPA_FEATURE_ENCRYPTION)); ASSERT3U(psize, !=, 0); enc_buf = zio_buf_alloc(psize); eabd = abd_get_from_buf(enc_buf, psize); abd_take_ownership_of_buf(eabd, B_TRUE); /* * For an explanation of what encryption parameters are stored * where, see the block comment in zio_crypt.c. */ if (ot == DMU_OT_INTENT_LOG) { zio_crypt_decode_params_bp(bp, salt, iv); } else { BP_SET_CRYPT(bp, B_TRUE); } /* Perform the encryption. This should not fail */ VERIFY0(spa_do_crypt_abd(B_TRUE, spa, &zio->io_bookmark, BP_GET_TYPE(bp), BP_GET_DEDUP(bp), BP_SHOULD_BYTESWAP(bp), salt, iv, mac, psize, zio->io_abd, eabd, &no_crypt)); /* encode encryption metadata into the bp */ if (ot == DMU_OT_INTENT_LOG) { /* * ZIL blocks store the MAC in the embedded checksum, so the * transform must always be applied. */ zio_crypt_encode_mac_zil(enc_buf, mac); zio_push_transform(zio, eabd, psize, psize, NULL); } else { BP_SET_CRYPT(bp, B_TRUE); zio_crypt_encode_params_bp(bp, salt, iv); zio_crypt_encode_mac_bp(bp, mac); if (no_crypt) { ASSERT3U(ot, ==, DMU_OT_DNODE); abd_free(eabd); } else { zio_push_transform(zio, eabd, psize, psize, NULL); } } return (zio); } /* * ========================================================================== * Generate and verify checksums * ========================================================================== */ static zio_t * zio_checksum_generate(zio_t *zio) { blkptr_t *bp = zio->io_bp; enum zio_checksum checksum; if (bp == NULL) { /* * This is zio_write_phys(). * We're either generating a label checksum, or none at all. */ checksum = zio->io_prop.zp_checksum; if (checksum == ZIO_CHECKSUM_OFF) return (zio); ASSERT(checksum == ZIO_CHECKSUM_LABEL); } else { if (BP_IS_GANG(bp) && zio->io_child_type == ZIO_CHILD_GANG) { ASSERT(!IO_IS_ALLOCATING(zio)); checksum = ZIO_CHECKSUM_GANG_HEADER; } else { checksum = BP_GET_CHECKSUM(bp); } } zio_checksum_compute(zio, checksum, zio->io_abd, zio->io_size); return (zio); } static zio_t * zio_checksum_verify(zio_t *zio) { zio_bad_cksum_t info; blkptr_t *bp = zio->io_bp; int error; ASSERT(zio->io_vd != NULL); if (bp == NULL) { /* * This is zio_read_phys(). * We're either verifying a label checksum, or nothing at all. */ if (zio->io_prop.zp_checksum == ZIO_CHECKSUM_OFF) return (zio); ASSERT(zio->io_prop.zp_checksum == ZIO_CHECKSUM_LABEL); } if ((error = zio_checksum_error(zio, &info)) != 0) { zio->io_error = error; if (error == ECKSUM && !(zio->io_flags & ZIO_FLAG_SPECULATIVE)) { int ret = zfs_ereport_start_checksum(zio->io_spa, zio->io_vd, &zio->io_bookmark, zio, zio->io_offset, zio->io_size, NULL, &info); if (ret != EALREADY) { mutex_enter(&zio->io_vd->vdev_stat_lock); zio->io_vd->vdev_stat.vs_checksum_errors++; mutex_exit(&zio->io_vd->vdev_stat_lock); } } } return (zio); } /* * Called by RAID-Z to ensure we don't compute the checksum twice. */ void zio_checksum_verified(zio_t *zio) { zio->io_pipeline &= ~ZIO_STAGE_CHECKSUM_VERIFY; } /* * ========================================================================== * Error rank. Error are ranked in the order 0, ENXIO, ECKSUM, EIO, other. * An error of 0 indicates success. ENXIO indicates whole-device failure, * which may be transient (e.g. unplugged) or permanent. ECKSUM and EIO * indicate errors that are specific to one I/O, and most likely permanent. * Any other error is presumed to be worse because we weren't expecting it. * ========================================================================== */ int zio_worst_error(int