diff --git a/cmd/zfs/zfs_main.c b/cmd/zfs/zfs_main.c index 81727224b04e..533b355fa858 100644 --- a/cmd/zfs/zfs_main.c +++ b/cmd/zfs/zfs_main.c @@ -1,9603 +1,9606 @@ // SPDX-License-Identifier: CDDL-1.0 /* * 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 https://opensource.org/licenses/CDDL-1.0. * 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 #include #ifdef HAVE_IDMAP #include #include #endif /* HAVE_IDMAP */ #include "zfs_iter.h" #include "zfs_util.h" #include "zfs_comutil.h" #include "zfs_projectutil.h" libzfs_handle_t *g_zfs; 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_rewrite(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 #ifdef __linux__ static int zfs_do_zone(int argc, char **argv); static int zfs_do_unzone(int argc, char **argv); #endif static int zfs_do_help(int argc, char **argv); enum zfs_options { ZFS_OPTION_JSON_NUMS_AS_INT = 1024 }; /* * 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_REWRITE, HELP_JAIL, HELP_UNJAIL, HELP_WAIT, HELP_ZONE, HELP_UNZONE, } 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 }, { "diff", zfs_do_diff, HELP_DIFF }, { 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 }, { "redact", zfs_do_redact, HELP_REDACT }, { NULL }, { "allow", zfs_do_allow, HELP_ALLOW }, { "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 }, { NULL }, { "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 }, { NULL }, { "program", zfs_do_channel_program, HELP_CHANNEL_PROGRAM }, { "rewrite", zfs_do_rewrite, HELP_REWRITE }, { "wait", zfs_do_wait, HELP_WAIT }, #ifdef __FreeBSD__ { NULL }, { "jail", zfs_do_jail, HELP_JAIL }, { "unjail", zfs_do_unjail, HELP_UNJAIL }, #endif #ifdef __linux__ { NULL }, { "zone", zfs_do_zone, HELP_ZONE }, { "unzone", zfs_do_unzone, HELP_UNZONE }, #endif }; #define NCOMMAND (sizeof (command_table) / sizeof (command_table[0])) #define MAX_CMD_LEN 256 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 [-Pnpuv] [-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] [-j [--json-int]] [-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] [-j [--json-int]] [-r|-d max] " "[-o property[,...]] [-s property]...\n\t " "[-S property]... [-t type[,...]] " "[filesystem|volume|snapshot] ...\n")); case HELP_MOUNT: return (gettext("\tmount [-j]\n" "\tmount [-flvO] [-o opts] <-a|-R filesystem|" "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 [-DLPbcehnpsVvw] " "[-i|-I snapshot]\n" "\t [-R [-X dataset[,dataset]...]] \n" "\tsend [-DnVvPLecw] [-i snapshot|bookmark] " "\n" "\tsend [-DnPpVvLec] [-i bookmark|snapshot] " "--redact \n" "\tsend [-nVvPe] -t \n" "\tsend [-PnVv] --saved filesystem\n")); case HELP_SET: return (gettext("\tset [-u] ... " " ...\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 [-rHp] ...\n")); case HELP_RELEASE: return (gettext("\trelease [-r] ...\n")); case HELP_DIFF: return (gettext("\tdiff [-FHth] " "[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 [-j]\n")); case HELP_REDACT: return (gettext("\tredact " " ...\n")); case HELP_REWRITE: - return (gettext("\trewrite [-rvx] [-o ] [-l ] " + return (gettext("\trewrite [-Prvx] [-o ] [-l ] " "\n")); case HELP_JAIL: return (gettext("\tjail \n")); case HELP_UNJAIL: return (gettext("\tunjail \n")); case HELP_WAIT: return (gettext("\twait [-t ] \n")); case HELP_ZONE: return (gettext("\tzone \n")); case HELP_UNZONE: return (gettext("\tunzone \n")); default: __builtin_unreachable(); } } 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(const 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%-22s ", 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 "); (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 __attribute__((noreturn)) 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, "%s", gettext("\nThe following properties are supported:\n")); (void) fprintf(fp, "\n\t%-21s %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%-22s ", "userused@..."); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, "\t%-22s ", "groupused@..."); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, "\t%-22s ", "projectused@..."); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, "\t%-22s ", "userobjused@..."); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, "\t%-22s ", "groupobjused@..."); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, "\t%-22s ", "projectobjused@..."); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, "\t%-22s ", "userquota@..."); (void) fprintf(fp, "YES NO | none\n"); (void) fprintf(fp, "\t%-22s ", "groupquota@..."); (void) fprintf(fp, "YES NO | none\n"); (void) fprintf(fp, "\t%-22s ", "projectquota@..."); (void) fprintf(fp, "YES NO | none\n"); (void) fprintf(fp, "\t%-22s ", "userobjquota@..."); (void) fprintf(fp, "YES NO | none\n"); (void) fprintf(fp, "\t%-22s ", "groupobjquota@..."); (void) fprintf(fp, "YES NO | none\n"); (void) fprintf(fp, "\t%-22s ", "projectobjquota@..."); (void) fprintf(fp, "YES NO | none\n"); (void) fprintf(fp, "\t%-22s ", "written@"); (void) fprintf(fp, " NO NO \n"); (void) fprintf(fp, "\t%-22s ", "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, "%s", 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"); (void) fprintf(fp, gettext("\nFor further help on a command or topic, " "run: %s\n"), "zfs help []"); } /* * 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 const 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(const 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(const 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(const char *done) { if (pt_shown) { (void) puts(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, NULL) != 0) { (void) fprintf(stderr, gettext("filesystem " "successfully created, but not shared\n")); ret = 1; } zfs_commit_shares(NULL); } 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); } /* * Return a default volblocksize for the pool which always uses more than * half of the data sectors. This primarily applies to dRAID which always * writes full stripe widths. */ static uint64_t default_volblocksize(zpool_handle_t *zhp, nvlist_t *props) { uint64_t volblocksize, asize = SPA_MINBLOCKSIZE; nvlist_t *tree, **vdevs; uint_t nvdevs; nvlist_t *config = zpool_get_config(zhp, NULL); if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &tree) != 0 || nvlist_lookup_nvlist_array(tree, ZPOOL_CONFIG_CHILDREN, &vdevs, &nvdevs) != 0) { return (ZVOL_DEFAULT_BLOCKSIZE); } for (int i = 0; i < nvdevs; i++) { nvlist_t *nv = vdevs[i]; uint64_t ashift, ndata, nparity; if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ASHIFT, &ashift) != 0) continue; if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_DRAID_NDATA, &ndata) == 0) { /* dRAID minimum allocation width */ asize = MAX(asize, ndata * (1ULL << ashift)); } else if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NPARITY, &nparity) == 0) { /* raidz minimum allocation width */ if (nparity == 1) asize = MAX(asize, 2 * (1ULL << ashift)); else asize = MAX(asize, 4 * (1ULL << ashift)); } else { /* mirror or (non-redundant) leaf vdev */ asize = MAX(asize, 1ULL << ashift); } } /* * Calculate the target volblocksize such that more than half * of the asize is used. The following table is for 4k sectors. * * n asize blksz used | n asize blksz used * -------------------------+--------------------------------- * 1 4,096 8,192 100% | 9 36,864 32,768 88% * 2 8,192 8,192 100% | 10 40,960 32,768 80% * 3 12,288 8,192 66% | 11 45,056 32,768 72% * 4 16,384 16,384 100% | 12 49,152 32,768 66% * 5 20,480 16,384 80% | 13 53,248 32,768 61% * 6 24,576 16,384 66% | 14 57,344 32,768 57% * 7 28,672 16,384 57% | 15 61,440 32,768 53% * 8 32,768 32,768 100% | 16 65,536 65,636 100% * * This is primarily a concern for dRAID which always allocates * a full stripe width. For dRAID the default stripe width is * n=8 in which case the volblocksize is set to 32k. Ignoring * compression there are no unused sectors. This same reasoning * applies to raidz[2,3] so target 4 sectors to minimize waste. */ uint64_t tgt_volblocksize = ZVOL_DEFAULT_BLOCKSIZE; while (tgt_volblocksize * 2 <= asize) tgt_volblocksize *= 2; const char *prop = zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE); if (nvlist_lookup_uint64(props, prop, &volblocksize) == 0) { /* Issue a warning when a non-optimal size is requested. */ if (volblocksize < ZVOL_DEFAULT_BLOCKSIZE) { (void) fprintf(stderr, gettext("Warning: " "volblocksize (%llu) is less than the default " "minimum block size (%llu).\nTo reduce wasted " "space a volblocksize of %llu is recommended.\n"), (u_longlong_t)volblocksize, (u_longlong_t)ZVOL_DEFAULT_BLOCKSIZE, (u_longlong_t)tgt_volblocksize); } else if (volblocksize < tgt_volblocksize) { (void) fprintf(stderr, gettext("Warning: " "volblocksize (%llu) is much less than the " "minimum allocation\nunit (%llu), which wastes " "at least %llu%% of space. To reduce wasted " "space,\nuse a larger volblocksize (%llu is " "recommended), fewer dRAID data disks\n" "per group, or smaller sector size (ashift).\n"), (u_longlong_t)volblocksize, (u_longlong_t)asize, (u_longlong_t)((100 * (asize - volblocksize)) / asize), (u_longlong_t)tgt_volblocksize); } } else { volblocksize = tgt_volblocksize; fnvlist_add_uint64(props, prop, volblocksize); } return (volblocksize); } /* * 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 '-u' flag prevents the newly created file system from being mounted. * * 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 nomount = B_FALSE; boolean_t verbose = B_FALSE; boolean_t parseable = B_FALSE; int ret = 1; nvlist_t *props; uint64_t intval; const char *strval; if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) nomem(); /* check options */ while ((c = getopt(argc, argv, ":PV:b:nso:puv")) != -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 'u': nomount = 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; } if (nomount && type != ZFS_TYPE_FILESYSTEM) { (void) fprintf(stderr, gettext("'-u' can only be " "used when creating a filesystem\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) { 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 (type == ZFS_TYPE_VOLUME) { const char *prop = zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE); uint64_t volblocksize = default_volblocksize(zpool_handle, real_props); if (volblocksize != ZVOL_DEFAULT_BLOCKSIZE && nvlist_lookup_string(props, prop, &strval) != 0) { char *tmp; if (asprintf(&tmp, "%llu", (u_longlong_t)volblocksize) == -1) nomem(); nvlist_add_string(props, prop, tmp); free(tmp); } /* * If volsize is not a multiple of volblocksize, round it * up to the nearest multiple of the volblocksize. */ 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; 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; const 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; } if (nomount) { ret = 0; goto error; } 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); if (error != 0) { zfs_close(zhp); return (-1); } } } 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_v2(fs_zhp, 0, 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_v2(zhp, 0, 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_v2(zhp, 0, 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_v2(zhp, 0, 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_v2(zhp, 0, 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_v2(zhp, 0, 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_v2(zhp, 0, 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); } /* * Generates an nvlist with output version for every command based on params. * Purpose of this is to add a version of JSON output, considering the schema * format might be updated for each command in future. * * Schema: * * "output_version": { * "command": string, * "vers_major": integer, * "vers_minor": integer, * } */ static nvlist_t * zfs_json_schema(int maj_v, int min_v) { nvlist_t *sch = NULL; nvlist_t *ov = NULL; char cmd[MAX_CMD_LEN]; snprintf(cmd, MAX_CMD_LEN, "zfs %s", current_command->name); sch = fnvlist_alloc(); ov = fnvlist_alloc(); fnvlist_add_string(ov, "command", cmd); fnvlist_add_uint32(ov, "vers_major", maj_v); fnvlist_add_uint32(ov, "vers_minor", min_v); fnvlist_add_nvlist(sch, "output_version", ov); fnvlist_free(ov); return (sch); } static void fill_dataset_info(nvlist_t *list, zfs_handle_t *zhp, boolean_t as_int) { char createtxg[ZFS_MAXPROPLEN]; zfs_type_t type = zfs_get_type(zhp); nvlist_add_string(list, "name", zfs_get_name(zhp)); switch (type) { case ZFS_TYPE_FILESYSTEM: fnvlist_add_string(list, "type", "FILESYSTEM"); break; case ZFS_TYPE_VOLUME: fnvlist_add_string(list, "type", "VOLUME"); break; case ZFS_TYPE_SNAPSHOT: fnvlist_add_string(list, "type", "SNAPSHOT"); break; case ZFS_TYPE_POOL: fnvlist_add_string(list, "type", "POOL"); break; case ZFS_TYPE_BOOKMARK: fnvlist_add_string(list, "type", "BOOKMARK"); break; default: fnvlist_add_string(list, "type", "UNKNOWN"); break; } if (type != ZFS_TYPE_POOL) fnvlist_add_string(list, "pool", zfs_get_pool_name(zhp)); if (as_int) { fnvlist_add_uint64(list, "createtxg", zfs_prop_get_int(zhp, ZFS_PROP_CREATETXG)); } else { if (zfs_prop_get(zhp, ZFS_PROP_CREATETXG, createtxg, sizeof (createtxg), NULL, NULL, 0, B_TRUE) == 0) fnvlist_add_string(list, "createtxg", createtxg); } if (type == ZFS_TYPE_SNAPSHOT) { char *snap = strdup(zfs_get_name(zhp)); char *ds = strsep(&snap, "@"); fnvlist_add_string(list, "dataset", ds); fnvlist_add_string(list, "snapshot_name", snap); free(ds); } } /* * zfs get [-rHp] [-j [--json-int]] [-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. * -j Display output in JSON format. * --json-int Display numbers as integers instead of strings. * * 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; nvlist_t *item, *d = NULL, *props = NULL; const char *strval; const char *sourceval; boolean_t received = is_recvd_column(cbp); int err = 0; if (cbp->cb_json) { d = fnvlist_lookup_nvlist(cbp->cb_jsobj, "datasets"); if (d == NULL) { fprintf(stderr, "datasets obj not found.\n"); exit(1); } props = fnvlist_alloc(); } 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_USERPROP) { 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; err = zprop_collect_property(zfs_get_name(zhp), cbp, zfs_prop_to_name(pl->pl_prop), buf, sourcetype, source, recvdval, props); } 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)); } err = zprop_collect_property(zfs_get_name(zhp), cbp, pl->pl_user_prop, buf, sourcetype, source, NULL, props); } 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)); } err = zprop_collect_property(zfs_get_name(zhp), cbp, pl->pl_user_prop, buf, sourcetype, source, NULL, props); } else { if (nvlist_lookup_nvlist(user_props, pl->pl_user_prop, &propval) != 0) { if (pl->pl_all) continue; sourcetype = ZPROP_SRC_NONE; strval = "-"; } else { strval = fnvlist_lookup_string(propval, ZPROP_VALUE); sourceval = fnvlist_lookup_string(propval, ZPROP_SOURCE); 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; err = zprop_collect_property(zfs_get_name(zhp), cbp, pl->pl_user_prop, strval, sourcetype, source, recvdval, props); } if (err != 0) return (err); } if (cbp->cb_json) { if (!nvlist_empty(props)) { item = fnvlist_alloc(); fill_dataset_info(item, zhp, cbp->cb_json_as_int); fnvlist_add_nvlist(item, "properties", props); fnvlist_add_nvlist(d, zfs_get_name(zhp), item); fnvlist_free(props); fnvlist_free(item); } else { fnvlist_free(props); } } 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 *fields; int ret = 0; int limit = 0; zprop_list_t fake_name = { 0 }; nvlist_t *data; /* * 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; struct option long_options[] = { {"json", no_argument, NULL, 'j'}, {"json-int", no_argument, NULL, ZFS_OPTION_JSON_NUMS_AS_INT}, {0, 0, 0, 0} }; /* check options */ while ((c = getopt_long(argc, argv, ":d:o:s:jrt:Hp", long_options, NULL)) != -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 'j': cb.cb_json = B_TRUE; cb.cb_jsobj = zfs_json_schema(0, 1); data = fnvlist_alloc(); fnvlist_add_nvlist(cb.cb_jsobj, "datasets", data); fnvlist_free(data); break; case ZFS_OPTION_JSON_NUMS_AS_INT: cb.cb_json_as_int = B_TRUE; cb.cb_literal = 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. */ memset(&cb.cb_columns, 0, sizeof (cb.cb_columns)); i = 0; for (char *tok; (tok = strsep(&optarg, ",")); ) { static const char *const col_subopts[] = { "name", "property", "value", "received", "source", "all" }; static const zfs_get_column_t col_subopt_col[] = { GET_COL_NAME, GET_COL_PROPERTY, GET_COL_VALUE, GET_COL_RECVD, GET_COL_SOURCE }; static const int col_subopt_flags[] = { 0, 0, 0, ZFS_ITER_RECVD_PROPS, 0 }; if (i == ZFS_GET_NCOLS) { (void) fprintf(stderr, gettext("too " "many fields given to -o " "option\n")); usage(B_FALSE); } for (c = 0; c < ARRAY_SIZE(col_subopts); ++c) if (strcmp(tok, col_subopts[c]) == 0) goto found; (void) fprintf(stderr, gettext("invalid column name '%s'\n"), tok); usage(B_FALSE); found: if (c >= 5) { if (i > 0) { (void) fprintf(stderr, gettext("\"all\" conflicts " "with specific fields " "given to -o option\n")); usage(B_FALSE); } memcpy(cb.cb_columns, col_subopt_col, sizeof (col_subopt_col)); flags |= ZFS_ITER_RECVD_PROPS; i = ZFS_GET_NCOLS; } else { cb.cb_columns[i++] = col_subopt_col[c]; flags |= col_subopt_flags[c]; } } break; case 's': cb.cb_sources = 0; for (char *tok; (tok = strsep(&optarg, ",")); ) { static const char *const source_opt[] = { "local", "default", "inherited", "received", "temporary", "none" }; static const int source_flg[] = { ZPROP_SRC_LOCAL, ZPROP_SRC_DEFAULT, ZPROP_SRC_INHERITED, ZPROP_SRC_RECEIVED, ZPROP_SRC_TEMPORARY, ZPROP_SRC_NONE }; for (i = 0; i < ARRAY_SIZE(source_opt); ++i) if (strcmp(tok, source_opt[i]) == 0) { cb.cb_sources |= source_flg[i]; goto found2; } (void) fprintf(stderr, gettext("invalid source '%s'\n"), tok); usage(B_FALSE); found2:; } break; case 't': types = 0; flags &= ~ZFS_ITER_PROP_LISTSNAPS; for (char *tok; (tok = strsep(&optarg, ",")); ) { static const char *const type_opts[] = { "filesystem", "fs", "volume", "vol", "snapshot", "snap", "bookmark", "all" }; static const int