e1, int e2) { static int zio_error_rank[] = { 0, ENXIO, ECKSUM, EIO }; int r1, r2; for (r1 = 0; r1 < sizeof (zio_error_rank) / sizeof (int); r1++) if (e1 == zio_error_rank[r1]) break; for (r2 = 0; r2 < sizeof (zio_error_rank) / sizeof (int); r2++) if (e2 == zio_error_rank[r2]) break; return (r1 > r2 ? e1 : e2); } /* * ========================================================================== * I/O completion * ========================================================================== */ static zio_t * zio_ready(zio_t *zio) { blkptr_t *bp = zio->io_bp; zio_t *pio, *pio_next; zio_link_t *zl = NULL; if (zio_wait_for_children(zio, ZIO_CHILD_GANG_BIT | ZIO_CHILD_DDT_BIT, ZIO_WAIT_READY)) { return (NULL); } if (zio->io_ready) { ASSERT(IO_IS_ALLOCATING(zio)); ASSERT(bp->blk_birth == zio->io_txg || BP_IS_HOLE(bp) || (zio->io_flags & ZIO_FLAG_NOPWRITE)); ASSERT(zio->io_children[ZIO_CHILD_GANG][ZIO_WAIT_READY] == 0); zio->io_ready(zio); } if (bp != NULL && bp != &zio->io_bp_copy) zio->io_bp_copy = *bp; if (zio->io_error != 0) { zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (zio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(IO_IS_ALLOCATING(zio)); ASSERT(zio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(zio->io_metaslab_class != NULL); /* * We were unable to allocate anything, unreserve and * issue the next I/O to allocate. */ metaslab_class_throttle_unreserve( zio->io_metaslab_class, zio->io_prop.zp_copies, zio->io_allocator, zio); zio_allocate_dispatch(zio->io_spa, zio->io_allocator); } } mutex_enter(&zio->io_lock); zio->io_state[ZIO_WAIT_READY] = 1; pio = zio_walk_parents(zio, &zl); mutex_exit(&zio->io_lock); /* * As we notify zio's parents, new parents could be added. * New parents go to the head of zio's io_parent_list, however, * so we will (correctly) not notify them. The remainder of zio's * io_parent_list, from 'pio_next' onward, cannot change because * all parents must wait for us to be done before they can be done. */ for (; pio != NULL; pio = pio_next) { pio_next = zio_walk_parents(zio, &zl); zio_notify_parent(pio, zio, ZIO_WAIT_READY, NULL); } if (zio->io_flags & ZIO_FLAG_NODATA) { if (BP_IS_GANG(bp)) { zio->io_flags &= ~ZIO_FLAG_NODATA; } else { ASSERT((uintptr_t)zio->io_abd < SPA_MAXBLOCKSIZE); zio->io_pipeline &= ~ZIO_VDEV_IO_STAGES; } } if (zio_injection_enabled && zio->io_spa->spa_syncing_txg == zio->io_txg) zio_handle_ignored_writes(zio); return (zio); } /* * Update the allocation throttle accounting. */ static void zio_dva_throttle_done(zio_t *zio) { zio_t *lio __maybe_unused = zio->io_logical; zio_t *pio = zio_unique_parent(zio); vdev_t *vd = zio->io_vd; int flags = METASLAB_ASYNC_ALLOC; ASSERT3P(zio->io_bp, !=, NULL); ASSERT3U(zio->io_type, ==, ZIO_TYPE_WRITE); ASSERT3U(zio->io_priority, ==, ZIO_PRIORITY_ASYNC_WRITE); ASSERT3U(zio->io_child_type, ==, ZIO_CHILD_VDEV); ASSERT(vd != NULL); ASSERT3P(vd, ==, vd->vdev_top); ASSERT(zio_injection_enabled || !(zio->io_flags & ZIO_FLAG_IO_RETRY)); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REPAIR)); ASSERT(zio->io_flags & ZIO_FLAG_IO_ALLOCATING); ASSERT(!(lio->io_flags & ZIO_FLAG_IO_REWRITE)); ASSERT(!(lio->io_orig_flags & ZIO_FLAG_NODATA)); /* * Parents of gang children can have two flavors -- ones that * allocated the gang header (will have ZIO_FLAG_IO_REWRITE set) * and ones that allocated the constituent blocks. The allocation * throttle needs to know the allocating parent zio so we must find * it here. */ if (pio->io_child_type == ZIO_CHILD_GANG) { /* * If our parent is a rewrite gang child then our grandparent * would have been the one that performed the allocation. */ if (pio->io_flags & ZIO_FLAG_IO_REWRITE) pio = zio_unique_parent(pio); flags |= METASLAB_GANG_CHILD; } ASSERT(IO_IS_ALLOCATING(pio)); ASSERT3P(zio, !