type_types[] = { ZFS_TYPE_FILESYSTEM, ZFS_TYPE_FILESYSTEM, ZFS_TYPE_VOLUME, ZFS_TYPE_VOLUME, ZFS_TYPE_SNAPSHOT, ZFS_TYPE_SNAPSHOT, ZFS_TYPE_BOOKMARK, ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK }; for (i = 0; i < ARRAY_SIZE(type_opts); ++i) if (strcmp(tok, type_opts[i]) == 0) { types |= type_types[i]; goto found3; } (void) fprintf(stderr, gettext("invalid type '%s'\n"), tok); usage(B_FALSE); found3:; } 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); } if (!cb.cb_json && cb.cb_json_as_int) { (void) fprintf(stderr, gettext("'--json-int' only works with" " '-j' option\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 (ret == 0 && cb.cb_json) zcmd_print_json(cb.cb_jsobj); else if (ret != 0 && cb.cb_json) nvlist_free(cb.cb_jsobj); 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_USERPROP) { 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[24]; (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) strlcpy(cb->cb_lastfs, zfs_get_name(zhp), sizeof (cb->cb_lastfs)); } 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 const char *const us_field_hdr[] = { "TYPE", "NAME", "USED", "QUOTA", "OBJUSED", "OBJQUOTA" }; static const char *const 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 const char *const 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(const char *field) { for (int 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) { const char *lvstr = ""; const 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, uint64_t default_quota) { us_cbdata_t *cb = (us_cbdata_t *)arg; zfs_userquota_prop_t prop = cb->cb_prop; char *name = NULL; const 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", default_quota); } 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", default_quota); } } 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 = -1; uint64_t val64 = -1; const 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: val32 = fnvpair_value_uint32(nvp); break; case DATA_TYPE_UINT64: val64 = fnvpair_value_uint64(nvp); break; case DATA_TYPE_STRING: strval = fnvpair_value_string(nvp); 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 = 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) putchar('\t'); else (void) fputs(" ", stdout); } if (scripted) (void) fputs(strval, stdout); 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) putchar('\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 (zfs_get_underlying_type(zhp) != 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; boolean_t cb_json; nvlist_t *cb_jsobj; boolean_t cb_json_as_int; } 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; color_start(ANSI_BOLD); for (; pl != NULL; pl = pl->pl_next) { if (!first) { (void) printf(" "); } else { first = B_FALSE; } right_justify = B_FALSE; if (pl->pl_prop != ZPROP_USERPROP) { 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); } color_end(); (void) printf("\n"); } /* * Decides on the color that the avail value should be printed in. * > 80% used = yellow * > 90% used = red */ static const char * zfs_list_avail_color(zfs_handle_t *zhp) { uint64_t used = zfs_prop_get_int(zhp, ZFS_PROP_USED); uint64_t avail = zfs_prop_get_int(zhp, ZFS_PROP_AVAILABLE); int percentage = (int)((double)avail / MAX(avail + used, 1) * 100); if (percentage > 20) return (NULL); else if (percentage > 10) return (ANSI_YELLOW); else return (ANSI_RED); } /* * Given a dataset and a list of fields, print out all the properties according * to the described layout, or return an nvlist containing all the fields, later * to be printed out as JSON object. */ static void collect_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; const char *propstr; boolean_t right_justify; nvlist_t *item, *d, *props; item = d = props = NULL; zprop_source_t sourcetype = ZPROP_SRC_NONE; char source[ZFS_MAX_DATASET_NAME_LEN]; if (cb->cb_json) { d = fnvlist_lookup_nvlist(cb->cb_jsobj, "datasets"); if (d == NULL) { fprintf(stderr, "datasets obj not found.\n"); exit(1); } item = fnvlist_alloc(); props = fnvlist_alloc(); fill_dataset_info(item, zhp, cb->cb_json_as_int); } for (; pl != NULL; pl = pl->pl_next) { if (!cb->cb_json && !first) { if (cb->cb_scripted) (void) putchar('\t'); else (void) fputs(" ", stdout); } 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_USERPROP) { if (zfs_prop_get(zhp, pl->pl_prop, property, sizeof (property), &sourcetype, source, sizeof (source), 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)) { sourcetype = ZPROP_SRC_LOCAL; if (zfs_prop_get_userquota(zhp, pl->pl_user_prop, property, sizeof (property), cb->cb_literal) != 0) { sourcetype = ZPROP_SRC_NONE; propstr = "-"; } else { propstr = property; } right_justify = B_TRUE; } else if (zfs_prop_written(pl->pl_user_prop)) { sourcetype = ZPROP_SRC_LOCAL; if (zfs_prop_get_written(zhp, pl->pl_user_prop, property, sizeof (property), cb->cb_literal) != 0) { sourcetype = ZPROP_SRC_NONE; propstr = "-"; } else { propstr = property; } right_justify = B_TRUE; } else { if (nvlist_lookup_nvlist(userprops, pl->pl_user_prop, &propval) != 0) { propstr = "-"; } else { propstr = fnvlist_lookup_string(propval, ZPROP_VALUE); strlcpy(source, fnvlist_lookup_string(propval, ZPROP_SOURCE), ZFS_MAX_DATASET_NAME_LEN); if (strcmp(source, zfs_get_name(zhp)) == 0) { sourcetype = ZPROP_SRC_LOCAL; } else if (strcmp(source, ZPROP_SOURCE_VAL_RECVD) == 0) { sourcetype = ZPROP_SRC_RECEIVED; } else { sourcetype = ZPROP_SRC_INHERITED; } } right_justify = B_FALSE; } if (cb->cb_json) { if (pl->pl_prop == ZFS_PROP_NAME) continue; const char *prop_name; if (pl->pl_prop != ZPROP_USERPROP) prop_name = zfs_prop_to_name(pl->pl_prop); else prop_name = pl->pl_user_prop; if (zprop_nvlist_one_property( prop_name, propstr, sourcetype, source, NULL, props, cb->cb_json_as_int) != 0) nomem(); } else { /* * zfs_list_avail_color() needs * ZFS_PROP_AVAILABLE + USED, so we need another * for() search for the USED part when no colors * wanted, we can skip the whole thing */ if (use_color() && pl->pl_prop == ZFS_PROP_AVAILABLE) { zprop_list_t *pl2 = cb->cb_proplist; for (; pl2 != NULL; pl2 = pl2->pl_next) { if (pl2->pl_prop == ZFS_PROP_USED) { color_start( zfs_list_avail_color(zhp)); /* * found it, no need for more * loops */ break; } } } /* * 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) fputs(propstr, stdout); else if (right_justify) { (void) printf("%*s", (int)pl->pl_width, propstr); } else { (void) printf("%-*s", (int)pl->pl_width, propstr); } if (pl->pl_prop == ZFS_PROP_AVAILABLE) color_end(); } } if (cb->cb_json) { fnvlist_add_nvlist(item, "properties", props); fnvlist_add_nvlist(d, zfs_get_name(zhp), item); fnvlist_free(props); fnvlist_free(item); } else (void) putchar('\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 && !cbp->cb_json) print_header(cbp); cbp->cb_first = B_FALSE; } collect_dataset(zhp, cbp); return (0); } static int zfs_do_list(int argc, char **argv) { int c; char default_fields[] = "name,used,available,referenced,mountpoint"; int types = ZFS_TYPE_DATASET; boolean_t types_specified = B_FALSE; char *fields = default_fields; list_cbdata_t cb = { 0 }; int limit = 0; int ret = 0; zfs_sort_column_t *sortcol = NULL; int flags = ZFS_ITER_PROP_LISTSNAPS | ZFS_ITER_ARGS_CAN_BE_PATHS; nvlist_t *data = NULL; struct option long_options[] = { {"json", no_argument, NULL, 'j'}, {"json-int", no_argument, NULL, ZFS_OPTION_JSON_NUMS_AS_INT}, {0, 0, 0, 0} }; /* check options */ while ((c = getopt_long(argc, argv, "jHS:d:o:prs:t:", long_options, NULL)) != -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 'j': cb.cb_json = B_TRUE; cb.cb_jsobj = zfs_json_schema(0, 1); data = fnvlist_alloc(); fnvlist_add_nvlist(cb.cb_jsobj, "datasets", data); fnvlist_free(data); break; case ZFS_OPTION_JSON_NUMS_AS_INT: cb.cb_json_as_int = B_TRUE; cb.cb_literal = B_TRUE; 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; for (char *tok; (tok = strsep(&optarg, ",")); ) { static const char *const type_subopts[] = { "filesystem", "fs", "volume", "vol", "snapshot", "snap", "bookmark", "all" }; static const int type_types[] = { ZFS_TYPE_FILESYSTEM, ZFS_TYPE_FILESYSTEM, ZFS_TYPE_VOLUME, ZFS_TYPE_VOLUME, ZFS_TYPE_SNAPSHOT, ZFS_TYPE_SNAPSHOT, ZFS_TYPE_BOOKMARK, ZFS_TYPE_DATASET | ZFS_TYPE_BOOKMARK }; for (c = 0; c < ARRAY_SIZE(type_subopts); ++c) if (strcmp(tok, type_subopts[c]) == 0) { types |= type_types[c]; goto found3; } (void) fprintf(stderr, gettext("invalid type '%s'\n"), tok); usage(B_FALSE); found3:; } 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 (!cb.cb_json && cb.cb_json_as_int) { (void) fprintf(stderr, gettext("'--json-int' only works with" " '-j' option\n")); usage(B_FALSE); } /* * 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; /* * If we are only going to list and sort by properties that are "fast" * then we can use "simple" mode and avoid populating the properties * nvlist. */ if (zfs_list_only_by_fast(cb.cb_proplist) && zfs_sort_only_by_fast(sortcol)) flags |= ZFS_ITER_SIMPLE; ret = zfs_for_each(argc, argv, flags, types, sortcol, &cb.cb_proplist, limit, list_callback, &cb); if (ret == 0 && cb.cb_json) zcmd_print_json(cb.cb_jsobj); else if (ret != 0 && cb.cb_json) nvlist_free(cb.cb_jsobj); 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. */ 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 */ 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: { zfs_handle_t *zhp = zfs_open(g_zfs, snap, ZFS_TYPE_SNAPSHOT); if (zhp == NULL) { (void) fprintf(stderr, gettext("provided snapshot %s " "does not exist\n"), snap); } else { zfs_close(zhp); } for (int i = 0; i < numrsnaps; i++) { zhp = zfs_open(g_zfs, rsnaps[i], ZFS_TYPE_SNAPSHOT); if (zhp == NULL) { (void) fprintf(stderr, gettext("provided " "snapshot %s does not exist\n"), rsnaps[i]); } else { zfs_close(zhp); } } 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; case ESRCH: (void) fprintf(stderr, gettext("attempted to resume redaction " " with a mismatched redaction list\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_v2(zhp, 0, 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_sorted_v2(zhp, 0, rollback_check, &cb, min_txg, 0)) != 0) goto out; if ((ret = zfs_iter_bookmarks_v2(zhp, 0, 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) { zprop_set_cbdata_t *cb = data; int ret = zfs_prop_set_list_flags(zhp, cb->cb_proplist, cb->cb_flags); if (ret != 0 || libzfs_errno(g_zfs) != EZFS_SUCCESS) { 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 (ret); } static int zfs_do_set(int argc, char **argv) { zprop_set_cbdata_t cb = { 0 }; int ds_start = -1; /* argv idx of first dataset arg */ int ret = 0; int i, c; /* check options */ while ((c = getopt(argc, argv, "u")) != -1) { switch (c) { case 'u': cb.cb_flags |= ZFS_SET_NOMOUNT; 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 arguments\n")); usage(B_FALSE); } if (argc < 2) { if (strchr(argv[0], '=') == 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 = 0; 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(&cb.cb_proplist, NV_UNIQUE_NAME, 0) != 0) nomem(); for (i = 0; i < ds_start; i++) { if (!parseprop(cb.cb_proplist, 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, &cb); error: nvlist_free(cb.cb_proplist); 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_v2(zhp, 0, 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); } /* * Array of prefixes to exclude – * a linear search, even if executed for each dataset, * is plenty good enough. */ typedef struct zfs_send_exclude_arg { size_t count; const char **list; } zfs_send_exclude_arg_t; static boolean_t zfs_do_send_exclude(zfs_handle_t *zhp, void *context) { zfs_send_exclude_arg_t *excludes = context; const char *name = zfs_get_name(zhp); for (size_t i = 0; i < excludes->count; ++i) { size_t len = strlen(excludes->list[i]); if (strncmp(name, excludes->list[i], len) == 0 && memchr("/@", name[len], sizeof ("/@"))) return (B_FALSE); } return (B_TRUE); } /* * 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; zfs_send_exclude_arg_t excludes = { 0 }; struct option long_options[] = { {"replicate", no_argument, NULL, 'R'}, {"skip-missing", no_argument, NULL, 's'}, {"redact", required_argument, NULL, 'd'}, {"props", no_argument, NULL, 'p'}, {"parsable", no_argument, NULL, 'P'}, {"dedup", no_argument, NULL, 'D'}, {"proctitle", no_argument, NULL, 'V'}, {"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'}, {"exclude", required_argument, NULL, 'X'}, {0, 0, 0, 0} }; /* check options */ while ((c = getopt_long(argc, argv, ":i:I:RsDpVvnPLeht:cwbd:SX:", long_options, NULL)) != -1) { switch (c) { case 'X': for (char *ds; (ds = strsep(&optarg, ",")) != NULL; ) { if (!zfs_name_valid(ds, ZFS_TYPE_DATASET) || strchr(ds, '/') == NULL) { (void) fprintf(stderr, gettext("-X %s: " "not a valid non-root dataset name" ".\n"), ds); usage(B_FALSE); } excludes.list = safe_realloc(excludes.list, sizeof (char *) * (excludes.count + 1)); excludes.list[excludes.count++] = ds; } break; 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 's': flags.skipmissing = 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.progressastitle = 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]); } free(excludes.list); usage(B_FALSE); break; case '?': 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]); } free(excludes.list); usage(B_FALSE); } } if ((flags.parsable || flags.progressastitle) && flags.verbosity == 0) flags.verbosity = 1; if (excludes.count > 0 && !flags.replicate) { free(excludes.list); (void) fprintf(stderr, gettext("Cannot specify " "dataset exclusion (-X) on a non-recursive " "send.\n")); return (1); } argc -= optind; argv += optind; if (resume_token != NULL) { if (fromname != NULL || flags.replicate || flags.props || flags.backup || flags.holds || flags.saved || redactbook != NULL) { free(excludes.list); (void) fprintf(stderr, gettext("invalid flags combined with -t\n")); usage(B_FALSE); } if (argc > 0) { free(excludes.list); (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } } else { if (argc < 1) { free(excludes.list); (void) fprintf(stderr, gettext("missing snapshot argument\n")); usage(B_FALSE); } if (argc > 1) { free(excludes.list); (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) { free(excludes.list); (void) fprintf(stderr, gettext("incompatible flags " "combined with saved send flag\n")); usage(B_FALSE); } if (strchr(argv[0], '@') != NULL) { free(excludes.list); (void) fprintf(stderr, gettext("saved send must " "specify the dataset with partially-received " "state\n")); usage(B_FALSE); } } if (flags.raw && redactbook != NULL) { free(excludes.list); (void) fprintf(stderr, gettext("Error: raw sends may not be redacted.\n")); return (1); } if (!flags.dryrun && isatty(STDOUT_FILENO)) { free(excludes.list); (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) { free(excludes.list); return (1); } err = zfs_send_saved(zhp, &flags, STDOUT_FILENO, resume_token); free(excludes.list); zfs_close(zhp); return (err != 0); } else if (resume_token != NULL) { free(excludes.list); return (zfs_send_resume(g_zfs, &flags, STDOUT_FILENO, resume_token)); } if (flags.skipmissing && !flags.replicate) { free(excludes.list); (void) fprintf(stderr, gettext("skip-missing flag can only be used in " "conjunction with replicate\n")); usage(B_FALSE); } /* * 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) { free(excludes.list); return (1); } err = zfs_send_one(zhp, fromname, STDOUT_FILENO, &flags, redactbook); free(excludes.list); 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")); free(excludes.list); return (1); } if (redactbook != NULL) { (void) fprintf(stderr, gettext("Error: multiple snapshots " "cannot be sent redacted.\n")); free(excludes.list); return (1); } if ((cp = strchr(argv[0], '@')) == NULL) { (void) fprintf(stderr, gettext("Error: " "Unsupported flag with filesystem or bookmark.\n")); free(excludes.list); return (1); } *cp = '\0'; toname = cp + 1; zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME); if (zhp == NULL) { free(excludes.list); 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)) { zfs_close(zhp); free(excludes.list); (void) fprintf(stderr, gettext("incremental source must be " "in same filesystem\n")); usage(B_FALSE); } fromname = cp + 1; if (strchr(fromname, '@') || strchr(fromname, '/')) { zfs_close(zhp); free(excludes.list); (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, excludes.count > 0 ? zfs_do_send_exclude : NULL, &excludes, 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); free(excludes.list); 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:dehMnuvFsAc")) != -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 'c': flags.heal = 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_RECEIVE_APPEND "receive:append" #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); } static int who_perm_compare(const void *larg, const void *rarg, void *unused) { (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); } static int deleg_perm_compare(const void *larg, const void *rarg, void *unused) { (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) { memset(fspset, 0, 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; memset(who_perm, 0, 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; memset(fsperm, 0, 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)); 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, sizeof (errbuf), 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, sizeof (errbuf), 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, sizeof (errbuf), 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_v2(zhp, 0, 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) strlcpy(parent, path, MIN(sizeof (parent), delim - path + 1)); 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, boolean_t parsable) { int i; nvpair_t *nvp = NULL; const char *const 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) { const 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]; const 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) { if (parsable) { (void) printf("%s\t%s\t%ld\n", zname, tagname, time); } else { (void) printf("%s\t%s\t%s\n", zname, tagname, tsbuf); } } else { if (parsable) { (void) printf("%-*s %-*s %ld\n", nwidth, zname, tagwidth, tagname, time); } 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 [-rHp] ... * * -r Lists holds that are set on the named snapshots recursively. * -H Scripted mode; elide headers and separate columns by tabs. * -p Display values in parsable (literal) format. */ static int zfs_do_holds(int argc, char **argv) { int c; boolean_t errors = B_FALSE; boolean_t scripted = B_FALSE; boolean_t recursive = B_FALSE; boolean_t parsable = B_FALSE; 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, "rHp")) != -1) { switch (c) { case 'r': recursive = B_TRUE; break; case 'H': scripted = B_TRUE; break; case 'p': parsable = 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); nvlist_t *nvl = fnvlist_alloc(); for (int 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 = B_TRUE; 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(1, argv + i, flags, types, NULL, NULL, limit, holds_callback, &cb); if (ret != 0) errors = B_TRUE; } /* * 2. print holds data */ print_holds(scripted, cb.cb_max_namelen, cb.cb_max_taglen, nvl, parsable); if (nvlist_empty(nvl)) (void) fprintf(stderr, gettext("no datasets available\n")); nvlist_free(nvl); return (errors); } #define CHECK_SPINNER 30 #define SPINNER_TIME 3 /* seconds */ #define MOUNT_TIME 1 /* seconds */ typedef struct get_all_state { char **ga_datasets; int ga_count; 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 const char *const 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_v2(zhp, 0, 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 int get_recursive_datasets(zfs_handle_t *zhp, void *data) { get_all_state_t *state = data; int len = strlen(zfs_get_name(zhp)); for (int i = 0; i < state->ga_count; ++i) { if (strcmp(state->ga_datasets[i], zfs_get_name(zhp)) == 0) return (get_one_dataset(zhp, data)); else if ((strncmp(state->ga_datasets[i], zfs_get_name(zhp), len) == 0) && state->ga_datasets[i][len] == '/') { (void) zfs_iter_filesystems_v2(zhp, 0, get_recursive_datasets, data); } } zfs_close(zhp); return (0); } static void get_all_datasets(get_all_state_t *state) { if (state->ga_verbose) set_progress_header(gettext("Reading ZFS config")); if (state->ga_datasets == NULL) (void) zfs_iter_root(g_zfs, get_one_dataset, state); else (void) zfs_iter_root(g_zfs, get_recursive_datasets, state); if (state->ga_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; enum sa_protocol 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, enum sa_protocol 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 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(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_unshare(zhp, mountpoint, NULL); 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: { enum sa_protocol prot[] = {SA_PROTOCOL_NFS, SA_NO_PROTOCOL}; shared_nfs = zfs_is_shared(zhp, NULL, prot); *prot = SA_PROTOCOL_SMB; shared_smb = zfs_is_shared(zhp, NULL, prot); 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); *prot = protocol; if (zfs_share(zhp, protocol == SA_NO_PROTOCOL ? NULL : prot)) return (1); } break; case OP_MOUNT: 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]; /* 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 enum sa_protocol sa_protocol_decode(const char *protocol) { for (enum sa_protocol i = 0; i < ARRAY_SIZE(sa_protocol_names); ++i) if (strcmp(protocol, sa_protocol_names[i]) == 0) return (i); (void) fputs(gettext("share type must be one of: "), stderr); for (enum sa_protocol i = 0; i < ARRAY_SIZE(sa_protocol_names); ++i) (void) fprintf(stderr, "%s%s", i != 0 ? ", " : "", sa_protocol_names[i]); (void) fputc('\n', stderr); usage(B_FALSE); } static int share_mount(int op, int argc, char **argv) { int do_all = 0; int recursive = 0; boolean_t verbose = B_FALSE; boolean_t json = B_FALSE; int c, ret = 0; char *options = NULL; int flags = 0; nvlist_t *jsobj, *data, *item; const uint_t mount_nthr = 512; uint_t nthr; jsobj = data = item = NULL; struct option long_options[] = { {"json", no_argument, NULL, 'j'}, {0, 0, 0, 0} }; /* check options */ while ((c = getopt_long(argc, argv, op == OP_MOUNT ? ":ajRlvo:Of" : "al", op == OP_MOUNT ? long_options : NULL, NULL)) != -1) { switch (c) { case 'a': do_all = 1; break; case 'R': recursive = 1; break; case 'v': verbose = B_TRUE; break; case 'l': flags |= MS_CRYPT; break; case 'j': json = B_TRUE; jsobj = zfs_json_schema(0, 1); data = fnvlist_alloc(); 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; if (json && argc != 0) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } /* check number of arguments */ if (do_all || recursive) { enum sa_protocol protocol = SA_NO_PROTOCOL; if (op == OP_SHARE && argc > 0) { protocol = sa_protocol_decode(argv[0]); argc--; argv++; } if (argc != 0 && do_all) { (void) fprintf(stderr, gettext("too many arguments\n")); usage(B_FALSE); } if (argc == 0 && recursive) { (void) fprintf(stderr, gettext("no dataset provided\n")); usage(B_FALSE); } start_progress_timer(); get_all_cb_t cb = { 0 }; get_all_state_t state = { 0 }; if (argc == 0) { state.ga_datasets = NULL; state.ga_count = -1; } else { zfs_handle_t *zhp; for (int i = 0; i < argc; i++) { zhp = zfs_open(g_zfs, argv[i], ZFS_TYPE_FILESYSTEM); if (zhp == NULL) usage(B_FALSE); zfs_close(zhp); } state.ga_datasets = argv; state.ga_count = argc; } state.ga_verbose = verbose; state.ga_cbp = &cb; get_all_datasets(&state); if (cb.cb_used == 0) { 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); /* For a 'zfs share -a' operation start with a clean slate. */ if (op == OP_SHARE) zfs_truncate_shares(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. */ nthr = op == OP_MOUNT && !(flags & MS_CRYPT) ? mount_nthr : 1; zfs_foreach_mountpoint(g_zfs, cb.cb_handles, cb.cb_used, share_mount_one_cb, &share_mount_state, nthr); zfs_commit_shares(NULL); 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) { FILE *mnttab; 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. */ if ((mnttab = fopen(MNTTAB, "re")) == NULL) { free(options); return (ENOENT); } while (getmntent(mnttab, &entry) == 0) { if (strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0 || strchr(entry.mnt_special, '@') != NULL) continue; if (json) { item = fnvlist_alloc(); fnvlist_add_string(item, "filesystem", entry.mnt_special); fnvlist_add_string(item, "mountpoint", entry.mnt_mountp); fnvlist_add_nvlist(data, entry.mnt_special, item); fnvlist_free(item); } else { (void) printf("%-30s %s\n", entry.mnt_special, entry.mnt_mountp); } } (void) fclose(mnttab); if (json) { fnvlist_add_nvlist(jsobj, "datasets", data); if (nvlist_empty(data)) fnvlist_free(jsobj); else zcmd_print_json(jsobj); fnvlist_free(data); } } 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, SA_NO_PROTOCOL, B_TRUE, options); zfs_commit_shares(NULL); zfs_close(zhp); } } free(options); return (ret); } /* * zfs mount -a * 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; static int unshare_unmount_compare(const void *larg, const void *rarg, void *unused) { (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; char *zfs_mntpnt, *entry_mntpnt; /* * Search for the given (major,minor) pair in the mount table. */ 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 the filesystem is mounted, check that the mountpoint matches * the one in the mnttab entry w.r.t. provided path. If it doesn't, * then we should not proceed further. */ entry_mntpnt = strdup(entry.mnt_mountp); if (zfs_is_mounted(zhp, &zfs_mntpnt)) { if (strcmp(zfs_mntpnt, entry_mntpnt) != 0) { (void) fprintf(stderr, gettext("cannot %s '%s': " "not an original mountpoint\n"), cmdname, path); free(zfs_mntpnt); free(entry_mntpnt); goto out; } free(zfs_mntpnt); } free(entry_mntpnt); 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, NULL, NULL)) { (void) fprintf(stderr, gettext("cannot unshare '%s': " "not currently shared\n"), path); } else { ret = zfs_unshare(zhp, path, NULL); zfs_commit_shares(NULL); } } 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. */ FILE *mnttab; 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; enum sa_protocol *protocol = NULL, single_protocol[] = {SA_NO_PROTOCOL, SA_NO_PROTOCOL}; if (op == OP_SHARE && argc > 0) { *single_protocol = sa_protocol_decode(argv[0]); protocol = single_protocol; 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(); if ((mnttab = fopen(MNTTAB, "re")) == NULL) { uu_avl_destroy(tree); uu_avl_pool_destroy(pool); return (ENOENT); } while (getmntent(mnttab, &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; break; 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); } } (void) fclose(mnttab); /* * 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_unshare(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 " "exports(5) or smb.conf(5) to unshare " "this filesystem\n")); ret = 1; } else if (!zfs_is_shared(zhp, NULL, NULL)) { (void) fprintf(stderr, gettext("cannot " "unshare '%s': not currently " "shared\n"), zfs_get_name(zhp)); ret = 1; } else if (zfs_unshareall(zhp, NULL) != 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(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(const 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, "FHth")) != -1) { switch (c) { case 'F': flags |= ZFS_DIFF_CLASSIFY; break; case 'H': flags |= ZFS_DIFF_PARSEABLE; break; case 't': flags |= ZFS_DIFF_TIMESTAMP; break; case 'h': flags |= ZFS_DIFF_NO_MANGLE; 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; 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; struct option long_options[] = { {"json", no_argument, NULL, 'j'}, {0, 0, 0, 0} }; /* check options */ while ((c = getopt_long(argc, argv, "nt:m:j", long_options, NULL)) != -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; } const char *poolname = argv[0]; const char *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; char *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, (const char **)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. */ const char *errstring = NULL; const char *msg = gettext("Channel program execution failed"); uint64_t instructions = 0; if (outnvl != NULL && nvlist_exists(outnvl, ZCP_RET_ERROR)) { const char *es = NULL; (void) nvlist_lookup_string(outnvl, ZCP_RET_ERROR, &es); if (es == NULL) errstring = strerror(ret); else errstring = es; 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_rewrite_file(const char *path, boolean_t verbose, zfs_rewrite_args_t *args) { int fd, ret = 0; fd = open(path, O_WRONLY); if (fd < 0) { ret = errno; (void) fprintf(stderr, gettext("failed to open %s: %s\n"), path, strerror(errno)); return (ret); } if (ioctl(fd, ZFS_IOC_REWRITE, args) < 0) { ret = errno; (void) fprintf(stderr, gettext("failed to rewrite %s: %s\n"), path, strerror(errno)); } else if (verbose) { printf("%s\n", path); } close(fd); return (ret); } static int zfs_rewrite_dir(const char *path, boolean_t verbose, boolean_t xdev, dev_t dev, zfs_rewrite_args_t *args, nvlist_t *dirs) { struct dirent *ent; DIR *dir; int ret = 0, err; dir = opendir(path); if (dir == NULL) { if (errno == ENOENT) return (0); ret = errno; (void) fprintf(stderr, gettext("failed to opendir %s: %s\n"), path, strerror(errno)); return (ret); } size_t plen = strlen(path) + 1; while ((ent = readdir(dir)) != NULL) { char *fullname; struct stat st; if (ent->d_type != DT_REG && ent->d_type != DT_DIR) continue; if (strcmp(ent->d_name, ".") == 0 || strcmp(ent->d_name, "..") == 0) continue; if (plen + strlen(ent->d_name) >= PATH_MAX) { (void) fprintf(stderr, gettext("path too long %s/%s\n"), path, ent->d_name); ret = ENAMETOOLONG; continue; } if (asprintf(&fullname, "%s/%s", path, ent->d_name) == -1) { (void) fprintf(stderr, gettext("failed to allocate memory\n")); ret = ENOMEM; continue; } if (xdev) { if (lstat(fullname, &st) < 0) { ret = errno; (void) fprintf(stderr, gettext("failed to stat %s: %s\n"), fullname, strerror(errno)); free(fullname); continue; } if (st.st_dev != dev) { free(fullname); continue; } } if (ent->d_type == DT_REG) { err = zfs_rewrite_file(fullname, verbose, args); if (err) ret = err; } else { /* DT_DIR */ fnvlist_add_uint64(dirs, fullname, dev); } free(fullname); } closedir(dir); return (ret); } static int zfs_rewrite_path(const char *path, boolean_t verbose, boolean_t recurse, boolean_t xdev, zfs_rewrite_args_t *args, nvlist_t *dirs) { struct stat st; int ret = 0; if (lstat(path, &st) < 0) { ret = errno; (void) fprintf(stderr, gettext("failed to stat %s: %s\n"), path, strerror(errno)); return (ret); } if (S_ISREG(st.st_mode)) { ret = zfs_rewrite_file(path, verbose, args); } else if (S_ISDIR(st.st_mode) && recurse) { ret = zfs_rewrite_dir(path, verbose, xdev, st.st_dev, args, dirs); } return (ret); } static int zfs_do_rewrite(int argc, char **argv) { int ret = 0, err, c; boolean_t recurse = B_FALSE, verbose = B_FALSE, xdev = B_FALSE; if (argc < 2) usage(B_FALSE); zfs_rewrite_args_t args; memset(&args, 0, sizeof (args)); - while ((c = getopt(argc, argv, "l:o:rvx")) != -1) { + while ((c = getopt(argc, argv, "Pl:o:rvx")) != -1) { switch (c) { + case 'P': + args.flags |= ZFS_REWRITE_PHYSICAL; + break; case 'l': args.len = strtoll(optarg, NULL, 0); break; case 'o': args.off = strtoll(optarg, NULL, 0); break; case 'r': recurse = B_TRUE; break; case 'v': verbose = B_TRUE; break; case 'x': xdev = B_TRUE; break; default: (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argv += optind; argc -= optind; if (argc == 0) { (void) fprintf(stderr, gettext("missing file or directory target(s)\n")); usage(B_FALSE); } nvlist_t *dirs = fnvlist_alloc(); for (int i = 0; i < argc; i++) { err = zfs_rewrite_path(argv[i], verbose, recurse, xdev, &args, dirs); if (err) ret = err; } nvpair_t *dir; while ((dir = nvlist_next_nvpair(dirs, NULL)) != NULL) { err = zfs_rewrite_dir(nvpair_name(dir), verbose, xdev, fnvpair_value_uint64(dir), &args, dirs); if (err) ret = err; fnvlist_remove_nvpair(dirs, dir); } fnvlist_free(dirs); return (ret); } static int zfs_do_wait(int argc, char **argv) { boolean_t enabled[ZFS_WAIT_NUM_ACTIVITIES]; int error = 0, i; int 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': /* Reset activities array */ memset(&enabled, 0, sizeof (enabled)); for (char *tok; (tok = strsep(&optarg, ",")); ) { static const char *const col_subopts[ ZFS_WAIT_NUM_ACTIVITIES] = { "deleteq" }; for (i = 0; i < ARRAY_SIZE(col_subopts); ++i) if (strcmp(tok, col_subopts[i]) == 0) { enabled[i] = B_TRUE; goto found; } (void) fprintf(stderr, gettext("invalid activity '%s'\n"), tok); usage(B_FALSE); found:; } 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) { int c; nvlist_t *jsobj = NULL, *zfs_ver = NULL; boolean_t json = B_FALSE; struct option long_options[] = { {"json", no_argument, NULL, 'j'}, {0, 0, 0, 0} }; while ((c = getopt_long(argc, argv, "j", long_options, NULL)) != -1) { switch (c) { case 'j': json = B_TRUE; jsobj = zfs_json_schema(0, 1); break; case '?': (void) fprintf(stderr, gettext("invalid option '%c'\n"), optopt); usage(B_FALSE); } } argc -= optind; if (argc != 0) { (void) fprintf(stderr, "too many arguments\n"); usage(B_FALSE); } if (json) { zfs_ver = zfs_version_nvlist(); if (zfs_ver) { fnvlist_add_nvlist(jsobj, "zfs_version", zfs_ver); zcmd_print_json(jsobj); fnvlist_free(zfs_ver); return (0); } else return (-1); } else return (zfs_version_print() != 0); } /* Display documentation */ static int zfs_do_help(int argc, char **argv) { char page[MAXNAMELEN]; if (argc < 3 || strcmp(argv[2], "zfs") == 0) strcpy(page, "zfs"); else if (strcmp(argv[2], "concepts") == 0 || strcmp(argv[2], "props") == 0) snprintf(page, sizeof (page), "zfs%s", argv[2]); else snprintf(page, sizeof (page), "zfs-%s", argv[2]); execlp("man", "man", page, NULL); fprintf(stderr, "couldn't run man program: %s", strerror(errno)); return (-1); } int main(int argc, char **argv) { int ret = 0; int i = 0; const 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_version_print() != 0); /* * Special case 'help' */ if (strcmp(cmdname, "help") == 0) return (zfs_do_help(argc, argv)); if ((g_zfs = libzfs_init()) == NULL) { (void) fprintf(stderr, "%s\n", libzfs_error_init(errno)); return (1); } zfs_save_arguments(argc, argv, history_str, sizeof (history_str)); libzfs_print_on_error(g_zfs, B_TRUE); zfs_setproctitle_init(argc, argv, environ); /* * Many commands modify input strings for string parsing reasons. * We create a copy to protect the original argv. */ newargv = safe_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); } /* * zfs zone nsfile filesystem * * Add or delete the given dataset to/from the namespace. */ #ifdef __linux__ static int zfs_do_zone_impl(int argc, char **argv, boolean_t attach) { zfs_handle_t *zhp; int ret; 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); } zhp = zfs_open(g_zfs, argv[2], ZFS_TYPE_FILESYSTEM); if (zhp == NULL) return (1); ret = (zfs_userns(zhp, argv[1], attach) != 0); zfs_close(zhp); return (ret); } static int zfs_do_zone(int argc, char **argv) { return (zfs_do_zone_impl(argc, argv, B_TRUE)); } static int zfs_do_unzone(int argc, char **argv) { return (zfs_do_zone_impl(argc, argv, B_FALSE)); } #endif #ifdef __FreeBSD__ #include #include /* * Attach/detach the given dataset to/from the given jail */ 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 */ 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 */ static int zfs_do_unjail(int argc, char **argv) { return (zfs_do_jail_impl(argc, argv, B_FALSE)); } #endif diff --git a/include/sys/dbuf.h b/include/sys/dbuf.h index 756459b2fbb5..baf3b1508335 100644 --- a/include/sys/dbuf.h +++ b/include/sys/dbuf.h @@ -1,522 +1,523 @@ // SPDX-License-Identifier: CDDL-1.0 /* * 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 https://opensource.org/licenses/CDDL-1.0. * 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, 2020 by Delphix. All rights reserved. * Copyright (c) 2013 by Saso Kiselkov. All rights reserved. * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved. */ #ifndef _SYS_DBUF_H #define _SYS_DBUF_H #include #include #include #include #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif #define IN_DMU_SYNC 2 /* * The simplified state transition diagram for dbufs looks like: * * +-------> READ ------+ * | | * | V * (alloc)-->UNCACHED CACHED-->EVICTING-->(free) * ^ | ^ ^ * | | | | * | +-------> FILL ------+ | * | | | | * | | | | * | +------> NOFILL -----+-----> UNCACHED * | | (Direct I/O) * +---------------+ * * DB_SEARCH is an invalid state for a dbuf. It is used by dbuf_free_range * to find all dbufs in a range of a dnode and must be less than any other * dbuf_states_t (see comment on dn_dbufs in dnode.h). */ typedef enum dbuf_states { DB_MARKER = -2, DB_SEARCH = -1, DB_UNCACHED, DB_FILL, DB_NOFILL, DB_READ, DB_CACHED, DB_EVICTING } dbuf_states_t; typedef enum dbuf_cached_state { DB_NO_CACHE = -1, DB_DBUF_CACHE, DB_DBUF_METADATA_CACHE, DB_CACHE_MAX } dbuf_cached_state_t; struct dnode; struct dmu_tx; /* * level = 0 means the user data * level = 1 means the single indirect block * etc. */ struct dmu_buf_impl; typedef enum override_states { DR_NOT_OVERRIDDEN, DR_IN_DMU_SYNC, DR_OVERRIDDEN } override_states_t; typedef enum db_lock_type { DLT_NONE, DLT_PARENT, DLT_OBJSET } db_lock_type_t; typedef struct dbuf_dirty_record { /* link on our parents dirty list */ list_node_t dr_dirty_node; /* transaction group this data will sync in */ uint64_t dr_txg; /* zio of outstanding write IO */ zio_t *dr_zio; /* pointer back to our dbuf */ struct dmu_buf_impl *dr_dbuf; /* list link for dbuf dirty records */ list_node_t dr_dbuf_node; /* * The dnode we are part of. Note that the dnode can not be moved or * evicted due to the hold that's added by dnode_setdirty() or * dmu_objset_sync_dnodes(), and released by dnode_rele_task() or * userquota_updates_task(). This hold is necessary for * dirty_lightweight_leaf-type dirty records, which don't have a hold * on a dbuf. */ dnode_t *dr_dnode; /* pointer to parent dirty record */ struct dbuf_dirty_record *dr_parent; /* How much space was changed to dsl_pool_dirty_space() for this? */ unsigned int dr_accounted; /* A copy of the bp that points to us */ blkptr_t dr_bp_copy; union dirty_types { struct dirty_indirect { /* protect access to list */ kmutex_t dr_mtx; /* Our list of dirty children */ list_t dr_children; } di; struct dirty_leaf { /* * dr_data is set when we dirty the buffer * so that we can retain the pointer even if it * gets COW'd in a subsequent transaction group. */ arc_buf_t *dr_data; override_states_t dr_override_state; uint8_t dr_copies; uint8_t dr_gang_copies; boolean_t dr_nopwrite; boolean_t dr_brtwrite; boolean_t dr_diowrite; + boolean_t dr_rewrite; boolean_t dr_has_raw_params; /* Override and raw params are mutually exclusive. */ union { blkptr_t dr_overridden_by; struct { /* * If dr_has_raw_params is set, the * following crypt params will be set * on the BP that's written. */ boolean_t dr_byteorder; uint8_t dr_salt[ZIO_DATA_SALT_LEN]; uint8_t dr_iv[ZIO_DATA_IV_LEN]; uint8_t dr_mac[ZIO_DATA_MAC_LEN]; }; }; } dl; struct dirty_lightweight_leaf { /* * This dirty record refers to a leaf (level=0) * block, whose dbuf has not been instantiated for * performance reasons. */ uint64_t dr_blkid; abd_t *dr_abd; zio_prop_t dr_props; zio_flag_t dr_flags; } dll; } dt; } dbuf_dirty_record_t; typedef struct dmu_buf_impl { /* * The following members are immutable, with the exception of * db.db_data, which is protected by db_mtx. */ /* the publicly visible structure */ dmu_buf_t db; /* the objset we belong to */ struct objset *db_objset; /* * Handle to safely access the dnode we belong to (NULL when evicted) * if dnode_move() is used on the platform, or just dnode otherwise. */ #if !defined(__linux__) && !defined(__FreeBSD__) #define USE_DNODE_HANDLE 1 struct dnode_handle *db_dnode_handle; #else struct dnode *db_dnode; #endif /* * our parent buffer; if the dnode points to us directly, * db_parent == db_dnode_handle->dnh_dnode->dn_dbuf * only accessed by sync thread ??? * (NULL when evicted) * May change from NULL to non-NULL under the protection of db_mtx * (see dbuf_check_blkptr()) */ struct dmu_buf_impl *db_parent; /* * link for hash table of all dmu_buf_impl_t's */ struct dmu_buf_impl *db_hash_next; /* * Our link on the owner dnodes's dn_dbufs list. * Protected by its dn_dbufs_mtx. Should be on the same cache line * as db_level and db_blkid for the best avl_add() performance. */ avl_node_t db_link; /* our block number */ uint64_t db_blkid; /* * Pointer to the blkptr_t which points to us. May be NULL if we * don't have one yet. (NULL when evicted) */ blkptr_t *db_blkptr; /* * Our indirection level. Data buffers have db_level==0. * Indirect buffers which point to data buffers have * db_level==1. etc. Buffers which contain dnodes have * db_level==0, since the dnodes are stored in a file. */ uint8_t db_level; /* This block was freed while a read or write was active. */ uint8_t db_freed_in_flight; /* * Evict user data as soon as the dirty and reference counts are equal. */ uint8_t db_user_immediate_evict; /* * dnode_evict_dbufs() or dnode_evict_bonus() tried to evict this dbuf, * but couldn't due to outstanding references. Evict once the refcount * drops to 0. */ uint8_t db_pending_evict; /* Number of TXGs in which this buffer is dirty. */ uint8_t db_dirtycnt; /* The buffer was partially read. More reads may follow. */ uint8_t db_partial_read; /* * Protects db_buf's contents if they contain an indirect block or data * block of the meta-dnode. We use this lock to protect the structure of * the block tree. This means that when modifying this dbuf's data, we * grab its rwlock. When modifying its parent's data (including the * blkptr to this dbuf), we grab the parent's rwlock. The lock ordering * for this lock is: * 1) dn_struct_rwlock * 2) db_rwlock * We don't currently grab multiple dbufs' db_rwlocks at once. */ krwlock_t db_rwlock; /* buffer holding our data */ arc_buf_t *db_buf; /* db_mtx protects the members below */ kmutex_t db_mtx; /* * Current state of the buffer */ dbuf_states_t db_state; /* In which dbuf cache this dbuf is, if any. */ dbuf_cached_state_t db_caching_status; /* * Refcount accessed by dmu_buf_{hold,rele}. * If nonzero, the buffer can't be destroyed. * Protected by db_mtx. */ zfs_refcount_t db_holds; kcondvar_t db_changed; dbuf_dirty_record_t *db_data_pending; /* List of dirty records for the buffer sorted newest to oldest. */ list_t db_dirty_records; /* Link in dbuf_cache or dbuf_metadata_cache */ multilist_node_t db_cache_link; uint64_t db_hash; /* User callback information. */ dmu_buf_user_t *db_user; } dmu_buf_impl_t; #define DBUF_HASH_MUTEX(h, idx) \ (&(h)->hash_mutexes[(idx) & ((h)->hash_mutex_mask)]) typedef struct dbuf_hash_table { uint64_t hash_table_mask; uint64_t hash_mutex_mask; dmu_buf_impl_t **hash_table; kmutex_t *hash_mutexes; } dbuf_hash_table_t; typedef void (*dbuf_prefetch_fn)(void *, uint64_t, uint64_t, boolean_t); extern kmem_cache_t *dbuf_dirty_kmem_cache; uint64_t dbuf_whichblock(const struct dnode *di, const int64_t level, const uint64_t offset); void dbuf_create_bonus(struct dnode *dn); int dbuf_spill_set_blksz(dmu_buf_t *db, uint64_t blksz, dmu_tx_t *tx); void dbuf_rm_spill(struct dnode *dn, dmu_tx_t *tx); dmu_buf_impl_t *dbuf_hold(struct dnode *dn, uint64_t blkid, const void *tag); dmu_buf_impl_t *dbuf_hold_level(struct dnode *dn, int level, uint64_t blkid, const void *tag); int dbuf_hold_impl(struct dnode *dn, uint8_t level, uint64_t blkid, boolean_t fail_sparse, boolean_t fail_uncached, const void *tag, dmu_buf_impl_t **dbp); int dbuf_prefetch_impl(struct dnode *dn, int64_t level, uint64_t blkid, zio_priority_t prio, arc_flags_t aflags, dbuf_prefetch_fn cb, void *arg); int dbuf_prefetch(struct dnode *dn, int64_t level, uint64_t blkid, zio_priority_t prio, arc_flags_t aflags); void dbuf_add_ref(dmu_buf_impl_t *db, const void *tag); boolean_t dbuf_try_add_ref(dmu_buf_t *db, objset_t *os, uint64_t obj, uint64_t blkid, const void *tag); uint64_t dbuf_refcount(dmu_buf_impl_t *db); void dbuf_rele(dmu_buf_impl_t *db, const void *tag); void dbuf_rele_and_unlock(dmu_buf_impl_t *db, const void *tag, boolean_t evicting); dmu_buf_impl_t *dbuf_find(struct objset *os, uint64_t object, uint8_t level, uint64_t blkid, uint64_t *hash_out); int dbuf_read(dmu_buf_impl_t *db, zio_t *zio, dmu_flags_t flags); void dmu_buf_will_clone_or_dio(dmu_buf_t *db, dmu_tx_t *tx); void dmu_buf_will_not_fill(dmu_buf_t *db, dmu_tx_t *tx); void dmu_buf_will_fill(dmu_buf_t *db, dmu_tx_t *tx, boolean_t canfail); void dmu_buf_will_fill_flags(dmu_buf_t *db, dmu_tx_t *tx, boolean_t canfail, dmu_flags_t flags); boolean_t dmu_buf_fill_done(dmu_buf_t *db, dmu_tx_t *tx, boolean_t failed); void dbuf_assign_arcbuf(dmu_buf_impl_t *db, arc_buf_t *buf, dmu_tx_t *tx, dmu_flags_t flags); dbuf_dirty_record_t *dbuf_dirty(dmu_buf_impl_t *db, dmu_tx_t *tx); dbuf_dirty_record_t *dbuf_dirty_lightweight(dnode_t *dn, uint64_t blkid, dmu_tx_t *tx); boolean_t dbuf_undirty(dmu_buf_impl_t *db, dmu_tx_t *tx); int dmu_buf_get_bp_from_dbuf(dmu_buf_impl_t *db, blkptr_t **bp); int dmu_buf_untransform_direct(dmu_buf_impl_t *db, spa_t *spa); void dmu_buf_write_embedded(dmu_buf_t *dbuf, void *data, bp_embedded_type_t etype, enum zio_compress comp, int uncompressed_size, int compressed_size, int byteorder, dmu_tx_t *tx); int dmu_lightweight_write_by_dnode(dnode_t *dn, uint64_t offset, abd_t *abd, const struct zio_prop *zp, zio_flag_t flags, dmu_tx_t *tx); void dmu_buf_redact(dmu_buf_t *dbuf, dmu_tx_t *tx); void dbuf_destroy(dmu_buf_impl_t *db); void dbuf_unoverride(dbuf_dirty_record_t *dr); void dbuf_sync_list(list_t *list, int level, dmu_tx_t *tx); void dbuf_release_bp(dmu_buf_impl_t *db); db_lock_type_t dmu_buf_lock_parent(dmu_buf_impl_t *db, krw_t rw, const void *tag); void dmu_buf_unlock_parent(dmu_buf_impl_t *db, db_lock_type_t type, const void *tag); void dbuf_free_range(struct dnode *dn, uint64_t start, uint64_t end, struct dmu_tx *); void dbuf_new_size(dmu_buf_impl_t *db, int size, dmu_tx_t *tx); void dbuf_stats_init(dbuf_hash_table_t *hash); void dbuf_stats_destroy(void); int dbuf_dnode_findbp(dnode_t *dn, uint64_t level, uint64_t blkid, blkptr_t *bp, uint16_t *datablkszsec, uint8_t *indblkshift); #ifdef USE_DNODE_HANDLE #define DB_DNODE(_db) ((_db)->db_dnode_handle->dnh_dnode) #define DB_DNODE_LOCK(_db) ((_db)->db_dnode_handle->dnh_zrlock) #define DB_DNODE_ENTER(_db) (zrl_add(&DB_DNODE_LOCK(_db))) #define DB_DNODE_EXIT(_db) (zrl_remove(&DB_DNODE_LOCK(_db))) #define DB_DNODE_HELD(_db) (!zrl_is_zero(&DB_DNODE_LOCK(_db))) #else #define DB_DNODE(_db) ((_db)->db_dnode) #define DB_DNODE_LOCK(_db) #define DB_DNODE_ENTER(_db) #define DB_DNODE_EXIT(_db) #define DB_DNODE_HELD(_db) (B_TRUE) #endif void dbuf_init(void); void dbuf_fini(void); void dbuf_cache_reduce_target_size(void); boolean_t dbuf_is_metadata(dmu_buf_impl_t *db); static inline dbuf_dirty_record_t * dbuf_find_dirty_lte(dmu_buf_impl_t *db, uint64_t txg) { dbuf_dirty_record_t *dr; for (dr = list_head(&db->db_dirty_records); dr != NULL && dr->dr_txg > txg; dr = list_next(&db->db_dirty_records, dr)) continue; return (dr); } static inline dbuf_dirty_record_t * dbuf_find_dirty_eq(dmu_buf_impl_t *db, uint64_t txg) { dbuf_dirty_record_t *dr; dr = dbuf_find_dirty_lte(db, txg); if (dr && dr->dr_txg == txg) return (dr); return (NULL); } #define DBUF_GET_BUFC_TYPE(_db) \ (dbuf_is_metadata(_db) ? ARC_BUFC_METADATA : ARC_BUFC_DATA) #define DBUF_IS_CACHEABLE(_db) (!(_db)->db_pending_evict && \ ((_db)->db_objset->os_primary_cache == ZFS_CACHE_ALL || \ (dbuf_is_metadata(_db) && \ ((_db)->db_objset->os_primary_cache == ZFS_CACHE_METADATA)))) boolean_t dbuf_is_l2cacheable(dmu_buf_impl_t *db, blkptr_t *db_bp); #ifdef ZFS_DEBUG /* * There should be a ## between the string literal and fmt, to make it * clear that we're joining two strings together, but gcc does not * support that preprocessor token. */ #define dprintf_dbuf(dbuf, fmt, ...) do { \ if (zfs_flags & ZFS_DEBUG_DPRINTF) { \ char __db_buf[32]; \ uint64_t __db_obj = (dbuf)->db.db_object; \ if (__db_obj == DMU_META_DNODE_OBJECT) \ (void) strlcpy(__db_buf, "mdn", sizeof (__db_buf)); \ else \ (void) snprintf(__db_buf, sizeof (__db_buf), "%lld", \ (u_longlong_t)__db_obj); \ dprintf_ds((dbuf)->db_objset->os_dsl_dataset, \ "obj=%s lvl=%u blkid=%lld " fmt, \ __db_buf, (dbuf)->db_level, \ (u_longlong_t)(dbuf)->db_blkid, __VA_ARGS__); \ } \ } while (0) #define dprintf_dbuf_bp(db, bp, fmt, ...) do { \ if (zfs_flags & ZFS_DEBUG_DPRINTF) { \ char *__blkbuf = kmem_alloc(BP_SPRINTF_LEN, KM_SLEEP); \ snprintf_blkptr(__blkbuf, BP_SPRINTF_LEN, bp); \ dprintf_dbuf(db, fmt " %s\n", __VA_ARGS__, __blkbuf); \ kmem_free(__blkbuf, BP_SPRINTF_LEN); \ } \ } while (0) #define DBUF_VERIFY(db) dbuf_verify(db) #else #define dprintf_dbuf(db, fmt, ...) #define dprintf_dbuf_bp(db, bp, fmt, ...) #define DBUF_VERIFY(db) #endif #ifdef __cplusplus } #endif #endif /* _SYS_DBUF_H */ diff --git a/include/sys/dmu.h b/include/sys/dmu.h index 7dc6daaf06e0..7c2024a16d8f 100644 --- a/include/sys/dmu.h +++ b/include/sys/dmu.h @@ -1,1179 +1,1180 @@ // SPDX-License-Identifier: CDDL-1.0 /* * 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 https://opensource.org/licenses/CDDL-1.0. * 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) 2012, Joyent, Inc. All rights reserved. * Copyright 2014 HybridCluster. All rights reserved. * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved. * Copyright 2013 Saso Kiselkov. All rights reserved. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2022 Hewlett Packard Enterprise Development LP. * Copyright (c) 2025, Klara, Inc. */ /* Portions Copyright 2010 Robert Milkowski */ #ifndef _SYS_DMU_H #define _SYS_DMU_H /* * This file describes the interface that the DMU provides for its * consumers. * * The DMU also interacts with the SPA. That interface is described in * dmu_spa.h. */ #include #include #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif struct page; struct vnode; struct spa; struct zilog; struct zio; struct blkptr; struct zap_cursor; struct dsl_dataset; struct dsl_pool; struct dnode; struct drr_begin; struct drr_end; struct zbookmark_phys; struct spa; struct nvlist; struct arc_buf; struct zio_prop; struct sa_handle; struct dsl_crypto_params; struct locked_range; typedef struct objset objset_t; typedef struct dmu_tx dmu_tx_t; typedef struct dsl_dir dsl_dir_t; typedef struct dnode dnode_t; typedef enum dmu_object_byteswap { DMU_BSWAP_UINT8, DMU_BSWAP_UINT16, DMU_BSWAP_UINT32, DMU_BSWAP_UINT64, DMU_BSWAP_ZAP, DMU_BSWAP_DNODE, DMU_BSWAP_OBJSET, DMU_BSWAP_ZNODE, DMU_BSWAP_OLDACL, DMU_BSWAP_ACL, /* * Allocating a new byteswap type number makes the on-disk format * incompatible with any other format that uses the same number. * * Data can usually be structured to work with one of the * DMU_BSWAP_UINT* or DMU_BSWAP_ZAP types. */ DMU_BSWAP_NUMFUNCS } dmu_object_byteswap_t; #define DMU_OT_NEWTYPE 0x80 #define DMU_OT_METADATA 0x40 #define DMU_OT_ENCRYPTED 0x20 #define DMU_OT_BYTESWAP_MASK 0x1f /* * Defines a uint8_t object type. Object types specify if the data * in the object is metadata (boolean) and how to byteswap the data * (dmu_object_byteswap_t). All of the types created by this method * are cached in the dbuf metadata cache. */ #define DMU_OT(byteswap, metadata, encrypted) \ (DMU_OT_NEWTYPE | \ ((metadata) ? DMU_OT_METADATA : 0) | \ ((encrypted) ? DMU_OT_ENCRYPTED : 0) | \ ((byteswap) & DMU_OT_BYTESWAP_MASK)) #define DMU_OT_IS_VALID(ot) (((ot) & DMU_OT_NEWTYPE) ? \ ((ot) & DMU_OT_BYTESWAP_MASK) < DMU_BSWAP_NUMFUNCS : \ (ot) < DMU_OT_NUMTYPES) #define DMU_OT_IS_METADATA_CACHED(ot) (((ot) & DMU_OT_NEWTYPE) ? \ B_TRUE : dmu_ot[(ot)].ot_dbuf_metadata_cache) /* * MDB doesn't have dmu_ot; it defines these macros itself. */ #ifndef ZFS_MDB #define DMU_OT_IS_METADATA_IMPL(ot) (dmu_ot[ot].ot_metadata) #define DMU_OT_IS_ENCRYPTED_IMPL(ot) (dmu_ot[ot].ot_encrypt) #define DMU_OT_BYTESWAP_IMPL(ot) (dmu_ot[ot].ot_byteswap) #endif #define DMU_OT_IS_METADATA(ot) (((ot) & DMU_OT_NEWTYPE) ? \ (((ot) & DMU_OT_METADATA) != 0) : \ DMU_OT_IS_METADATA_IMPL(ot)) #define DMU_OT_IS_DDT(ot) \ ((ot) == DMU_OT_DDT_ZAP) #define DMU_OT_IS_CRITICAL(ot, level) \ (DMU_OT_IS_METADATA(ot) && \ ((ot) != DMU_OT_DNODE || (level) > 0) && \ (ot) != DMU_OT_DIRECTORY_CONTENTS && \ (ot) != DMU_OT_SA) /* Note: ztest uses DMU_OT_UINT64_OTHER as a proxy for file blocks */ #define DMU_OT_IS_FILE(ot) \ ((ot) == DMU_OT_PLAIN_FILE_CONTENTS || (ot) == DMU_OT_UINT64_OTHER) #define DMU_OT_IS_ENCRYPTED(ot) (((ot) & DMU_OT_NEWTYPE) ? \ (((ot) & DMU_OT_ENCRYPTED) != 0) : \ DMU_OT_IS_ENCRYPTED_IMPL(ot)) /* * These object types use bp_fill != 1 for their L0 bp's. Therefore they can't * have their data embedded (i.e. use a BP_IS_EMBEDDED() bp), because bp_fill * is repurposed for embedded BPs. */ #define DMU_OT_HAS_FILL(ot) \ ((ot) == DMU_OT_DNODE || (ot) == DMU_OT_OBJSET) #define DMU_OT_BYTESWAP(ot) (((ot) & DMU_OT_NEWTYPE) ? \ ((ot) & DMU_OT_BYTESWAP_MASK) : \ DMU_OT_BYTESWAP_IMPL(ot)) typedef enum dmu_object_type { DMU_OT_NONE, /* general: */ DMU_OT_OBJECT_DIRECTORY, /* ZAP */ DMU_OT_OBJECT_ARRAY, /* UINT64 */ DMU_OT_PACKED_NVLIST, /* UINT8 (XDR by nvlist_pack/unpack) */ DMU_OT_PACKED_NVLIST_SIZE, /* UINT64 */ DMU_OT_BPOBJ, /* UINT64 */ DMU_OT_BPOBJ_HDR, /* UINT64 */ /* spa: */ DMU_OT_SPACE_MAP_HEADER, /* UINT64 */ DMU_OT_SPACE_MAP, /* UINT64 */ /* zil: */ DMU_OT_INTENT_LOG, /* UINT64 */ /* dmu: */ DMU_OT_DNODE, /* DNODE */ DMU_OT_OBJSET, /* OBJSET */ /* dsl: */ DMU_OT_DSL_DIR, /* UINT64 */ DMU_OT_DSL_DIR_CHILD_MAP, /* ZAP */ DMU_OT_DSL_DS_SNAP_MAP, /* ZAP */ DMU_OT_DSL_PROPS, /* ZAP */ DMU_OT_DSL_DATASET, /* UINT64 */ /* zpl: */ DMU_OT_ZNODE, /* ZNODE */ DMU_OT_OLDACL, /* Old ACL */ DMU_OT_PLAIN_FILE_CONTENTS, /* UINT8 */ DMU_OT_DIRECTORY_CONTENTS, /* ZAP */ DMU_OT_MASTER_NODE, /* ZAP */ DMU_OT_UNLINKED_SET, /* ZAP */ /* zvol: */ DMU_OT_ZVOL, /* UINT8 */ DMU_OT_ZVOL_PROP, /* ZAP */ /* other; for testing only! */ DMU_OT_PLAIN_OTHER, /* UINT8 */ DMU_OT_UINT64_OTHER, /* UINT64 */ DMU_OT_ZAP_OTHER, /* ZAP */ /* new object