=, zio->io_logical); ASSERT(zio->io_logical != NULL); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REPAIR)); ASSERT0(zio->io_flags & ZIO_FLAG_NOPWRITE); ASSERT(zio->io_metaslab_class != NULL); mutex_enter(&pio->io_lock); metaslab_group_alloc_decrement(zio->io_spa, vd->vdev_id, pio, flags, pio->io_allocator, B_TRUE); mutex_exit(&pio->io_lock); metaslab_class_throttle_unreserve(zio->io_metaslab_class, 1, pio->io_allocator, pio); /* * Call into the pipeline to see if there is more work that * needs to be done. If there is work to be done it will be * dispatched to another taskq thread. */ zio_allocate_dispatch(zio->io_spa, pio->io_allocator); } static zio_t * zio_done(zio_t *zio) { /* * Always attempt to keep stack usage minimal here since * we can be called recursively up to 19 levels deep. */ const uint64_t psize = zio->io_size; zio_t *pio, *pio_next; zio_link_t *zl = NULL; /* * If our children haven't all completed, * wait for them and then repeat this pipeline stage. */ if (zio_wait_for_children(zio, ZIO_CHILD_ALL_BITS, ZIO_WAIT_DONE)) { return (NULL); } /* * If the allocation throttle is enabled, then update the accounting. * We only track child I/Os that are part of an allocating async * write. We must do this since the allocation is performed * by the logical I/O but the actual write is done by child I/Os. */ if (zio->io_flags & ZIO_FLAG_IO_ALLOCATING && zio->io_child_type == ZIO_CHILD_VDEV) { ASSERT(zio->io_metaslab_class != NULL); ASSERT(zio->io_metaslab_class->mc_alloc_throttle_enabled); zio_dva_throttle_done(zio); } /* * If the allocation throttle is enabled, verify that * we have decremented the refcounts for every I/O that was throttled. */ if (zio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(zio->io_type == ZIO_TYPE_WRITE); ASSERT(zio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(zio->io_bp != NULL); metaslab_group_alloc_verify(zio->io_spa, zio->io_bp, zio, zio->io_allocator); VERIFY(zfs_refcount_not_held( &zio->io_metaslab_class->mc_alloc_slots[zio->io_allocator], zio)); } for (int c = 0; c < ZIO_CHILD_TYPES; c++) for (int w = 0; w < ZIO_WAIT_TYPES; w++) ASSERT(zio->io_children[c][w] == 0); if (zio->io_bp != NULL && !BP_IS_EMBEDDED(zio->io_bp)) { ASSERT(zio->io_bp->blk_pad[0] == 0); ASSERT(zio->io_bp->blk_pad[1] == 0); ASSERT(bcmp(zio->io_bp, &zio->io_bp_copy, sizeof (blkptr_t)) == 0 || (zio->io_bp == zio_unique_parent(zio)->io_bp)); if (zio->io_type == ZIO_TYPE_WRITE && !BP_IS_HOLE(zio->io_bp) && zio->io_bp_override == NULL && !(zio->io_flags & ZIO_FLAG_IO_REPAIR)) { ASSERT3U(zio->io_prop.zp_copies, <=, BP_GET_NDVAS(zio->io_bp)); ASSERT(BP_COUNT_GANG(zio->io_bp) == 0 || (BP_COUNT_GANG(zio->io_bp) == BP_GET_NDVAS(zio->io_bp))); } if (zio->io_flags & ZIO_FLAG_NOPWRITE) VERIFY(BP_EQUAL(zio->io_bp, &zio->io_bp_orig)); } /* * If there were child vdev/gang/ddt errors, they apply to us now. */ zio_inherit_child_errors(zio, ZIO_CHILD_VDEV); zio_inherit_child_errors(zio, ZIO_CHILD_GANG); zio_inherit_child_errors(zio, ZIO_CHILD_DDT); /* * If the I/O on the transformed data was successful, generate any * checksum reports now while we still have the transformed data. */ if (zio->io_error == 0) { while (zio->io_cksum_report != NULL) { zio_cksum_report_t *zcr = zio->io_cksum_report; uint64_t align = zcr->zcr_align; uint64_t asize = P2ROUNDUP(psize, align); abd_t *adata = zio->io_abd; if (asize != psize) { adata = abd_alloc(asize, B_TRUE); abd_copy(adata, zio->io_abd, psize); abd_zero_off(adata, psize, asize - psize); } zio->io_cksum_report = zcr->zcr_next; zcr->zcr_next = NULL; zcr->zcr_finish(zcr, adata); zfs_ereport_free_checksum(zcr); if (asize != psize) abd_free(adata); } } zio_pop_transforms(zio); /* note: may set zio->io_error */ vdev_stat_update(zio, psize); /* * If this I/O is attached to a particular vdev is slow, exceeding * 30 seconds to complete, post an error described the I/O delay. * We ignore these errors if the device is currently unavailable. */ if (zio->io_delay >= MSEC2NSEC(zio_slow_io_ms)) { if (zio->io_vd != NULL && !vdev_is_dead(zio->io_vd)) { /* * We want to only increment our slow IO counters if * the IO is valid (i.e. not if the drive is removed). * * zfs_ereport_post() will also do these checks, but * it can also ratelimit and have other failures, so we * need to increment the slow_io counters independent * of it. */ if (zfs_ereport_is_valid(FM_EREPORT_ZFS_DELAY, zio->io_spa, zio->io_vd, zio)) { mutex_enter(&zio->io_vd->vdev_stat_lock); zio->io_vd->vdev_stat.vs_slow_ios++; mutex_exit(&zio->io_vd->vdev_stat_lock); (void) zfs_ereport_post(FM_EREPORT_ZFS_DELAY, zio->io_spa, zio->io_vd, &zio->io_bookmark, zio, 0); } } } if (zio->io_error) { /* * If this I/O is attached to a particular vdev, * generate an error message describing the I/O failure * at the block level. We ignore these errors if the * device is currently unavailable. */ if (zio->io_error != ECKSUM && zio->io_vd != NULL && !vdev_is_dead(zio->io_vd)) { int ret = zfs_ereport_post(FM_EREPORT_ZFS_IO, zio->io_spa, zio->io_vd, &zio->io_bookmark, zio, 0); if (ret != EALREADY) { mutex_enter(&zio->io_vd->vdev_stat_lock); if (zio->io_type == ZIO_TYPE_READ) zio->io_vd->vdev_stat.vs_read_errors++; else if (zio->io_type == ZIO_TYPE_WRITE) zio->io_vd->vdev_stat.vs_write_errors++; mutex_exit(&zio->io_vd->vdev_stat_lock); } } if ((zio->io_error == EIO || !(zio->io_flags & (ZIO_FLAG_SPECULATIVE | ZIO_FLAG_DONT_PROPAGATE))) && zio == zio->io_logical) { /* * For logical I/O requests, tell the SPA to log the * error and generate a logical data ereport. */ spa_log_error(zio->io_spa, &zio->io_bookmark); (void) zfs_ereport_post(FM_EREPORT_ZFS_DATA, zio->io_spa, NULL, &zio->io_bookmark, zio, 0); } } if (zio->io_error && zio == zio->io_logical) { /* * Determine whether zio should be reexecuted. This will * propagate all the way to the root via zio_notify_parent(). */ ASSERT(zio->io_vd == NULL && zio->io_bp != NULL); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); if (IO_IS_ALLOCATING(zio) && !(zio->io_flags & ZIO_FLAG_CANFAIL)) { if (zio->io_error != ENOSPC) zio->io_reexecute |= ZIO_REEXECUTE_NOW; else zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; } if ((zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_FREE) && !(zio->io_flags & ZIO_FLAG_SCAN_THREAD) && zio->io_error == ENXIO && spa_load_state(zio->io_spa) == SPA_LOAD_NONE && spa_get_failmode(zio->io_spa) != ZIO_FAILURE_MODE_CONTINUE) zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; if (!