types: */ DMU_OT_ERROR_LOG, /* ZAP */ DMU_OT_SPA_HISTORY, /* UINT8 */ DMU_OT_SPA_HISTORY_OFFSETS, /* spa_his_phys_t */ DMU_OT_POOL_PROPS, /* ZAP */ DMU_OT_DSL_PERMS, /* ZAP */ DMU_OT_ACL, /* ACL */ DMU_OT_SYSACL, /* SYSACL */ DMU_OT_FUID, /* FUID table (Packed NVLIST UINT8) */ DMU_OT_FUID_SIZE, /* FUID table size UINT64 */ DMU_OT_NEXT_CLONES, /* ZAP */ DMU_OT_SCAN_QUEUE, /* ZAP */ DMU_OT_USERGROUP_USED, /* ZAP */ DMU_OT_USERGROUP_QUOTA, /* ZAP */ DMU_OT_USERREFS, /* ZAP */ DMU_OT_DDT_ZAP, /* ZAP */ DMU_OT_DDT_STATS, /* ZAP */ DMU_OT_SA, /* System attr */ DMU_OT_SA_MASTER_NODE, /* ZAP */ DMU_OT_SA_ATTR_REGISTRATION, /* ZAP */ DMU_OT_SA_ATTR_LAYOUTS, /* ZAP */ DMU_OT_SCAN_XLATE, /* ZAP */ DMU_OT_DEDUP, /* fake dedup BP from ddt_bp_create() */ DMU_OT_DEADLIST, /* ZAP */ DMU_OT_DEADLIST_HDR, /* UINT64 */ DMU_OT_DSL_CLONES, /* ZAP */ DMU_OT_BPOBJ_SUBOBJ, /* UINT64 */ /* * Do not allocate new object types here. Doing so makes the on-disk * format incompatible with any other format that uses the same object * type number. * * When creating an object which does not have one of the above types * use the DMU_OTN_* type with the correct byteswap and metadata * values. * * The DMU_OTN_* types do not have entries in the dmu_ot table, * use the DMU_OT_IS_METADATA() and DMU_OT_BYTESWAP() macros instead * of indexing into dmu_ot directly (this works for both DMU_OT_* types * and DMU_OTN_* types). */ DMU_OT_NUMTYPES, /* * Names for valid types declared with DMU_OT(). */ DMU_OTN_UINT8_DATA = DMU_OT(DMU_BSWAP_UINT8, B_FALSE, B_FALSE), DMU_OTN_UINT8_METADATA = DMU_OT(DMU_BSWAP_UINT8, B_TRUE, B_FALSE), DMU_OTN_UINT16_DATA = DMU_OT(DMU_BSWAP_UINT16, B_FALSE, B_FALSE), DMU_OTN_UINT16_METADATA = DMU_OT(DMU_BSWAP_UINT16, B_TRUE, B_FALSE), DMU_OTN_UINT32_DATA = DMU_OT(DMU_BSWAP_UINT32, B_FALSE, B_FALSE), DMU_OTN_UINT32_METADATA = DMU_OT(DMU_BSWAP_UINT32, B_TRUE, B_FALSE), DMU_OTN_UINT64_DATA = DMU_OT(DMU_BSWAP_UINT64, B_FALSE, B_FALSE), DMU_OTN_UINT64_METADATA = DMU_OT(DMU_BSWAP_UINT64, B_TRUE, B_FALSE), DMU_OTN_ZAP_DATA = DMU_OT(DMU_BSWAP_ZAP, B_FALSE, B_FALSE), DMU_OTN_ZAP_METADATA = DMU_OT(DMU_BSWAP_ZAP, B_TRUE, B_FALSE), DMU_OTN_UINT8_ENC_DATA = DMU_OT(DMU_BSWAP_UINT8, B_FALSE, B_TRUE), DMU_OTN_UINT8_ENC_METADATA = DMU_OT(DMU_BSWAP_UINT8, B_TRUE, B_TRUE), DMU_OTN_UINT16_ENC_DATA = DMU_OT(DMU_BSWAP_UINT16, B_FALSE, B_TRUE), DMU_OTN_UINT16_ENC_METADATA = DMU_OT(DMU_BSWAP_UINT16, B_TRUE, B_TRUE), DMU_OTN_UINT32_ENC_DATA = DMU_OT(DMU_BSWAP_UINT32, B_FALSE, B_TRUE), DMU_OTN_UINT32_ENC_METADATA = DMU_OT(DMU_BSWAP_UINT32, B_TRUE, B_TRUE), DMU_OTN_UINT64_ENC_DATA = DMU_OT(DMU_BSWAP_UINT64, B_FALSE, B_TRUE), DMU_OTN_UINT64_ENC_METADATA = DMU_OT(DMU_BSWAP_UINT64, B_TRUE, B_TRUE), DMU_OTN_ZAP_ENC_DATA = DMU_OT(DMU_BSWAP_ZAP, B_FALSE, B_TRUE), DMU_OTN_ZAP_ENC_METADATA = DMU_OT(DMU_BSWAP_ZAP, B_TRUE, B_TRUE), } dmu_object_type_t; /* * These flags are for the dmu_tx_assign() function and describe what to do if * the transaction is full. See the comment above dmu_tx_assign() for more * details on the meaning of these flags. */ typedef enum { /* * If the tx cannot be assigned to a transaction for any reason, do * not block but return immediately. */ DMU_TX_NOWAIT = 0, /* * Assign the tx to the open transaction. If the open transaction is * full, or the write throttle is active, block until the next * transaction and try again. If the pool suspends while waiting * and failmode=continue, return an error. */ DMU_TX_WAIT = (1 << 0), /* If the write throttle would prevent the assignment, ignore it. */ DMU_TX_NOTHROTTLE = (1 << 1), /* * With DMU_TX_WAIT, always block if the pool suspends during * assignment, regardless of the value of the failmode= property. */ DMU_TX_SUSPEND = (1 << 2), } dmu_tx_flag_t; void byteswap_uint64_array(void *buf, size_t size); void byteswap_uint32_array(void *buf, size_t size); void byteswap_uint16_array(void *buf, size_t size); void byteswap_uint8_array(void *buf, size_t size); void zap_byteswap(void *buf, size_t size); void zfs_oldacl_byteswap(void *buf, size_t size); void zfs_acl_byteswap(void *buf, size_t size); void zfs_znode_byteswap(void *buf, size_t size); #define DS_FIND_SNAPSHOTS (1<<0) #define DS_FIND_CHILDREN (1<<1) #define DS_FIND_SERIALIZE (1<<2) /* * The maximum number of bytes that can be accessed as part of one * operation, including metadata. */ #define DMU_MAX_ACCESS (64 * 1024 * 1024) /* 64MB */ #define DMU_MAX_DELETEBLKCNT (20480) /* ~5MB of indirect blocks */ #define DMU_USERUSED_OBJECT (-1ULL) #define DMU_GROUPUSED_OBJECT (-2ULL) #define DMU_PROJECTUSED_OBJECT (-3ULL) /* * Zap prefix for object accounting in DMU_{USER,GROUP,PROJECT}USED_OBJECT. */ #define DMU_OBJACCT_PREFIX "obj-" #define DMU_OBJACCT_PREFIX_LEN 4 /* * artificial blkids for bonus buffer and spill blocks */ #define DMU_BONUS_BLKID (-1ULL) #define DMU_SPILL_BLKID (-2ULL) /* * Public routines to create, destroy, open, and close objsets. */ typedef void dmu_objset_create_sync_func_t(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx); int dmu_objset_hold(const char *name, const void *tag, objset_t **osp); int dmu_objset_own(const char *name, dmu_objset_type_t type, boolean_t readonly, boolean_t key_required, const void *tag, objset_t **osp); void dmu_objset_rele(objset_t *os, const void *tag); void dmu_objset_disown(objset_t *os, boolean_t key_required, const void *tag); int dmu_objset_open_ds(struct dsl_dataset *ds, objset_t **osp); void dmu_objset_evict_dbufs(objset_t *os); int dmu_objset_create(const char *name, dmu_objset_type_t type, uint64_t flags, struct dsl_crypto_params *dcp, dmu_objset_create_sync_func_t func, void *arg); int dsl_destroy_snapshots_nvl(struct nvlist *snaps, boolean_t defer, struct nvlist *errlist); int dmu_objset_snapshot_one(const char *fsname, const char *snapname); int dmu_objset_find(const char *name, int func(const char *, void *), void *arg, int flags); void dmu_objset_byteswap(void *buf, size_t size); int dsl_dataset_rename_snapshot(const char *fsname, const char *oldsnapname, const char *newsnapname, boolean_t recursive); typedef struct dmu_buf { uint64_t db_object; /* object that this buffer is part of */ uint64_t db_offset; /* byte offset in this object */ uint64_t db_size; /* size of buffer in bytes */ void *db_data; /* data in buffer */ } dmu_buf_t; /* * The names of zap entries in the DIRECTORY_OBJECT of the MOS. */ #define DMU_POOL_DIRECTORY_OBJECT 1 #define DMU_POOL_CONFIG "config" #define DMU_POOL_FEATURES_FOR_WRITE "features_for_write" #define DMU_POOL_FEATURES_FOR_READ "features_for_read" #define DMU_POOL_FEATURE_DESCRIPTIONS "feature_descriptions" #define DMU_POOL_FEATURE_ENABLED_TXG "feature_enabled_txg" #define DMU_POOL_ROOT_DATASET "root_dataset" #define DMU_POOL_SYNC_BPOBJ "sync_bplist" #define DMU_POOL_ERRLOG_SCRUB "errlog_scrub" #define DMU_POOL_ERRLOG_LAST "errlog_last" #define DMU_POOL_SPARES "spares" #define DMU_POOL_DEFLATE "deflate" #define DMU_POOL_HISTORY "history" #define DMU_POOL_PROPS "pool_props" #define DMU_POOL_L2CACHE "l2cache" #define DMU_POOL_TMP_USERREFS "tmp_userrefs" #define DMU_POOL_DDT "DDT-%s-%s-%s" #define DMU_POOL_DDT_LOG "DDT-log-%s-%u" #define DMU_POOL_DDT_STATS "DDT-statistics" #define DMU_POOL_DDT_DIR "DDT-%s" #define DMU_POOL_CREATION_VERSION "creation_version" #define DMU_POOL_SCAN "scan" #define DMU_POOL_ERRORSCRUB "error_scrub" #define DMU_POOL_LAST_SCRUBBED_TXG "last_scrubbed_txg" #define DMU_POOL_FREE_BPOBJ "free_bpobj" #define DMU_POOL_BPTREE_OBJ "bptree_obj" #define DMU_POOL_EMPTY_BPOBJ "empty_bpobj" #define DMU_POOL_CHECKSUM_SALT "org.illumos:checksum_salt" #define DMU_POOL_VDEV_ZAP_MAP "com.delphix:vdev_zap_map" #define DMU_POOL_REMOVING "com.delphix:removing" #define DMU_POOL_OBSOLETE_BPOBJ "com.delphix:obsolete_bpobj" #define DMU_POOL_CONDENSING_INDIRECT "com.delphix:condensing_indirect" #define DMU_POOL_ZPOOL_CHECKPOINT "com.delphix:zpool_checkpoint" #define DMU_POOL_LOG_SPACEMAP_ZAP "com.delphix:log_spacemap_zap" #define DMU_POOL_DELETED_CLONES "com.delphix:deleted_clones" #define DMU_POOL_TXG_LOG_TIME_MINUTES "com.klaraystems:txg_log_time:minutes" #define DMU_POOL_TXG_LOG_TIME_DAYS "com.klaraystems:txg_log_time:days" #define DMU_POOL_TXG_LOG_TIME_MONTHS "com.klaraystems:txg_log_time:months" /* * Allocate an object from this objset. The range of object numbers * available is (0, DN_MAX_OBJECT). Object 0 is the meta-dnode. * * The transaction must be assigned to a txg. The newly allocated * object will be "held" in the transaction (ie. you can modify the * newly allocated object in this transaction). * * dmu_object_alloc() chooses an object and returns it in *objectp. * * dmu_object_claim() allocates a specific object number. If that * number is already allocated, it fails and returns EEXIST. * * Return 0 on success, or ENOSPC or EEXIST as specified above. */ uint64_t dmu_object_alloc(objset_t *os, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); uint64_t dmu_object_alloc_ibs(objset_t *os, dmu_object_type_t ot, int blocksize, int indirect_blockshift, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx); uint64_t dmu_object_alloc_dnsize(objset_t *os, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonus_type, int bonus_len, int dnodesize, dmu_tx_t *tx); uint64_t dmu_object_alloc_hold(objset_t *os, dmu_object_type_t ot, int blocksize, int indirect_blockshift, dmu_object_type_t bonustype, int bonuslen, int dnodesize, dnode_t **allocated_dnode, const void *tag, dmu_tx_t *tx); int dmu_object_claim(objset_t *os, uint64_t object, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); int dmu_object_claim_dnsize(objset_t *os, uint64_t object, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonus_type, int bonus_len, int dnodesize, dmu_tx_t *tx); int dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *txp); int dmu_object_reclaim_dnsize(objset_t *os, uint64_t object, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonustype, int bonuslen, int dnodesize, boolean_t keep_spill, dmu_tx_t *tx); int dmu_object_rm_spill(objset_t *os, uint64_t object, dmu_tx_t *tx); /* * Free an object from this objset. * * The object's data will be freed as well (ie. you don't need to call * dmu_free(object, 0, -1, tx)). * * The object need not be held in the transaction. * * If there are any holds on this object's buffers (via dmu_buf_hold()), * or tx holds on the object (via dmu_tx_hold_object()), you can not * free it; it fails and returns EBUSY. * * If the object is not allocated, it fails and returns ENOENT. * * Return 0 on success, or EBUSY or ENOENT as specified above. */ int dmu_object_free(objset_t *os, uint64_t object, dmu_tx_t *tx); /* * Find the next allocated or free object. * * The objectp parameter is in-out. It will be updated to be the next * object which is allocated. Ignore objects which have not been * modified since txg. * * XXX Can only be called on a objset with no dirty data. * * Returns 0 on success, or ENOENT if there are no more objects. */ int dmu_object_next(objset_t *os, uint64_t *objectp, boolean_t hole, uint64_t txg); /* * Set the number of levels on a dnode. nlevels must be greater than the * current number of levels or an EINVAL will be returned. */ int dmu_object_set_nlevels(objset_t *os, uint64_t object, int nlevels, dmu_tx_t *tx); /* * Set the data blocksize for an object. * * The object cannot have any blocks allocated beyond the first. If * the first block is allocated already, the new size must be greater * than the current block size. If these conditions are not met, * ENOTSUP will be returned. * * Returns 0 on success, or EBUSY if there are any holds on the object * contents, or ENOTSUP as described above. */ int dmu_object_set_blocksize(objset_t *os, uint64_t object, uint64_t size, int ibs, dmu_tx_t *tx); /* * Manually set the maxblkid on a dnode. This will adjust nlevels accordingly * to accommodate the change. When calling this function, the caller must * ensure that the object's nlevels can sufficiently support the new maxblkid. */ int dmu_object_set_maxblkid(objset_t *os, uint64_t object, uint64_t maxblkid, dmu_tx_t *tx); /* * Set the checksum property on a dnode. The new checksum algorithm will * apply to all newly written blocks; existing blocks will not be affected. */ void dmu_object_set_checksum(objset_t *os, uint64_t object, uint8_t checksum, dmu_tx_t *tx); /* * Set the compress property on a dnode. The new compression algorithm will * apply to all newly written blocks; existing blocks will not be affected. */ void dmu_object_set_compress(objset_t *os, uint64_t object, uint8_t compress, dmu_tx_t *tx); /* * Get an estimated cache size for an object. Caller must expect races. */ int dmu_object_cached_size(objset_t *os, uint64_t object, uint64_t *l1sz, uint64_t *l2sz); void dmu_write_embedded(objset_t *os, uint64_t object, uint64_t offset, void *data, uint8_t etype, uint8_t comp, int uncompressed_size, int compressed_size, int byteorder, dmu_tx_t *tx); void dmu_redact(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, dmu_tx_t *tx); /* * Decide how to write a block: checksum, compression, number of copies, etc. */ #define WP_NOFILL 0x1 #define WP_DMU_SYNC 0x2 #define WP_SPILL 0x4 #define WP_DIRECT_WR 0x8 void dmu_write_policy(objset_t *os, dnode_t *dn, int level, int wp, struct zio_prop *zp); /* * DB_RF_* are to be used for dbuf_read() or in limited other cases. */ typedef enum dmu_flags { DB_RF_MUST_SUCCEED = 0, /* Suspend on I/O errors. */ DB_RF_CANFAIL = 1 << 0, /* Return on I/O errors. */ DB_RF_HAVESTRUCT = 1 << 1, /* dn_struct_rwlock is locked. */ DB_RF_NEVERWAIT = 1 << 2, DMU_READ_PREFETCH = 0, /* Try speculative prefetch. */ DMU_READ_NO_PREFETCH = 1 << 3, /* Don't prefetch speculatively. */ DB_RF_NOPREFETCH = DMU_READ_NO_PREFETCH, DMU_READ_NO_DECRYPT = 1 << 4, /* Don't decrypt. */ DB_RF_NO_DECRYPT = DMU_READ_NO_DECRYPT, DMU_DIRECTIO = 1 << 5, /* Bypass ARC. */ DMU_UNCACHEDIO = 1 << 6, /* Reduce caching. */ DMU_PARTIAL_FIRST = 1 << 7, /* First partial access. */ DMU_PARTIAL_MORE = 1 << 8, /* Following partial access. */ DMU_KEEP_CACHING = 1 << 9, /* Don't affect caching. */ } dmu_flags_t; /* * The bonus data is accessed more or less like a regular buffer. * You must dmu_bonus_hold() to get the buffer, which will give you a * dmu_buf_t with db_offset==-1ULL, and db_size = the size of the bonus * data. As with any normal buffer, you must call dmu_buf_will_dirty() * before modifying it, and the * object must be held in an assigned transaction before calling * dmu_buf_will_dirty. You may use dmu_buf_set_user() on the bonus * buffer as well. You must release what you hold with dmu_buf_rele(). * * Returns ENOENT, EIO, or 0. */ int dmu_bonus_hold(objset_t *os, uint64_t object, const void *tag, dmu_buf_t **dbp); int dmu_bonus_hold_by_dnode(dnode_t *dn, const void *tag, dmu_buf_t **dbp, dmu_flags_t flags); int dmu_bonus_max(void); int dmu_set_bonus(dmu_buf_t *, int, dmu_tx_t *); int dmu_set_bonustype(dmu_buf_t *, dmu_object_type_t, dmu_tx_t *); dmu_object_type_t dmu_get_bonustype(dmu_buf_t *); int dmu_rm_spill(objset_t *, uint64_t, dmu_tx_t *); /* * Special spill buffer support used by "SA" framework */ int dmu_spill_hold_by_bonus(dmu_buf_t *bonus, dmu_flags_t flags, const void *tag, dmu_buf_t **dbp); int dmu_spill_hold_by_dnode(dnode_t *dn, dmu_flags_t flags, const void *tag, dmu_buf_t **dbp); int dmu_spill_hold_existing(dmu_buf_t *bonus, const void *tag, dmu_buf_t **dbp); /* * Obtain the DMU buffer from the specified object which contains the * specified offset. dmu_buf_hold() puts a "hold" on the buffer, so * that it will remain in memory. You must release the hold with * dmu_buf_rele(). You must not access the dmu_buf_t after releasing * what you hold. You must have a hold on any dmu_buf_t* you pass to the DMU. * * You must call dmu_buf_read, dmu_buf_will_dirty, or dmu_buf_will_fill * on the returned buffer before reading or writing the buffer's * db_data. The comments for those routines describe what particular * operations are valid after calling them. * * The object number must be a valid, allocated object number. */ int dmu_buf_hold(objset_t *os, uint64_t object, uint64_t offset, const void *tag, dmu_buf_t **, dmu_flags_t flags); int dmu_buf_hold_array(objset_t *os, uint64_t object, uint64_t offset, uint64_t length, int read, const void *tag, int *numbufsp, dmu_buf_t ***dbpp); int dmu_buf_hold_noread(objset_t *os, uint64_t object, uint64_t offset, const void *tag, dmu_buf_t **dbp); int dmu_buf_hold_by_dnode(dnode_t *dn, uint64_t offset, const void *tag, dmu_buf_t **dbp, dmu_flags_t flags); int dmu_buf_hold_array_by_dnode(dnode_t *dn, uint64_t offset, uint64_t length, boolean_t read, const void *tag, int *numbufsp, dmu_buf_t ***dbpp, dmu_flags_t flags); int dmu_buf_hold_noread_by_dnode(dnode_t *dn, uint64_t offset, const void *tag, dmu_buf_t **dbp); /* * Add a reference to a dmu buffer that has already been held via * dmu_buf_hold() in the current context. */ void dmu_buf_add_ref(dmu_buf_t *db, const void *tag); /* * Attempt to add a reference to a dmu buffer that is in an unknown state, * using a pointer that may have been invalidated by eviction processing. * The request will succeed if the passed in dbuf still represents the * same os/object/blkid, is ineligible for eviction, and has at least * one hold by a user other than the syncer. */ boolean_t dmu_buf_try_add_ref(dmu_buf_t *, objset_t *os, uint64_t object, uint64_t blkid, const void *tag); void dmu_buf_rele(dmu_buf_t *db, const void *tag); uint64_t dmu_buf_refcount(dmu_buf_t *db); uint64_t dmu_buf_user_refcount(dmu_buf_t *db); /* * dmu_buf_hold_array holds the DMU buffers which contain all bytes in a * range of an object. A pointer to an array of dmu_buf_t*'s is * returned (in *dbpp). * * dmu_buf_rele_array releases the hold on an array of dmu_buf_t*'s, and * frees the array. The hold on the array of buffers MUST be released * with dmu_buf_rele_array. You can NOT release the hold on each buffer * individually with dmu_buf_rele. */ int dmu_buf_hold_array_by_bonus(dmu_buf_t *db, uint64_t