(zio->io_flags & ZIO_FLAG_CANFAIL) && !zio->io_reexecute) zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; /* * Here is a possibly good place to attempt to do * either combinatorial reconstruction or error correction * based on checksums. It also might be a good place * to send out preliminary ereports before we suspend * processing. */ } /* * If there were logical child errors, they apply to us now. * We defer this until now to avoid conflating logical child * errors with errors that happened to the zio itself when * updating vdev stats and reporting FMA events above. */ zio_inherit_child_errors(zio, ZIO_CHILD_LOGICAL); if ((zio->io_error || zio->io_reexecute) && IO_IS_ALLOCATING(zio) && zio->io_gang_leader == zio && !(zio->io_flags & (ZIO_FLAG_IO_REWRITE | ZIO_FLAG_NOPWRITE))) zio_dva_unallocate(zio, zio->io_gang_tree, zio->io_bp); zio_gang_tree_free(&zio->io_gang_tree); /* * Godfather I/Os should never suspend. */ if ((zio->io_flags & ZIO_FLAG_GODFATHER) && (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND)) zio->io_reexecute &= ~ZIO_REEXECUTE_SUSPEND; if (zio->io_reexecute) { /* * This is a logical I/O that wants to reexecute. * * Reexecute is top-down. When an i/o fails, if it's not * the root, it simply notifies its parent and sticks around. * The parent, seeing that it still has children in zio_done(), * does the same. This percolates all the way up to the root. * The root i/o will reexecute or suspend the entire tree. * * This approach ensures that zio_reexecute() honors * all the original i/o dependency relationships, e.g. * parents not executing until children are ready. */ ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); zio->io_gang_leader = NULL; mutex_enter(&zio->io_lock); zio->io_state[ZIO_WAIT_DONE] = 1; mutex_exit(&zio->io_lock); /* * "The Godfather" I/O monitors its children but is * not a true parent to them. It will track them through * the pipeline but severs its ties whenever they get into * trouble (e.g. suspended). This allows "The Godfather" * I/O to return status without blocking. */ zl = NULL; for (pio = zio_walk_parents(zio, &zl); pio != NULL; pio = pio_next) { zio_link_t *remove_zl = zl; pio_next = zio_walk_parents(zio, &zl); if ((pio->io_flags & ZIO_FLAG_GODFATHER) && (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND)) { zio_remove_child(pio, zio, remove_zl); /* * This is a rare code path, so we don't * bother with "next_to_execute". */ zio_notify_parent(pio, zio, ZIO_WAIT_DONE, NULL); } } if ((pio = zio_unique_parent(zio)) != NULL) { /* * We're not a root i/o, so there's nothing to do * but notify our parent. Don't propagate errors * upward since we haven't permanently failed yet. */ ASSERT(!(zio->io_flags & ZIO_FLAG_GODFATHER)); zio->io_flags |= ZIO_FLAG_DONT_PROPAGATE; /* * This is a rare code path, so we don't bother with * "next_to_execute". */ zio_notify_parent(pio, zio, ZIO_WAIT_DONE, NULL); } else if (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND) { /* * We'd fail again if we reexecuted now, so suspend * until conditions improve (e.g. device comes online). */ zio_suspend(zio->io_spa, zio, ZIO_SUSPEND_IOERR); } else { /* * Reexecution is potentially a huge amount of work. * Hand it off to the otherwise-unused claim taskq. */ ASSERT(taskq_empty_ent(&zio->io_tqent)); spa_taskq_dispatch_ent(zio->io_spa, ZIO_TYPE_CLAIM, ZIO_TASKQ_ISSUE, (task_func_t *)zio_reexecute, zio, 0, &zio->io_tqent); } return (NULL); } ASSERT(zio->io_child_count == 0); ASSERT(zio->io_reexecute == 0); ASSERT(zio->io_error == 0 || (zio->io_flags & ZIO_FLAG_CANFAIL)); /* * Report any checksum errors, since the I/O is complete. */ while (zio->io_cksum_report != NULL) { zio_cksum_report_t *zcr = zio->io_cksum_report; zio->io_cksum_report = zcr->zcr_next; zcr->zcr_next = NULL; zcr->zcr_finish(zcr, NULL); zfs_ereport_free_checksum(zcr); } if (zio->io_flags & ZIO_FLAG_FASTWRITE && zio->io_bp && !BP_IS_HOLE(zio->io_bp) && !BP_IS_EMBEDDED(zio->io_bp) && !(zio->io_flags & ZIO_FLAG_NOPWRITE)) { metaslab_fastwrite_unmark(zio->io_spa, zio->io_bp); } /* * It is the responsibility of the done callback to ensure that this * particular zio is no longer discoverable for adoption, and as * such, cannot acquire any new parents. */ if (zio->io_done) zio->io_done(zio); mutex_enter(&zio->io_lock); zio->io_state[ZIO_WAIT_DONE] = 1; mutex_exit(&zio->io_lock); /* * We are done executing this zio. We may want to execute a parent * next. See the comment in zio_notify_parent(). */ zio_t *next_to_execute = NULL; zl = NULL; for (pio = zio_walk_parents(zio, &zl); pio != NULL; pio = pio_next) { zio_link_t *remove_zl = zl; pio_next = zio_walk_parents(zio, &zl); zio_remove_child(pio, zio, remove_zl); zio_notify_parent(pio, zio, ZIO_WAIT_DONE, &next_to_execute); } if (zio->io_waiter != NULL) { mutex_enter(&zio->io_lock); zio->io_executor = NULL; cv_broadcast(&zio->io_cv); mutex_exit(&zio->io_lock); } else { zio_destroy(zio); } return (next_to_execute); } /* * ========================================================================== * I/O pipeline definition * ========================================================================== */ static zio_pipe_stage_t *zio_pipeline[] = { NULL, zio_read_bp_init, zio_write_bp_init, zio_free_bp_init, zio_issue_async, zio_write_compress, zio_encrypt, zio_checksum_generate, zio_nop_write, zio_ddt_read_start, zio_ddt_read_done, zio_ddt_write, zio_ddt_free, zio_gang_assemble, zio_gang_issue, zio_dva_throttle, zio_dva_allocate, zio_dva_free, zio_dva_claim, zio_ready, zio_vdev_io_start, zio_vdev_io_done, zio_vdev_io_assess, zio_checksum_verify, zio_done }; /* * Compare two zbookmark_phys_t's to see which we would reach first in a * pre-order traversal of the object tree. * * This is simple in every case aside from the meta-dnode object. For all other * objects, we traverse them in order (object 1 before object 2, and so on). * However, all of these objects are traversed while traversing object 0, since * the data it points to is the list of objects. Thus, we need to convert to a * canonical representation so we can compare meta-dnode bookmarks to * non-meta-dnode bookmarks. * * We do this by calculating "equivalents" for each field of the zbookmark. * zbookmarks outside of the meta-dnode use their own object and level, and * calculate the level 0 equivalent (the first L0 blkid that is contained in the * blocks this bookmark refers to) by multiplying their blkid by their span * (the number of L0 blocks contained within one block at their level). * zbookmarks inside the meta-dnode calculate their object equivalent * (which is L0equiv * dnodes per data block), use 0 for their L0equiv, and use * level + 1<<31 (any value larger than a level could ever be) for their level. * This causes them to always compare before a bookmark in their object * equivalent, compare appropriately to bookmarks in other objects, and to * compare appropriately to other bookmarks in the meta-dnode. */ int zbookmark_compare(uint16_t dbss1, uint8_t ibs1, uint16_t dbss2, uint8_t ibs2, const zbookmark_phys_t *zb1, const zbookmark_phys_t *zb2) { /* * These variables represent the "equivalent" values for the zbookmark, * after converting zbookmarks inside the meta dnode to their * normal-object equivalents. */ uint64_t zb1obj, zb2obj; uint64_t zb1L0, zb2L0; uint64_t zb1level, zb2level; if (zb1->zb_object == zb2->zb_object && zb1->zb_level == zb2->zb_level && zb1->zb_blkid == zb2->zb_blkid) return (0); IMPLY(zb1->zb_level > 0, ibs1 >= SPA_MINBLOCKSHIFT); IMPLY(zb2->zb_level > 0, ibs2 >= SPA_MINBLOCKSHIFT); /* * BP_SPANB calculates the span in blocks. */ zb1L0 = (zb1->zb_blkid) * BP_SPANB(ibs1, zb1->zb_level); zb2L0 = (zb2->zb_blkid) * BP_SPANB(ibs2, zb2->zb_level); if (zb1->zb_object == DMU_META_DNODE_OBJECT) { zb1obj = zb1L0 * (dbss1 << (SPA_MINBLOCKSHIFT - DNODE_SHIFT)); zb1L0 = 0; zb1level = zb1->zb_level + COMPARE_META_LEVEL; } else { zb1obj = zb1->zb_object; zb1level = zb1->zb_level; } if (zb2->zb_object == DMU_META_DNODE_OBJECT) { zb2obj = zb2L0 * (dbss2 << (SPA_MINBLOCKSHIFT - DNODE_SHIFT)); zb2L0 = 0; zb2level = zb2->zb_level + COMPARE_META_LEVEL; } else { zb2obj = zb2->zb_object; zb2level = zb2->zb_level; } /* Now that we have a canonical representation, do the comparison. */ if (zb1obj != zb2obj) return (zb1obj < zb2obj ? -1 : 1); else if (zb1L0 != zb2L0) return (zb1L0 < zb2L0 ? -1 : 1); else if (zb1level != zb2level) return (zb1level > zb2level ? -1 : 1); /* * This can (theoretically) happen if the bookmarks have the same object * and level, but different blkids, if the block sizes are not the same. * There is presently no way to change the indirect block sizes */ return (0); } /* * This function checks the following: given that last_block is the place that * our traversal stopped last time, does that guarantee that we've visited * every node under subtree_root? Therefore, we can't just use the raw output * of zbookmark_compare. We have to pass in a modified version of * subtree_root; by incrementing the block id, and then checking whether * last_block is before or equal to that, we can tell whether or not having * visited last_block implies that all of subtree_root's children have been * visited. */ boolean_t zbookmark_subtree_completed(const dnode_phys_t *dnp, const zbookmark_phys_t *subtree_root, const zbookmark_phys_t *last_block) { zbookmark_phys_t mod_zb = *subtree_root; mod_zb.zb_blkid++; ASSERT(last_block->zb_level == 0); /* The objset_phys_t isn't before anything. */ if (dnp == NULL) return (B_FALSE); /* * We pass in 1ULL << (DNODE_BLOCK_SHIFT - SPA_MINBLOCKSHIFT) for the * data block size in sectors, because that variable is only used if * the bookmark refers to a block in the meta-dnode. Since we don't * know without examining it what object it refers to, and there's no * harm in passing in this value in other cases, we always pass it in. * * We pass in 0 for the indirect block size shift because zb2 must be * level 0. The indirect block size is only used to calculate the span * of the bookmark, but since the bookmark must be level 0, the span is * always 1, so the math works out. * * If you make changes to how the zbookmark_compare code works, be sure * to make sure that this code still works afterwards. */ return (zbookmark_compare(dnp->dn_datablkszsec, dnp->dn_indblkshift, 1ULL << (DNODE_BLOCK_SHIFT - SPA_MINBLOCKSHIFT), 0, &mod_zb, last_block) <= 0); } EXPORT_SYMBOL(zio_type_name); EXPORT_SYMBOL(zio_buf_alloc); EXPORT_SYMBOL(zio_data_buf_alloc); EXPORT_SYMBOL(zio_buf_free); EXPORT_SYMBOL(zio_data_buf_free); /* BEGIN CSTYLED */ ZFS_MODULE_PARAM(zfs_zio, zio_, slow_io_ms, INT, ZMOD_RW, "Max I/O completion time (milliseconds) before marking it as slow"); ZFS_MODULE_PARAM(zfs_zio, zio_, requeue_io_start_cut_in_line, INT, ZMOD_RW, "Prioritize requeued I/O"); ZFS_MODULE_PARAM(zfs, zfs_, sync_pass_deferred_free, INT, ZMOD_RW, "Defer frees starting in this pass"); ZFS_MODULE_PARAM(zfs, zfs_, sync_pass_dont_compress, INT, ZMOD_RW, "Don't compress starting in this pass"); ZFS_MODULE_PARAM(zfs, zfs_, sync_pass_rewrite, INT, ZMOD_RW, "Rewrite new bps starting in this pass"); ZFS_MODULE_PARAM(zfs_zio, zio_, dva_throttle_enabled, INT, ZMOD_RW, "Throttle block allocations in the ZIO pipeline"); ZFS_MODULE_PARAM(zfs_zio, zio_, deadman_log_all, INT, ZMOD_RW, "Log all slow ZIOs, not just those with vdevs"); /* END CSTYLED */ diff --git a/tests/zfs-tests/tests/functional/cli_root/zfs_copies/zfs_copies_002_pos.ksh b/tests/zfs-tests/tests/functional/cli_root/zfs_copies/zfs_copies_002_pos.ksh index c88e300412f1..6e293ca63829 100755 --- a/tests/zfs-tests/tests/functional/cli_root/zfs_copies/zfs_copies_002_pos.ksh +++ b/tests/zfs-tests/tests/functional/cli_root/zfs_copies/zfs_copies_002_pos.ksh @@ -1,119 +1,119 @@ #!/bin/ksh # # 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) 2016 by Delphix. All rights reserved. # . $STF_SUITE/include/libtest.shlib . $STF_SUITE/tests/functional/cli_root/zfs_copies/zfs_copies.kshlib # # DESCRIPTION: # Verify that the space used by multiple copies is charged correctly # # STRATEGY: # 1. Create filesystems with copies set as 2,3 respectively; # 2. Copy specified size data into each filesystem; # 3. Verify that the space is charged as expected with zfs list, ls -s, df(1m), # du(1) commands; # verify_runnable "both" function cleanup { typeset val for val in 1 2 3; do if datasetexists $TESTPOOL/fs_$val; then log_must zfs destroy $TESTPOOL/fs_$val fi done } log_assert "Verify that the space used by multiple copies is charged correctly." log_onexit cleanup for val in 1 2 3; do log_must zfs create -o copies=$val $TESTPOOL/fs_$val log_must mkfile $FILESIZE /$TESTPOOL/fs_$val/$FILE done # # Sync up the filesystem # sync # # Verify 'zfs list' can correctly list the space charged # log_note "Verify 'zfs list' can correctly list the space charged." fsize=${FILESIZE%[m|M]} for val in 1 2 3; do used=$(get_prop used $TESTPOOL/fs_$val) check_used $used $val done log_note "Verify 'ls -s' can correctly list the space charged." if is_linux || is_freebsd; then blksize=1024 else blksize=512 fi for val in 1 2 3; do blks=`ls -ls /$TESTPOOL/fs_$val/$FILE | awk '{print $1}'` (( used = blks * $blksize )) # bytes check_used $used $val done -log_note "Verify df(1M) can correctly display the space charged." +log_note "Verify df(1) can correctly display the space charged." for val in 1 2 3; do if is_freebsd; then used=`df -m /$TESTPOOL/fs_$val | grep $TESTPOOL/fs_$val \ | awk -v fs=fs_$val '$4 ~ fs {print $3}'` else used=`df -F zfs -k /$TESTPOOL/fs_$val/$FILE | grep $TESTPOOL/fs_$val \ | awk '{print $3}'` (( used = used * 1024 )) # kb -> bytes fi check_used $used $val done log_note "Verify du(1) can correctly display the space charged." for val in 1 2 3; do if is_freebsd; then used=`du -h /$TESTPOOL/fs_$val/$FILE | awk '{print $1}'` else used=`du -k /$TESTPOOL/fs_$val/$FILE | awk '{print $1}'` (( used = used * 1024 )) # kb -> bytes fi check_used $used $val done log_pass "The space used by multiple copies is charged correctly as expected. "