offset, uint64_t length, boolean_t read, const void *tag, int *numbufsp, dmu_buf_t ***dbpp); void dmu_buf_rele_array(dmu_buf_t **, int numbufs, const void *tag); typedef void dmu_buf_evict_func_t(void *user_ptr); /* * A DMU buffer user object may be associated with a dbuf for the * duration of its lifetime. This allows the user of a dbuf (client) * to attach private data to a dbuf (e.g. in-core only data such as a * dnode_children_t, zap_t, or zap_leaf_t) and be optionally notified * when that dbuf has been evicted. Clients typically respond to the * eviction notification by freeing their private data, thus ensuring * the same lifetime for both dbuf and private data. * * The mapping from a dmu_buf_user_t to any client private data is the * client's responsibility. All current consumers of the API with private * data embed a dmu_buf_user_t as the first member of the structure for * their private data. This allows conversions between the two types * with a simple cast. Since the DMU buf user API never needs access * to the private data, other strategies can be employed if necessary * or convenient for the client (e.g. using container_of() to do the * conversion for private data that cannot have the dmu_buf_user_t as * its first member). * * Eviction callbacks are executed without the dbuf mutex held or any * other type of mechanism to guarantee that the dbuf is still available. * For this reason, users must assume the dbuf has already been freed * and not reference the dbuf from the callback context. * * Users requesting "immediate eviction" are notified as soon as the dbuf * is only referenced by dirty records (dirties == holds). Otherwise the * notification occurs after eviction processing for the dbuf begins. */ typedef struct dmu_buf_user { /* * Asynchronous user eviction callback state. */ taskq_ent_t dbu_tqent; /* Size of user data, for inclusion in dbuf_cache accounting. */ uint64_t dbu_size; /* * This instance's eviction function pointers. * * dbu_evict_func_sync is called synchronously and then * dbu_evict_func_async is executed asynchronously on a taskq. */ dmu_buf_evict_func_t *dbu_evict_func_sync; dmu_buf_evict_func_t *dbu_evict_func_async; #ifdef ZFS_DEBUG /* * Pointer to user's dbuf pointer. NULL for clients that do * not associate a dbuf with their user data. * * The dbuf pointer is cleared upon eviction so as to catch * use-after-evict bugs in clients. */ dmu_buf_t **dbu_clear_on_evict_dbufp; #endif } dmu_buf_user_t; /* * Initialize the given dmu_buf_user_t instance with the eviction function * evict_func, to be called when the user is evicted. * * NOTE: This function should only be called once on a given dmu_buf_user_t. * To allow enforcement of this, dbu must already be zeroed on entry. */ static inline void dmu_buf_init_user(dmu_buf_user_t *dbu, dmu_buf_evict_func_t *evict_func_sync, dmu_buf_evict_func_t *evict_func_async, dmu_buf_t **clear_on_evict_dbufp __maybe_unused) { ASSERT(dbu->dbu_evict_func_sync == NULL); ASSERT(dbu->dbu_evict_func_async == NULL); /* must have at least one evict func */ IMPLY(evict_func_sync == NULL, evict_func_async != NULL); dbu->dbu_evict_func_sync = evict_func_sync; dbu->dbu_evict_func_async = evict_func_async; taskq_init_ent(&dbu->dbu_tqent); #ifdef ZFS_DEBUG dbu->dbu_clear_on_evict_dbufp = clear_on_evict_dbufp; #endif } /* * Attach user data to a dbuf and mark it for normal (when the dbuf's * data is cleared or its reference count goes to zero) eviction processing. * * Returns NULL on success, or the existing user if another user currently * owns the buffer. */ void *dmu_buf_set_user(dmu_buf_t *db, dmu_buf_user_t *user); /* * Attach user data to a dbuf and mark it for immediate (its dirty and * reference counts are equal) eviction processing. * * Returns NULL on success, or the existing user if another user currently * owns the buffer. */ void *dmu_buf_set_user_ie(dmu_buf_t *db, dmu_buf_user_t *user); /* * Replace the current user of a dbuf. * * If given the current user of a dbuf, replaces the dbuf's user with * "new_user" and returns the user data pointer that was replaced. * Otherwise returns the current, and unmodified, dbuf user pointer. */ void *dmu_buf_replace_user(dmu_buf_t *db, dmu_buf_user_t *old_user, dmu_buf_user_t *new_user); /* * Remove the specified user data for a DMU buffer. * * Returns the user that was removed on success, or the current user if * another user currently owns the buffer. */ void *dmu_buf_remove_user(dmu_buf_t *db, dmu_buf_user_t *user); /* * User data size accounting. This can be used to artifically inflate the size * of the dbuf during cache accounting, so that dbuf_evict_thread evicts enough * to satisfy memory reclaim requests. It's not used for anything else, and * defaults to 0. */ uint64_t dmu_buf_user_size(dmu_buf_t *db); void dmu_buf_add_user_size(dmu_buf_t *db, uint64_t nadd); void dmu_buf_sub_user_size(dmu_buf_t *db, uint64_t nsub); /* * Returns the user data (dmu_buf_user_t *) associated with this dbuf. */ void *dmu_buf_get_user(dmu_buf_t *db); objset_t *dmu_buf_get_objset(dmu_buf_t *db); /* Block until any in-progress dmu buf user evictions complete. */ void dmu_buf_user_evict_wait(void); /* * Returns the blkptr associated with this dbuf, or NULL if not set. */ struct blkptr *dmu_buf_get_blkptr(dmu_buf_t *db); /* * Indicate that you are going to modify the buffer's data (db_data). * * The transaction (tx) must be assigned to a txg (ie. you've called * dmu_tx_assign()). The buffer's object must be held in the tx * (ie. you've called dmu_tx_hold_object(tx, db->db_object)). */ void dmu_buf_will_dirty(dmu_buf_t *db, dmu_tx_t *tx); void dmu_buf_will_dirty_flags(dmu_buf_t *db, dmu_tx_t *tx, dmu_flags_t flags); +void dmu_buf_will_rewrite(dmu_buf_t *db, dmu_tx_t *tx); boolean_t dmu_buf_is_dirty(dmu_buf_t *db, dmu_tx_t *tx); void dmu_buf_set_crypt_params(dmu_buf_t *db_fake, boolean_t byteorder, const uint8_t *salt, const uint8_t *iv, const uint8_t *mac, dmu_tx_t *tx); /* * You must create a transaction, then hold the objects which you will * (or might) modify as part of this transaction. Then you must assign * the transaction to a transaction group. Once the transaction has * been assigned, you can modify buffers which belong to held objects as * part of this transaction. You can't modify buffers before the * transaction has been assigned; you can't modify buffers which don't * belong to objects which this transaction holds; you can't hold * objects once the transaction has been assigned. You may hold an * object which you are going to free (with dmu_object_free()), but you * don't have to. * * You can abort the transaction before it has been assigned. * * Note that you may hold buffers (with dmu_buf_hold) at any time, * regardless of transaction state. */ #define DMU_NEW_OBJECT (-1ULL) #define DMU_OBJECT_END (-1ULL) dmu_tx_t *dmu_tx_create(objset_t *os); void dmu_tx_hold_write(dmu_tx_t *tx, uint64_t object, uint64_t off, int len); void dmu_tx_hold_write_by_dnode(dmu_tx_t *tx, dnode_t *dn, uint64_t off, int len); void dmu_tx_hold_append(dmu_tx_t *tx, uint64_t object, uint64_t off, int len); void dmu_tx_hold_append_by_dnode(dmu_tx_t *tx, dnode_t *dn, uint64_t off, int len); void dmu_tx_hold_clone_by_dnode(dmu_tx_t *tx, dnode_t *dn, uint64_t off, uint64_t len, uint_t blksz); void dmu_tx_hold_free(dmu_tx_t *tx, uint64_t object, uint64_t off, uint64_t len); void dmu_tx_hold_free_by_dnode(dmu_tx_t *tx, dnode_t *dn, uint64_t off, uint64_t len); void dmu_tx_hold_zap(dmu_tx_t *tx, uint64_t object, int add, const char *name); void dmu_tx_hold_zap_by_dnode(dmu_tx_t *tx, dnode_t *dn, int add, const char *name); void dmu_tx_hold_bonus(dmu_tx_t *tx, uint64_t object); void dmu_tx_hold_bonus_by_dnode(dmu_tx_t *tx, dnode_t *dn); void dmu_tx_hold_spill(dmu_tx_t *tx, uint64_t object); void dmu_tx_hold_sa(dmu_tx_t *tx, struct sa_handle *hdl, boolean_t may_grow); void dmu_tx_hold_sa_create(dmu_tx_t *tx, int total_size); void dmu_tx_abort(dmu_tx_t *tx); int dmu_tx_assign(dmu_tx_t *tx, dmu_tx_flag_t flags); void dmu_tx_wait(dmu_tx_t *tx); void dmu_tx_commit(dmu_tx_t *tx); void dmu_tx_mark_netfree(dmu_tx_t *tx); /* * To register a commit callback, dmu_tx_callback_register() must be called. * * dcb_data is a pointer to caller private data that is passed on as a * callback parameter. The caller is responsible for properly allocating and * freeing it. * * When registering a callback, the transaction must be already created, but * it cannot be committed or aborted. It can be assigned to a txg or not. * * The callback will be called after the transaction has been safely written * to stable storage and will also be called if the dmu_tx is aborted. * If there is any error which prevents the transaction from being committed to * disk, the callback will be called with a value of error != 0. * * When multiple callbacks are registered to the transaction, the callbacks * will be called in reverse order to let Lustre, the only user of commit * callback currently, take the fast path of its commit callback handling. */ typedef void dmu_tx_callback_func_t(void *dcb_data, int error); void dmu_tx_callback_register(dmu_tx_t *tx, dmu_tx_callback_func_t *dcb_func, void *dcb_data); void dmu_tx_do_callbacks(list_t *cb_list, int error); /* * Free up the data blocks for a defined range of a file. If size is * -1, the range from offset to end-of-file is freed. */ int dmu_free_range(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, dmu_tx_t *tx); int dmu_free_long_range(objset_t *os, uint64_t object, uint64_t offset, uint64_t size); int dmu_free_long_object(objset_t *os, uint64_t object); /* * Convenience functions. * * Canfail routines will return 0 on success, or an errno if there is a * nonrecoverable I/O error. */ int dmu_read(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, void *buf, dmu_flags_t flags); int dmu_read_by_dnode(dnode_t *dn, uint64_t offset, uint64_t size, void *buf, dmu_flags_t flags); void dmu_write(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, const void *buf, dmu_tx_t *tx); int dmu_write_by_dnode(dnode_t *dn, uint64_t offset, uint64_t size, const void *buf, dmu_tx_t *tx, dmu_flags_t flags); void dmu_prealloc(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, dmu_tx_t *tx); #ifdef _KERNEL int dmu_read_uio(objset_t *os, uint64_t object, zfs_uio_t *uio, uint64_t size, dmu_flags_t flags); int dmu_read_uio_dbuf(dmu_buf_t *zdb, zfs_uio_t *uio, uint64_t size, dmu_flags_t flags); int dmu_read_uio_dnode(dnode_t *dn, zfs_uio_t *uio, uint64_t size, dmu_flags_t flags); int dmu_write_uio(objset_t *os, uint64_t object, zfs_uio_t *uio, uint64_t size, dmu_tx_t *tx, dmu_flags_t flags); int dmu_write_uio_dbuf(dmu_buf_t *zdb, zfs_uio_t *uio, uint64_t size, dmu_tx_t *tx, dmu_flags_t flags); int dmu_write_uio_dnode(dnode_t *dn, zfs_uio_t *uio, uint64_t size, dmu_tx_t *tx, dmu_flags_t flags); #endif struct arc_buf *dmu_request_arcbuf(dmu_buf_t *handle, int size); void dmu_return_arcbuf(struct arc_buf *buf); int dmu_assign_arcbuf_by_dnode(dnode_t *dn, uint64_t offset, struct arc_buf *buf, dmu_tx_t *tx, dmu_flags_t flags); int dmu_assign_arcbuf_by_dbuf(dmu_buf_t *handle, uint64_t offset, struct arc_buf *buf, dmu_tx_t *tx, dmu_flags_t flags); #define dmu_assign_arcbuf dmu_assign_arcbuf_by_dbuf extern uint_t zfs_max_recordsize; /* * Asynchronously try to read in the data. */ void dmu_prefetch(objset_t *os, uint64_t object, int64_t level, uint64_t offset, uint64_t len, enum zio_priority pri); void dmu_prefetch_by_dnode(dnode_t *dn, int64_t level, uint64_t offset, uint64_t len, enum zio_priority pri); void dmu_prefetch_dnode(objset_t *os, uint64_t object, enum zio_priority pri); int dmu_prefetch_wait(objset_t *os, uint64_t object, uint64_t offset, uint64_t size); typedef struct dmu_object_info { /* All sizes are in bytes unless otherwise indicated. */ uint32_t doi_data_block_size; uint32_t doi_metadata_block_size; dmu_object_type_t doi_type; dmu_object_type_t doi_bonus_type; uint64_t doi_bonus_size; uint8_t doi_indirection; /* 2 = dnode->indirect->data */ uint8_t doi_checksum; uint8_t doi_compress; uint8_t doi_nblkptr; uint8_t doi_pad[4]; uint64_t doi_dnodesize; uint64_t doi_physical_blocks_512; /* data + metadata, 512b blks */ uint64_t doi_max_offset; uint64_t doi_fill_count; /* number of non-empty blocks */ } dmu_object_info_t; typedef void (*const arc_byteswap_func_t)(void *buf, size_t size); typedef struct dmu_object_type_info { dmu_object_byteswap_t ot_byteswap; boolean_t ot_metadata; boolean_t ot_dbuf_metadata_cache; boolean_t ot_encrypt; const char *ot_name; } dmu_object_type_info_t; typedef const struct dmu_object_byteswap_info { arc_byteswap_func_t ob_func; const char *ob_name; } dmu_object_byteswap_info_t; extern const dmu_object_type_info_t dmu_ot[DMU_OT_NUMTYPES]; extern dmu_object_byteswap_info_t dmu_ot_byteswap[DMU_BSWAP_NUMFUNCS]; /* * Get information on a DMU object. * * Return 0 on success or ENOENT if object is not allocated. * * If doi is NULL, just indicates whether the object exists. */ int dmu_object_info(objset_t *os, uint64_t object, dmu_object_info_t *doi); void __dmu_object_info_from_dnode(struct dnode *dn, dmu_object_info_t *doi); /* Like dmu_object_info, but faster if you have a held dnode in hand. */ void dmu_object_info_from_dnode(dnode_t *dn, dmu_object_info_t *doi); /* Like dmu_object_info, but faster if you have a held dbuf in hand. */ void dmu_object_info_from_db(dmu_buf_t *db, dmu_object_info_t *doi); /* * Like dmu_object_info_from_db, but faster still when you only care about * the size. */ void dmu_object_size_from_db(dmu_buf_t *db, uint32_t *blksize, u_longlong_t *nblk512); void dmu_object_dnsize_from_db(dmu_buf_t *db, int *dnsize); typedef enum { DDS_FLAG_ENCRYPTED = (1<<0), DDS_FLAG_HAS_ENCRYPTED = (1<<7), } dmu_objset_flag_t; typedef struct dmu_objset_stats { uint64_t dds_num_clones; /* number of clones of this */ uint64_t dds_creation_txg; uint64_t dds_guid; dmu_objset_type_t dds_type; uint8_t dds_is_snapshot; uint8_t dds_inconsistent; uint8_t dds_redacted; char dds_origin[ZFS_MAX_DATASET_NAME_LEN]; uint8_t dds_flags; /* dmu_objset_flag_t */ } dmu_objset_stats_t; /* * Get stats on a dataset. */ void dmu_objset_fast_stat(objset_t *os, dmu_objset_stats_t *stat); /* * Add entries to the nvlist for all the objset's properties. See * zfs_prop_table[] and zfs(1m) for details on the properties. */ void dmu_objset_stats(objset_t *os, struct nvlist *nv); /* * Get the space usage statistics for statvfs(). * * refdbytes is the amount of space "referenced" by this objset. * availbytes is the amount of space available to this objset, taking * into account quotas & reservations, assuming that no other objsets * use the space first. These values correspond to the 'referenced' and * 'available' properties, described in the zfs(1m) manpage. * * usedobjs and availobjs are the number of objects currently allocated, * and available. */ void dmu_objset_space(objset_t *os, uint64_t *refdbytesp, uint64_t *availbytesp, uint64_t *usedobjsp, uint64_t *availobjsp); /* * The fsid_guid is a 56-bit ID that can change to avoid collisions. * (Contrast with the ds_guid which is a 64-bit ID that will never * change, so there is a small probability that it will collide.) */ uint64_t dmu_objset_fsid_guid(objset_t *os); /* * Get the [cm]time for an objset's snapshot dir */ inode_timespec_t dmu_objset_snap_cmtime(objset_t *os); int dmu_objset_is_snapshot(objset_t *os); extern struct spa *dmu_objset_spa(objset_t *os); extern struct zilog *dmu_objset_zil(objset_t *os); extern struct dsl_pool *dmu_objset_pool(objset_t *os); extern struct dsl_dataset *dmu_objset_ds(objset_t *os); extern void dmu_objset_name(objset_t *os, char *buf); extern dmu_objset_type_t dmu_objset_type(objset_t *os); extern uint64_t dmu_objset_id(objset_t *os); extern uint64_t dmu_objset_dnodesize(objset_t *os); extern zfs_sync_type_t dmu_objset_syncprop(objset_t *os); extern zfs_logbias_op_t dmu_objset_logbias(objset_t *os); extern int dmu_objset_blksize(objset_t *os); extern int dmu_snapshot_list_next(objset_t *os, int namelen, char *name, uint64_t *id, uint64_t *offp, boolean_t *case_conflict); extern int dmu_snapshot_lookup(objset_t *os, const char *name, uint64_t *val); extern int dmu_snapshot_realname(objset_t *os, const char *name, char *real, int maxlen, boolean_t *conflict); extern int dmu_dir_list_next(objset_t *os, int namelen, char *name, uint64_t *idp, uint64_t *offp); typedef struct zfs_file_info { uint64_t zfi_user; uint64_t zfi_group; uint64_t zfi_project; uint64_t zfi_generation; } zfs_file_info_t; typedef int file_info_cb_t(dmu_object_type_t bonustype, const void *data, struct zfs_file_info *zoi); extern void dmu_objset_register_type(dmu_objset_type_t ost, file_info_cb_t *cb); extern void dmu_objset_set_user(objset_t *os, void *user_ptr); extern void *dmu_objset_get_user(objset_t *os); /* * Return the txg number for the given assigned transaction. */ uint64_t dmu_tx_get_txg(dmu_tx_t *tx); /* * Synchronous write. * If a parent zio is provided this function initiates a write on the * provided buffer as a child of the parent zio. * In the absence of a parent zio, the write is completed synchronously. * At write completion, blk is filled with the bp of the written block. * Note that while the data covered by this function will be on stable * storage when the write completes this new data does not become a * permanent part of the file until the associated transaction commits. */ /* * {zfs,zvol,ztest}_get_done() args */ typedef struct zgd { struct lwb *zgd_lwb; struct blkptr *zgd_bp; dmu_buf_t *zgd_db; struct zfs_locked_range *zgd_lr; void *zgd_private; } zgd_t; typedef void dmu_sync_cb_t(zgd_t *arg, int error); int dmu_sync(struct zio *zio, uint64_t txg, dmu_sync_cb_t *done, zgd_t *zgd); /* * Find the next hole or data block in file starting at *off * Return found offset in *off. Return ESRCH for end of file. */ int dmu_offset_next(objset_t *os, uint64_t object, boolean_t hole, uint64_t *off); int dmu_read_l0_bps(objset_t *os, uint64_t object, uint64_t offset, uint64_t length, struct blkptr *bps, size_t *nbpsp); int dmu_brt_clone(objset_t *os, uint64_t object, uint64_t offset, uint64_t length, dmu_tx_t *tx, const struct blkptr *bps, size_t nbps); /* * Initial setup and final teardown. */ extern void dmu_init(void); extern void dmu_fini(void); typedef void (*dmu_traverse_cb_t)(objset_t *os, void *arg, struct blkptr *bp, uint64_t object, uint64_t offset, int len); void dmu_traverse_objset(objset_t *os, uint64_t txg_start, dmu_traverse_cb_t cb, void *arg); int dmu_diff(const char *tosnap_name, const char *fromsnap_name, zfs_file_t *fp, offset_t *offp); /* CRC64 table */ #define ZFS_CRC64_POLY 0xC96C5795D7870F42ULL /* ECMA-182, reflected form */ extern uint64_t zfs_crc64_table[256]; extern uint_t dmu_prefetch_max; #ifdef __cplusplus } #endif #endif /* _SYS_DMU_H */ diff --git a/include/sys/fs/zfs.h b/include/sys/fs/zfs.h index c8deb5be419e..fc359c10365a 100644 --- a/include/sys/fs/zfs.h +++ b/include/sys/fs/zfs.h @@ -1,2000 +1,2003 @@ // SPDX-License-Identifier: CDDL-1.0 /* * 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 https://opensource.org/licenses/CDDL-1.0. * 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, 2014, 2016, 2024 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 * Copyright (c) 2021, Colm Buckley * Copyright (c) 2022 Hewlett Packard Enterprise Development LP. * Copyright (c) 2024, Klara, Inc. */ #ifndef _SYS_FS_ZFS_H #define _SYS_FS_ZFS_H extern __attribute__((visibility("default"))) #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_INVALID = 0, 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_VDEV = (1 << 5), } 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_MAXNAMELEN_NEW 1024 #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, ZPROP_USERPROP = ZPROP_INVAL, 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_PROP_SNAPSHOTS_CHANGED, ZFS_PROP_PREFETCH, ZFS_PROP_VOLTHREADING, ZFS_PROP_DIRECT, ZFS_PROP_LONGNAME, ZFS_PROP_DEFAULTUSERQUOTA, ZFS_PROP_DEFAULTGROUPQUOTA, ZFS_PROP_DEFAULTPROJECTQUOTA, ZFS_PROP_DEFAULTUSEROBJQUOTA, ZFS_PROP_DEFAULTGROUPOBJQUOTA, ZFS_PROP_DEFAULTPROJECTOBJQUOTA, 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; _SYS_FS_ZFS_H const char *const 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_PROP_COMPATIBILITY, ZPOOL_PROP_BCLONEUSED, ZPOOL_PROP_BCLONESAVED, ZPOOL_PROP_BCLONERATIO, ZPOOL_PROP_DEDUP_TABLE_SIZE, ZPOOL_PROP_DEDUP_TABLE_QUOTA, ZPOOL_PROP_DEDUPCACHED, ZPOOL_PROP_LAST_SCRUBBED_TXG, ZPOOL_NUM_PROPS } zpool_prop_t; /* Small enough to not hog a whole line of printout in zpool(8). */ #define ZPROP_MAX_COMMENT 32 #define ZPROP_BOOLEAN_NA 2 #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 /* * VDEV 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 usr/src/common/zfs/zpool_prop.c. */ typedef enum { VDEV_PROP_INVAL = -1, VDEV_PROP_USERPROP = VDEV_PROP_INVAL, VDEV_PROP_NAME, VDEV_PROP_CAPACITY, VDEV_PROP_STATE, VDEV_PROP_GUID, VDEV_PROP_ASIZE, VDEV_PROP_PSIZE, VDEV_PROP_ASHIFT, VDEV_PROP_SIZE, VDEV_PROP_FREE, VDEV_PROP_ALLOCATED, VDEV_PROP_COMMENT, VDEV_PROP_EXPANDSZ, VDEV_PROP_FRAGMENTATION, VDEV_PROP_BOOTSIZE, VDEV_PROP_PARITY, VDEV_PROP_PATH, VDEV_PROP_DEVID, VDEV_PROP_PHYS_PATH, VDEV_PROP_ENC_PATH, VDEV_PROP_FRU, VDEV_PROP_PARENT, VDEV_PROP_CHILDREN, VDEV_PROP_NUMCHILDREN, VDEV_PROP_READ_ERRORS, VDEV_PROP_WRITE_ERRORS, VDEV_PROP_CHECKSUM_ERRORS, VDEV_PROP_INITIALIZE_ERRORS, VDEV_PROP_OPS_NULL, VDEV_PROP_OPS_READ, VDEV_PROP_OPS_WRITE, VDEV_PROP_OPS_FREE, VDEV_PROP_OPS_CLAIM, VDEV_PROP_OPS_TRIM, VDEV_PROP_BYTES_NULL, VDEV_PROP_BYTES_READ, VDEV_PROP_BYTES_WRITE, VDEV_PROP_BYTES_FREE, VDEV_PROP_BYTES_CLAIM, VDEV_PROP_BYTES_TRIM, VDEV_PROP_REMOVING, VDEV_PROP_ALLOCATING, VDEV_PROP_FAILFAST, VDEV_PROP_CHECKSUM_N, VDEV_PROP_CHECKSUM_T, VDEV_PROP_IO_N, VDEV_PROP_IO_T, VDEV_PROP_RAIDZ_EXPANDING, VDEV_PROP_SLOW_IO_N, VDEV_PROP_SLOW_IO_T, VDEV_PROP_TRIM_SUPPORT, VDEV_PROP_TRIM_ERRORS, VDEV_PROP_SLOW_IOS, VDEV_NUM_PROPS } vdev_prop_t; /* * Dataset property functions shared between libzfs and kernel. */ _SYS_FS_ZFS_H const char *zfs_prop_default_string(zfs_prop_t); _SYS_FS_ZFS_H uint64_t zfs_prop_default_numeric(zfs_prop_t); _SYS_FS_ZFS_H boolean_t zfs_prop_readonly(zfs_prop_t); _SYS_FS_ZFS_H boolean_t zfs_prop_visible(zfs_prop_t prop); _SYS_FS_ZFS_H boolean_t zfs_prop_inheritable(zfs_prop_t); _SYS_FS_ZFS_H boolean_t zfs_prop_setonce(zfs_prop_t); _SYS_FS_ZFS_H boolean_t zfs_prop_encryption_key_param(zfs_prop_t); _SYS_FS_ZFS_H boolean_t zfs_prop_valid_keylocation(const char *, boolean_t); _SYS_FS_ZFS_H const char *zfs_prop_to_name(zfs_prop_t); _SYS_FS_ZFS_H zfs_prop_t zfs_name_to_prop(const char *); _SYS_FS_ZFS_H boolean_t zfs_prop_user(const char *); _SYS_FS_ZFS_H boolean_t zfs_prop_userquota(const char *); _SYS_FS_ZFS_H boolean_t zfs_prop_written(const char *); _SYS_FS_ZFS_H int zfs_prop_index_to_string(zfs_prop_t, uint64_t, const char **); _SYS_FS_ZFS_H int zfs_prop_string_to_index(zfs_prop_t, const char *, uint64_t *); _SYS_FS_ZFS_H uint64_t zfs_prop_random_value(zfs_prop_t, uint64_t seed); _SYS_FS_ZFS_H boolean_t zfs_prop_valid_for_type(int, zfs_type_t, boolean_t); /* * Pool property functions shared between libzfs and kernel. */ _SYS_FS_ZFS_H zpool_prop_t zpool_name_to_prop(const char *); _SYS_FS_ZFS_H const char *zpool_prop_to_name(zpool_prop_t); _SYS_FS_ZFS_H const char *zpool_prop_default_string(zpool_prop_t); _SYS_FS_ZFS_H uint64_t zpool_prop_default_numeric(zpool_prop_t); _SYS_FS_ZFS_H boolean_t zpool_prop_readonly(zpool_prop_t); _SYS_FS_ZFS_H boolean_t zpool_prop_setonce(zpool_prop_t); _SYS_FS_ZFS_H boolean_t zpool_prop_feature(const char *); _SYS_FS_ZFS_H boolean_t zpool_prop_unsupported(const char *); _SYS_FS_ZFS_H int zpool_prop_index_to_string(zpool_prop_t, uint64_t, const char **); _SYS_FS_ZFS_H int zpool_prop_string_to_index(zpool_prop_t, const char *, uint64_t *); _SYS_FS_ZFS_H uint64_t zpool_prop_random_value(zpool_prop_t, uint64_t seed); /* * VDEV property functions shared between libzfs and kernel. */ _SYS_FS_ZFS_H vdev_prop_t vdev_name_to_prop(const char *); _SYS_FS_ZFS_H boolean_t vdev_prop_user(const char *name); _SYS_FS_ZFS_H const char *vdev_prop_to_name(vdev_prop_t); _SYS_FS_ZFS_H const char *vdev_prop_default_string(vdev_prop_t); _SYS_FS_ZFS_H uint64_t vdev_prop_default_numeric(vdev_prop_t); _SYS_FS_ZFS_H boolean_t vdev_prop_readonly(vdev_prop_t prop); _SYS_FS_ZFS_H int vdev_prop_index_to_string(vdev_prop_t, uint64_t, const char **); _SYS_FS_ZFS_H int vdev_prop_string_to_index(vdev_prop_t, const char *, uint64_t *); _SYS_FS_ZFS_H boolean_t zpool_prop_vdev(const char *name); _SYS_FS_ZFS_H uint64_t vdev_prop_random_value(vdev_prop_t prop, 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_SOME, ZFS_REDUNDANT_METADATA_NONE } 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_DIRECT_DISABLED = 0, ZFS_DIRECT_STANDARD, ZFS_DIRECT_ALWAYS } zfs_direct_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; typedef enum { ZFS_PREFETCH_NONE = 0, ZFS_PREFETCH_METADATA = 1, ZFS_PREFETCH_ALL = 2 } zfs_prefetch_type_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_SPARE_ID "spareid" #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_RAIDZ_EXPAND_STATS "raidz_expand_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" #define ZPOOL_CONFIG_VDEV_REBUILD_ACTIVE_QUEUE "vdev_rebuild_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" #define ZPOOL_CONFIG_VDEV_REBUILD_PEND_QUEUE "vdev_rebuild_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" #define ZPOOL_CONFIG_VDEV_REBUILD_LAT_HISTO "vdev_rebuild_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_IND_REBUILD_HISTO "vdev_ind_rebuild_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" #define ZPOOL_CONFIG_VDEV_AGG_REBUILD_HISTO "vdev_agg_rebuild_histo" /* Number of slow IOs */ #define ZPOOL_CONFIG_VDEV_SLOW_IOS "vdev_slow_ios" /* Number of Direct I/O write verify errors */ #define ZPOOL_CONFIG_VDEV_DIO_VERIFY_ERRORS "vdev_dio_verify_errors" /* 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_RAIDZ_EXPANDING "raidz_expanding" #define ZPOOL_CONFIG_RAIDZ_EXPAND_TXGS "raidz_expand_txgs" #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_NONALLOCATING "non_allocating" #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_ROOT_ZAP "com.klarasystems:vdev_zap_root" #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" #define ZPOOL_CONFIG_COMPATIBILITY "compatibility" /* * 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_META_ERRORS "verify_meta_errors" #define ZPOOL_CONFIG_LOAD_DATA_ERRORS "verify_data_errors" #define ZPOOL_CONFIG_REWIND_TIME "seconds_of_rewind" /* dRAID configuration */ #define ZPOOL_CONFIG_DRAID_NDATA "draid_ndata" #define ZPOOL_CONFIG_DRAID_NSPARES "draid_nspares" #define ZPOOL_CONFIG_DRAID_NGROUPS "draid_ngroups" #define VDEV_TYPE_ROOT "root" #define VDEV_TYPE_MIRROR "mirror" #define VDEV_TYPE_REPLACING "replacing" #define VDEV_TYPE_RAIDZ "raidz" #define VDEV_TYPE_DRAID "draid" #define VDEV_TYPE_DRAID_SPARE "dspare" #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" #define VDEV_RAIDZ_MAXPARITY 3 #define VDEV_DRAID_MAXPARITY 3 #define VDEV_DRAID_MIN_CHILDREN 2 #define VDEV_DRAID_MAX_CHILDREN UINT8_MAX /* 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" #define VDEV_TOP_ZAP_RAIDZ_EXPAND_STATE \ "org.openzfs:raidz_expand_state" #define VDEV_TOP_ZAP_RAIDZ_EXPAND_START_TIME \ "org.openzfs:raidz_expand_start_time" #define VDEV_TOP_ZAP_RAIDZ_EXPAND_END_TIME \ "org.openzfs:raidz_expand_end_time" #define VDEV_TOP_ZAP_RAIDZ_EXPAND_BYTES_COPIED \ "org.openzfs:raidz_expand_bytes_copied" /* 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" /* * Settings for zpool compatibility features files */ #define ZPOOL_SYSCONF_COMPAT_D SYSCONFDIR "/zfs/compatibility.d" #define ZPOOL_DATA_COMPAT_D PKGDATADIR "/compatibility.d" #define ZPOOL_COMPAT_MAXSIZE 16384 /* * Hard-wired compatibility settings */ #define ZPOOL_COMPAT_LEGACY "legacy" #define ZPOOL_COMPAT_OFF "off" /* * 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_ERRORSCRUB, 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_FROM_LAST_TXG, 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_FLUSH, ZIO_TYPE_TRIM, ZIO_TYPES } zio_type_t; /* * Compatibility: _IOCTL was renamed to _FLUSH; keep the old name available to * user programs. */ #define ZIO_TYPE_IOCTL ZIO_TYPE_FLUSH /* * ZIO priority types. Needed to interpret vdev statistics below. * * NOTE: PLEASE UPDATE THE ENUM STRINGS IN zfs_valstr.c IF YOU ADD ANOTHER * VALUE. */ typedef enum zio_priority { ZIO_PRIORITY_SYNC_READ, ZIO_PRIORITY_SYNC_WRITE, /* ZIL */ ZIO_PRIORITY_ASYNC_READ, /* prefetch */ ZIO_PRIORITY_ASYNC_WRITE, /* spa_sync() */ ZIO_PRIORITY_SCRUB, /* asynchronous scrub/resilver reads */ ZIO_PRIORITY_REMOVAL, /* reads/writes for vdev removal */ ZIO_PRIORITY_INITIALIZING, /* initializing I/O */ ZIO_PRIORITY_TRIM, /* trim I/O (discard) */ ZIO_PRIORITY_REBUILD, /* reads/writes for vdev rebuild */ ZIO_PRIORITY_NUM_QUEUEABLE, ZIO_PRIORITY_NOW, /* non-queued i/os (e.g. free) */ } zio_priority_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_skipped; /* total bytes skipped by scanner */ 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 */ /* error scrub values stored on disk */ uint64_t pss_error_scrub_func; /* pool_scan_func_t */ uint64_t pss_error_scrub_state; /* dsl_scan_state_t */ uint64_t pss_error_scrub_start; /* error scrub start time */ uint64_t pss_error_scrub_end; /* error scrub end time */ uint64_t pss_error_scrub_examined; /* error blocks issued I/O */ /* error blocks to be issued I/O */ uint64_t pss_error_scrub_to_be_examined; /* error scrub values not stored on disk */ /* error scrub pause time in milliseconds */ uint64_t pss_pass_error_scrub_pause; } 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 struct pool_raidz_expand_stat { uint64_t pres_state; /* dsl_scan_state_t */ uint64_t pres_expanding_vdev; uint64_t pres_start_time; uint64_t pres_end_time; uint64_t pres_to_reflow; /* bytes that need to be moved */ uint64_t pres_reflowed; /* bytes moved so far */ uint64_t pres_waiting_for_resilver; } pool_raidz_expand_stat_t; typedef enum dsl_scan_state { DSS_NONE, DSS_SCANNING, DSS_FINISHED, DSS_CANCELED, DSS_ERRORSCRUBBING, 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 */ uint64_t vrs_pass_bytes_skipped; /* bytes skipped 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 */ uint64_t vs_noalloc; /* allocations halted? */ uint64_t vs_pspace; /* physical capacity */ uint64_t vs_dio_verify_errors; /* DIO write verify errors */ } vdev_stat_t; #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))) /* * 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_UNINIT, 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_DEVDIR "/dev" #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 16384 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 - 89/128 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 */ ZFS_IOC_VDEV_GET_PROPS, /* 0x5a55 */ ZFS_IOC_VDEV_SET_PROPS, /* 0x5a56 */ ZFS_IOC_POOL_SCRUB, /* 0x5a57 */ ZFS_IOC_POOL_PREFETCH, /* 0x5a58 */ ZFS_IOC_DDT_PRUNE, /* 0x5a59 */ /* * 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_USERNS_ATTACH = ZFS_IOC_JAIL, /* 0x85 (Linux) */ ZFS_IOC_UNJAIL, /* 0x86 (FreeBSD) */ ZFS_IOC_USERNS_DETACH = ZFS_IOC_UNJAIL, /* 0x86 (Linux) */ 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]) #ifdef __linux__ /* * IOCTLs to update and retrieve additional file level attributes on * Linux. */ #define ZFS_IOC_GETDOSFLAGS _IOR(0x83, 1, uint64_t) #define ZFS_IOC_SETDOSFLAGS _IOW(0x83, 2, uint64_t) /* * Additional file level attributes, that are stored * in the upper half of z_pflags */ #define ZFS_READONLY 0x0000000100000000ull #define ZFS_HIDDEN 0x0000000200000000ull #define ZFS_SYSTEM 0x0000000400000000ull #define ZFS_ARCHIVE 0x0000000800000000ull #define ZFS_IMMUTABLE 0x0000001000000000ull #define ZFS_NOUNLINK 0x0000002000000000ull #define ZFS_APPENDONLY 0x0000004000000000ull #define ZFS_NODUMP 0x0000008000000000ull #define ZFS_OPAQUE 0x0000010000000000ull #define ZFS_AV_QUARANTINED 0x0000020000000000ull #define ZFS_AV_MODIFIED 0x0000040000000000ull #define ZFS_REPARSE 0x0000080000000000ull #define ZFS_OFFLINE 0x0000100000000000ull #define ZFS_SPARSE 0x0000200000000000ull #define ZFS_DOS_FL_USER_VISIBLE (ZFS_IMMUTABLE | ZFS_APPENDONLY | \ ZFS_NOUNLINK | ZFS_ARCHIVE | ZFS_NODUMP | ZFS_SYSTEM | \ ZFS_HIDDEN | ZFS_READONLY | ZFS_REPARSE | ZFS_OFFLINE | \ ZFS_SPARSE) #endif typedef struct zfs_rewrite_args { uint64_t off; uint64_t len; uint64_t flags; uint64_t arg; } zfs_rewrite_args_t; +/* zfs_rewrite_args flags */ +#define ZFS_REWRITE_PHYSICAL 0x1 /* Preserve logical birth time. */ + #define ZFS_IOC_REWRITE _IOW(0x83, 3, zfs_rewrite_args_t) /* * 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_ERR_VDEV_NOTSUP, ZFS_ERR_NOT_USER_NAMESPACE, ZFS_ERR_RESUME_EXISTS, ZFS_ERR_CRYPTO_NOTSUP, ZFS_ERR_RAIDZ_EXPAND_IN_PROGRESS, ZFS_ERR_ASHIFT_MISMATCH, ZFS_ERR_STREAM_LARGE_MICROZAP, } 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_RAIDZ_EXPAND, ZPOOL_WAIT_NUM_ACTIVITIES } zpool_wait_activity_t; typedef enum { ZFS_WAIT_DELETEQ, ZFS_WAIT_NUM_ACTIVITIES } zfs_wait_activity_t; typedef enum { ZPOOL_PREFETCH_NONE = 0, ZPOOL_PREFETCH_DDT } zpool_prefetch_type_t; typedef enum { ZPOOL_DDT_PRUNE_NONE, ZPOOL_DDT_PRUNE_AGE, /* in seconds */ ZPOOL_DDT_PRUNE_PERCENTAGE, /* 1 - 100 */ } zpool_ddt_prune_unit_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_OUTPUT_SIZE "out_size" #define ZPOOL_HIST_DSNAME "dsname" #define ZPOOL_HIST_DSID "dsid" #define ZPOOL_HIST_ERRNO "errno" #define ZPOOL_HIST_ELAPSED_NS "elapsed_ns" /* * Special nvlist name that will not have its args recorded in the pool's * history log. */ #define ZPOOL_HIDDEN_ARGS "hidden_args" /* * The following is used when invoking ZFS_IOC_POOL_GET_PROPS. */ #define ZPOOL_GET_PROPS_NAMES "get_props_names" /* * Opt-in property names used with ZPOOL_GET_PROPS_NAMES. * For example, properties that are hidden or expensive to compute. */ #define ZPOOL_DEDUPCACHED_PROP_NAME "dedupcached" /* * 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_REGUID. */ #define ZPOOL_REGUID_GUID "guid" /* * 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_VDEV_GET_PROP. */ #define ZPOOL_VDEV_PROPS_GET_VDEV "vdevprops_get_vdev" #define ZPOOL_VDEV_PROPS_GET_PROPS "vdevprops_get_props" /* * The following are names used when invoking ZFS_IOC_VDEV_SET_PROP. */ #define ZPOOL_VDEV_PROPS_SET_VDEV "vdevprops_set_vdev" #define ZPOOL_VDEV_PROPS_SET_PROPS "vdevprops_set_props" /* * The following are names used when invoking ZFS_IOC_WAIT_FS. */ #define ZFS_WAIT_ACTIVITY "wait_activity" #define ZFS_WAIT_WAITED "wait_waited" /* * The following are names used when invoking ZFS_IOC_POOL_PREFETCH. */ #define ZPOOL_PREFETCH_TYPE "prefetch_type" /* * The following are names used when invoking ZFS_IOC_DDT_PRUNE. */ #define DDT_PRUNE_UNIT "ddt_prune_unit" #define DDT_PRUNE_AMOUNT "ddt_prune_amount" /* * 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_ONLINE_SPARE 0x10 #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" /* * We currently support block sizes from 512 bytes to 16MB. * The benefits of larger blocks, and thus larger IO, need to be weighed * against the cost of COWing a giant block to modify one byte, and the * large latency of reading or writing a large block. * * The recordsize property can not be set larger than zfs_max_recordsize * (default 16MB on 64-bit and 1MB on 32-bit). See the comment near * zfs_max_recordsize in dsl_dataset.c for details. * * Note that although the LSIZE field of the blkptr_t can store sizes up * to 32MB, the dnode's dn_datablkszsec can only store sizes up to * 32MB - 512 bytes. Therefore, we limit SPA_MAXBLOCKSIZE to 16MB. */ #define SPA_MINBLOCKSHIFT 9 #define SPA_OLD_MAXBLOCKSHIFT 17 #define SPA_MAXBLOCKSHIFT 24 #define SPA_MINBLOCKSIZE (1ULL << SPA_MINBLOCKSHIFT) #define SPA_OLD_MAXBLOCKSIZE (1ULL << SPA_OLD_MAXBLOCKSHIFT) #define SPA_MAXBLOCKSIZE (1ULL << SPA_MAXBLOCKSHIFT) /* supported encryption algorithms */ enum zio_encrypt { ZIO_CRYPT_INHERIT = 0, ZIO_CRYPT_ON, ZIO_CRYPT_OFF, ZIO_CRYPT_AES_128_CCM, ZIO_CRYPT_AES_192_CCM, ZIO_CRYPT_AES_256_CCM, ZIO_CRYPT_AES_128_GCM, ZIO_CRYPT_AES_192_GCM, ZIO_CRYPT_AES_256_GCM, ZIO_CRYPT_FUNCTIONS }; #define ZIO_CRYPT_ON_VALUE ZIO_CRYPT_AES_256_GCM #define ZIO_CRYPT_DEFAULT ZIO_CRYPT_OFF /* * xattr namespace prefixes. These are forbidden in xattr names. * * For cross-platform compatibility, xattrs in the user namespace should not be * prefixed with the namespace name, but for backwards compatibility with older * ZFS on Linux versions we do prefix the namespace. */ #define ZFS_XA_NS_FREEBSD_PREFIX "freebsd:" #define ZFS_XA_NS_FREEBSD_PREFIX_LEN strlen("freebsd:") #define ZFS_XA_NS_LINUX_SECURITY_PREFIX "security." #define ZFS_XA_NS_LINUX_SECURITY_PREFIX_LEN strlen("security.") #define ZFS_XA_NS_LINUX_SYSTEM_PREFIX "system." #define ZFS_XA_NS_LINUX_SYSTEM_PREFIX_LEN strlen("system.") #define ZFS_XA_NS_LINUX_TRUSTED_PREFIX "trusted." #define ZFS_XA_NS_LINUX_TRUSTED_PREFIX_LEN strlen("trusted.") #define ZFS_XA_NS_LINUX_USER_PREFIX "user." #define ZFS_XA_NS_LINUX_USER_PREFIX_LEN strlen("user.") #define ZFS_XA_NS_PREFIX_MATCH(ns, name) \ (strncmp(name, ZFS_XA_NS_##ns##_PREFIX, \ ZFS_XA_NS_##ns##_PREFIX_LEN) == 0) #define ZFS_XA_NS_PREFIX_FORBIDDEN(name) \ (ZFS_XA_NS_PREFIX_MATCH(FREEBSD, name) || \ ZFS_XA_NS_PREFIX_MATCH(LINUX_SECURITY, name) || \ ZFS_XA_NS_PREFIX_MATCH(LINUX_SYSTEM, name) || \ ZFS_XA_NS_PREFIX_MATCH(LINUX_TRUSTED, name) || \ ZFS_XA_NS_PREFIX_MATCH(LINUX_USER, name)) #ifdef __cplusplus } #endif #endif /* _SYS_FS_ZFS_H */ diff --git a/include/sys/zio.h b/include/sys/zio.h index b139c9de4852..a3368034695b 100644 --- a/include/sys/zio.h +++ b/include/sys/zio.h @@ -1,755 +1,756 @@ // SPDX-License-Identifier: CDDL-1.0 /* * 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 https://opensource.org/licenses/CDDL-1.0. * 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 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2012, 2024 by Delphix. All rights reserved. * Copyright (c) 2013 by Saso Kiselkov. All rights reserved. * Copyright (c) 2013, Joyent, Inc. All rights reserved. * Copyright 2016 Toomas Soome * Copyright (c) 2019, Allan Jude * Copyright (c) 2019, 2023, 2024, Klara Inc. * Copyright (c) 2019-2020, Michael Niewöhner * Copyright (c) 2024 by George Melikov. All rights reserved. */ #ifndef _ZIO_H #define _ZIO_H #include #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif /* * Embedded checksum */ #define ZEC_MAGIC 0x210da7ab10c7a11ULL typedef struct zio_eck { uint64_t zec_magic; /* for validation, endianness */ zio_cksum_t zec_cksum; /* 256-bit checksum */ } zio_eck_t; /* * Gang block headers are self-checksumming and contain an array * of block pointers. The old gang block size has enough room for 3 blkptrs, * while new gang blocks can store more. * * Layout: * +--------+--------+--------+-----+---------+-----------+ * | | | | | | | * | blkptr | blkptr | blkptr | ... | padding | zio_eck_t | * | 1 | 2 | 3 | | | | * +--------+--------+--------+-----+---------+-----------+ * 128B 128B 128B 88B 40B */ #define SPA_OLD_GANGBLOCKSIZE SPA_MINBLOCKSIZE typedef void zio_gbh_phys_t; static inline uint64_t gbh_nblkptrs(uint64_t size) { ASSERT(IS_P2ALIGNED(size, sizeof (blkptr_t))); return ((size - sizeof (zio_eck_t)) / sizeof (blkptr_t)); } static inline zio_eck_t * gbh_eck(zio_gbh_phys_t *gbh, uint64_t size) { ASSERT(IS_P2ALIGNED(size, sizeof (blkptr_t))); return ((zio_eck_t *)((uintptr_t)gbh + size - sizeof (zio_eck_t))); } static inline blkptr_t * gbh_bp(zio_gbh_phys_t *gbh, int bp) { return (&((blkptr_t *)gbh)[bp]); } enum zio_checksum { ZIO_CHECKSUM_INHERIT = 0, ZIO_CHECKSUM_ON, ZIO_CHECKSUM_OFF, ZIO_CHECKSUM_LABEL, ZIO_CHECKSUM_GANG_HEADER, ZIO_CHECKSUM_ZILOG, ZIO_CHECKSUM_FLETCHER_2, ZIO_CHECKSUM_FLETCHER_4, ZIO_CHECKSUM_SHA256, ZIO_CHECKSUM_ZILOG2, ZIO_CHECKSUM_NOPARITY, ZIO_CHECKSUM_SHA512, ZIO_CHECKSUM_SKEIN, ZIO_CHECKSUM_EDONR, ZIO_CHECKSUM_BLAKE3, ZIO_CHECKSUM_FUNCTIONS }; /* * The number of "legacy" compression functions which can be set on individual * objects. */ #define ZIO_CHECKSUM_LEGACY_FUNCTIONS ZIO_CHECKSUM_ZILOG2 #define ZIO_CHECKSUM_ON_VALUE ZIO_CHECKSUM_FLETCHER_4 #define ZIO_CHECKSUM_DEFAULT ZIO_CHECKSUM_ON #define ZIO_CHECKSUM_MASK 0xffULL #define ZIO_CHECKSUM_VERIFY (1U << 8) #define ZIO_DEDUPCHECKSUM ZIO_CHECKSUM_SHA256 /* macros defining encryption lengths */ #define ZIO_OBJSET_MAC_LEN 32 #define ZIO_DATA_IV_LEN 12 #define ZIO_DATA_SALT_LEN 8 #define ZIO_DATA_MAC_LEN 16 /* * The number of "legacy" compression functions which can be set on individual * objects. */ #define ZIO_COMPRESS_LEGACY_FUNCTIONS ZIO_COMPRESS_LZ4 /* * The meaning of "compress = on" selected by the compression features enabled * on a given pool. */ #define ZIO_COMPRESS_LEGACY_ON_VALUE ZIO_COMPRESS_LZJB #define ZIO_COMPRESS_LZ4_ON_VALUE ZIO_COMPRESS_LZ4 #define ZIO_COMPRESS_DEFAULT ZIO_COMPRESS_ON #define BOOTFS_COMPRESS_VALID(compress) \ ((compress) == ZIO_COMPRESS_LZJB || \ (compress) == ZIO_COMPRESS_LZ4 || \ (compress) == ZIO_COMPRESS_GZIP_1 || \ (compress) == ZIO_COMPRESS_GZIP_2 || \ (compress) == ZIO_COMPRESS_GZIP_3 || \ (compress) == ZIO_COMPRESS_GZIP_4 || \ (compress) == ZIO_COMPRESS_GZIP_5 || \ (compress) == ZIO_COMPRESS_GZIP_6 || \ (compress) == ZIO_COMPRESS_GZIP_7 || \ (compress) == ZIO_COMPRESS_GZIP_8 || \ (compress) == ZIO_COMPRESS_GZIP_9 || \ (compress) == ZIO_COMPRESS_ZLE || \ (compress) == ZIO_COMPRESS_ZSTD || \ (compress) == ZIO_COMPRESS_ON || \ (compress) == ZIO_COMPRESS_OFF) #define ZIO_COMPRESS_ALGO(x) (x & SPA_COMPRESSMASK) #define ZIO_COMPRESS_LEVEL(x) ((x & ~SPA_COMPRESSMASK) >> SPA_COMPRESSBITS) #define ZIO_COMPRESS_RAW(type, level) (type | ((level) << SPA_COMPRESSBITS)) #define ZIO_COMPLEVEL_ZSTD(level) \ ZIO_COMPRESS_RAW(ZIO_COMPRESS_ZSTD, level) #define ZIO_FAILURE_MODE_WAIT 0 #define ZIO_FAILURE_MODE_CONTINUE 1 #define ZIO_FAILURE_MODE_PANIC 2 typedef enum zio_suspend_reason { ZIO_SUSPEND_NONE = 0, ZIO_SUSPEND_IOERR, ZIO_SUSPEND_MMP, } zio_suspend_reason_t; /* * This was originally an enum type. However, those are 32-bit and there is no * way to make a 64-bit enum type. Since we ran out of bits for flags, we were * forced to upgrade it to a uint64_t. * * NOTE: PLEASE UPDATE THE BITFIELD STRINGS IN zfs_valstr.c IF YOU ADD ANOTHER * FLAG. */ typedef uint64_t zio_flag_t; /* * Flags inherited by gang, ddt, and vdev children, * and that must be equal for two zios to aggregate */ #define ZIO_FLAG_DONT_AGGREGATE (1ULL << 0) #define ZIO_FLAG_IO_REPAIR (1ULL << 1) #define ZIO_FLAG_SELF_HEAL (1ULL << 2) #define ZIO_FLAG_RESILVER (1ULL << 3) #define ZIO_FLAG_SCRUB (1ULL << 4) #define ZIO_FLAG_SCAN_THREAD (1ULL << 5) #define ZIO_FLAG_PHYSICAL (1ULL << 6) #define ZIO_FLAG_AGG_INHERIT (ZIO_FLAG_CANFAIL - 1) /* * Flags inherited by ddt, gang, and vdev children. */ #define ZIO_FLAG_CANFAIL (1ULL << 7) /* must be first for INHERIT */ #define ZIO_FLAG_SPECULATIVE (1ULL << 8) #define ZIO_FLAG_CONFIG_WRITER (1ULL << 9) #define ZIO_FLAG_DONT_RETRY (1ULL << 10) #define ZIO_FLAG_NODATA (1ULL << 12) #define ZIO_FLAG_INDUCE_DAMAGE (1ULL << 13) #define ZIO_FLAG_ALLOC_THROTTLED (1ULL << 14) #define ZIO_FLAG_DDT_INHERIT (ZIO_FLAG_IO_RETRY - 1) #define ZIO_FLAG_GANG_INHERIT (ZIO_FLAG_IO_RETRY - 1) /* * Flags inherited by vdev children. */ #define ZIO_FLAG_IO_RETRY (1ULL << 15) /* must be first for INHERIT */ #define ZIO_FLAG_PROBE (1ULL << 16) #define ZIO_FLAG_TRYHARD (1ULL << 17) #define ZIO_FLAG_OPTIONAL (1ULL << 18) #define ZIO_FLAG_DIO_READ (1ULL << 19) #define ZIO_FLAG_VDEV_INHERIT (ZIO_FLAG_DONT_QUEUE - 1) /* * Flags not inherited by any children. */ #define ZIO_FLAG_DONT_QUEUE (1ULL << 20) /* must be first for INHERIT */ #define ZIO_FLAG_DONT_PROPAGATE (1ULL << 21) #define ZIO_FLAG_IO_BYPASS (1ULL << 22) #define ZIO_FLAG_IO_REWRITE (1ULL << 23) #define ZIO_FLAG_RAW_COMPRESS (1ULL << 24) #define ZIO_FLAG_RAW_ENCRYPT (1ULL << 25) #define ZIO_FLAG_GANG_CHILD (1ULL << 26) #define ZIO_FLAG_DDT_CHILD (1ULL << 27) #define ZIO_FLAG_GODFATHER (1ULL << 28) #define ZIO_FLAG_NOPWRITE (1ULL << 29) #define ZIO_FLAG_REEXECUTED (1ULL << 30) #define ZIO_FLAG_DELEGATED (1ULL << 31) #define ZIO_FLAG_PREALLOCATED (1ULL << 32) #define ZIO_ALLOCATOR_NONE (-1) #define ZIO_HAS_ALLOCATOR(zio) ((zio)->io_allocator != ZIO_ALLOCATOR_NONE) #define ZIO_FLAG_MUSTSUCCEED 0 #define ZIO_FLAG_RAW (ZIO_FLAG_RAW_COMPRESS | ZIO_FLAG_RAW_ENCRYPT) #define ZIO_DDT_CHILD_FLAGS(zio) \ (((zio)->io_flags & ZIO_FLAG_DDT_INHERIT) | \ ZIO_FLAG_DDT_CHILD | ZIO_FLAG_CANFAIL) #define ZIO_GANG_CHILD_FLAGS(zio) \ (((zio)->io_flags & ZIO_FLAG_GANG_INHERIT) | \ ZIO_FLAG_GANG_CHILD | ZIO_FLAG_CANFAIL) #define ZIO_VDEV_CHILD_FLAGS(zio) \ (((zio)->io_flags & ZIO_FLAG_VDEV_INHERIT) | \ ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_CANFAIL) #define ZIO_CHILD_BIT(x) (1U << (x)) #define ZIO_CHILD_BIT_IS_SET(val, x) ((val) & (1U << (x))) enum zio_child { ZIO_CHILD_VDEV = 0, ZIO_CHILD_GANG, ZIO_CHILD_DDT, ZIO_CHILD_LOGICAL, ZIO_CHILD_TYPES }; #define ZIO_CHILD_VDEV_BIT ZIO_CHILD_BIT(ZIO_CHILD_VDEV) #define ZIO_CHILD_GANG_BIT ZIO_CHILD_BIT(ZIO_CHILD_GANG) #define ZIO_CHILD_DDT_BIT ZIO_CHILD_BIT(ZIO_CHILD_DDT) #define ZIO_CHILD_LOGICAL_BIT ZIO_CHILD_BIT(ZIO_CHILD_LOGICAL) #define ZIO_CHILD_ALL_BITS \ (ZIO_CHILD_VDEV_BIT | ZIO_CHILD_GANG_BIT | \ ZIO_CHILD_DDT_BIT | ZIO_CHILD_LOGICAL_BIT) enum zio_wait_type { ZIO_WAIT_READY = 0, ZIO_WAIT_DONE, ZIO_WAIT_TYPES }; typedef void zio_done_func_t(zio_t *zio); extern int zio_exclude_metadata; extern int zio_dva_throttle_enabled; extern const char *const zio_type_name[ZIO_TYPES]; /* * A bookmark is a four-tuple that uniquely * identifies any block in the pool. By convention, the meta-objset (MOS) * is objset 0, and the meta-dnode is object 0. This covers all blocks * except root blocks and ZIL blocks, which are defined as follows: * * Root blocks (objset_phys_t) are object 0, level -1: . * ZIL blocks are bookmarked . * dmu_sync()ed ZIL data blocks are bookmarked . * dnode visit bookmarks are . * * Note: this structure is called a bookmark because its original purpose * was to remember where to resume a pool-wide traverse. * * Note: this structure is passed between userland and the kernel, and is * stored on disk (by virtue of being incorporated into other on-disk * structures, e.g. dsl_scan_phys_t). * * If the head_errlog feature is enabled a different on-disk format for error * logs is used. This introduces the use of an error bookmark, a four-tuple * that uniquely identifies any error block * in the pool. The birth transaction group is used to track whether the block * has been overwritten by newer data or added to a snapshot since its marking * as an error. */ struct zbookmark_phys { uint64_t zb_objset; uint64_t zb_object; int64_t zb_level; uint64_t zb_blkid; }; struct zbookmark_err_phys { uint64_t zb_object; int64_t zb_level; uint64_t zb_blkid; uint64_t zb_birth; }; #define SET_BOOKMARK(zb, objset, object, level, blkid) \ { \ (zb)->zb_objset = objset; \ (zb)->zb_object = object; \ (zb)->zb_level = level; \ (zb)->zb_blkid = blkid; \ } #define ZB_DESTROYED_OBJSET (-1ULL) #define ZB_ROOT_OBJECT (0ULL) #define ZB_ROOT_LEVEL (-1LL) #define ZB_ROOT_BLKID (0ULL) #define ZB_ZIL_OBJECT (0ULL) #define ZB_ZIL_LEVEL (-2LL) #define ZB_DNODE_LEVEL (-3LL) #define ZB_DNODE_BLKID (0ULL) #define ZB_IS_ZERO(zb) \ ((zb)->zb_objset == 0 && (zb)->zb_object == 0 && \ (zb)->zb_level == 0 && (zb)->zb_blkid == 0) #define ZB_IS_ROOT(zb) \ ((zb)->zb_object == ZB_ROOT_OBJECT && \ (zb)->zb_level == ZB_ROOT_LEVEL && \ (zb)->zb_blkid == ZB_ROOT_BLKID) typedef struct zio_prop { enum zio_checksum zp_checksum; enum zio_compress zp_compress; uint8_t zp_complevel; uint8_t zp_level; uint8_t zp_copies; uint8_t zp_gang_copies; dmu_object_type_t zp_type; boolean_t zp_dedup; boolean_t zp_dedup_verify; boolean_t zp_nopwrite; boolean_t zp_brtwrite; boolean_t zp_encrypt; boolean_t zp_byteorder; boolean_t zp_direct_write; + boolean_t zp_rewrite; uint8_t zp_salt[ZIO_DATA_SALT_LEN]; uint8_t zp_iv[ZIO_DATA_IV_LEN]; uint8_t zp_mac[ZIO_DATA_MAC_LEN]; uint32_t zp_zpl_smallblk; dmu_object_type_t zp_storage_type; } zio_prop_t; typedef struct zio_cksum_report zio_cksum_report_t; typedef void zio_cksum_finish_f(zio_cksum_report_t *rep, const abd_t *good_data); typedef void zio_cksum_free_f(void *cbdata, size_t size); struct zio_bad_cksum; /* defined in zio_checksum.h */ struct dnode_phys; struct abd; struct zio_cksum_report { struct zio_cksum_report *zcr_next; nvlist_t *zcr_ereport; nvlist_t *zcr_detector; void *zcr_cbdata; size_t zcr_cbinfo; /* passed to zcr_free() */ uint64_t zcr_sector; uint64_t zcr_align; uint64_t zcr_length; zio_cksum_finish_f *zcr_finish; zio_cksum_free_f *zcr_free; /* internal use only */ struct zio_bad_cksum *zcr_ckinfo; /* information from failure */ }; typedef struct zio_vsd_ops { zio_done_func_t *vsd_free; } zio_vsd_ops_t; typedef struct zio_gang_node { zio_gbh_phys_t *gn_gbh; uint64_t gn_gangblocksize; uint64_t gn_allocsize; struct zio_gang_node *gn_child[]; } zio_gang_node_t; typedef zio_t *zio_gang_issue_func_t(zio_t *zio, blkptr_t *bp, zio_gang_node_t *gn, struct abd *data, uint64_t offset); typedef void zio_transform_func_t(zio_t *zio, struct abd *data, uint64_t size); typedef struct zio_transform { struct abd *zt_orig_abd; uint64_t zt_orig_size; uint64_t zt_bufsize; zio_transform_func_t *zt_transform; struct zio_transform *zt_next; } zio_transform_t; typedef zio_t *zio_pipe_stage_t(zio_t *zio); /* * The io_post flags describe additional actions that a parent IO should * consider or perform on behalf of a child. They are distinct from io_flags * because the child must be able to propagate them to the parent. The normal * io_flags are local to the zio, not protected by any lock, and not modifiable * by children; the reexecute flags are protected by io_lock, modifiable by * children, and always propagated -- even when ZIO_FLAG_DONT_PROPAGATE is set. */ #define ZIO_POST_REEXECUTE (1 << 0) #define ZIO_POST_SUSPEND (1 << 1) #define ZIO_POST_DIO_CHKSUM_ERR (1 << 2) /* * The io_trim flags are used to specify the type of TRIM to perform. They * only apply to ZIO_TYPE_TRIM zios are distinct from io_flags. */ enum trim_flag { ZIO_TRIM_SECURE = 1U << 0, }; typedef struct zio_alloc_list { list_t zal_list; uint64_t zal_size; } zio_alloc_list_t; typedef struct zio_link { zio_t *zl_parent; zio_t *zl_child; list_node_t zl_parent_node; list_node_t zl_child_node; } zio_link_t; enum zio_qstate { ZIO_QS_NONE = 0, ZIO_QS_QUEUED, ZIO_QS_ACTIVE, }; struct zio { /* Core information about this I/O */ zbookmark_phys_t io_bookmark; zio_prop_t io_prop; zio_type_t io_type; enum zio_child io_child_type; enum trim_flag io_trim_flags; zio_priority_t io_priority; uint8_t io_post; uint8_t io_state[ZIO_WAIT_TYPES]; uint64_t io_txg; spa_t *io_spa; blkptr_t *io_bp; blkptr_t *io_bp_override; blkptr_t io_bp_copy; list_t io_parent_list; list_t io_child_list; zio_t *io_logical; zio_transform_t *io_transform_stack; /* Callback info */ zio_done_func_t *io_ready; zio_done_func_t *io_children_ready; zio_done_func_t *io_done; void *io_private; int64_t io_prev_space_delta; /* DMU private */ blkptr_t io_bp_orig; /* io_lsize != io_orig_size iff this is a raw write */ uint64_t io_lsize; /* Data represented by this I/O */ struct abd *io_abd; struct abd *io_orig_abd; uint64_t io_size; uint64_t io_orig_size; /* Stuff for the vdev stack */ vdev_t *io_vd; void *io_vsd; const zio_vsd_ops_t *io_vsd_ops; metaslab_class_t *io_metaslab_class; /* dva throttle class */ enum zio_qstate io_queue_state; /* vdev queue state */ union { list_node_t l; avl_node_t a; } io_queue_node ____cacheline_aligned; /* allocator and vdev queues */ avl_node_t io_offset_node; /* vdev offset queues */ uint64_t io_offset; hrtime_t io_timestamp; /* submitted at */ hrtime_t io_queued_timestamp; hrtime_t io_target_timestamp; hrtime_t io_delta; /* vdev queue service delta */ hrtime_t io_delay; /* Device access time (disk or */ /* file). */ zio_alloc_list_t io_alloc_list; /* Internal pipeline state */ zio_flag_t io_flags; enum zio_stage io_stage; enum zio_stage io_pipeline; zio_flag_t io_orig_flags; enum zio_stage io_orig_stage; enum zio_stage io_orig_pipeline; enum zio_stage io_pipeline_trace; int io_error; int io_child_error[ZIO_CHILD_TYPES]; uint64_t io_children[ZIO_CHILD_TYPES][ZIO_WAIT_TYPES]; uint64_t *io_stall; zio_t *io_gang_leader; zio_gang_node_t *io_gang_tree; void *io_executor; void *io_waiter; void *io_bio; kmutex_t io_lock; kcondvar_t io_cv; int io_allocator; /* FMA state */ zio_cksum_report_t *io_cksum_report; uint64_t io_ena; /* Taskq dispatching state */ taskq_ent_t io_tqent; }; enum blk_verify_flag { BLK_VERIFY_ONLY, BLK_VERIFY_LOG, BLK_VERIFY_HALT }; enum blk_config_flag { BLK_CONFIG_HELD, // SCL_VDEV held for writer BLK_CONFIG_NEEDED, // SCL_VDEV should be obtained for reader BLK_CONFIG_NEEDED_TRY, // Try with SCL_VDEV for reader BLK_CONFIG_SKIP, // skip checks which require SCL_VDEV }; extern int zio_bookmark_compare(const void *, const void *); extern zio_t *zio_null(zio_t *pio, spa_t *spa, vdev_t *vd, zio_done_func_t *done, void *priv, zio_flag_t flags); extern zio_t *zio_root(spa_t *spa, zio_done_func_t *done, void *priv, zio_flag_t flags); extern void zio_destroy(zio_t *zio); extern zio_t *zio_read(zio_t *pio, spa_t *spa, const blkptr_t *bp, struct abd *data, uint64_t lsize, zio_done_func_t *done, void *priv, zio_priority_t priority, zio_flag_t flags, const zbookmark_phys_t *zb); extern zio_t *zio_write(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, struct abd *data, uint64_t size, uint64_t psize, const zio_prop_t *zp, zio_done_func_t *ready, zio_done_func_t *children_ready, zio_done_func_t *done, void *priv, zio_priority_t priority, zio_flag_t flags, const zbookmark_phys_t *zb); extern zio_t *zio_rewrite(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, struct abd *data, uint64_t size, zio_done_func_t *done, void *priv, zio_priority_t priority, zio_flag_t flags, zbookmark_phys_t *zb); extern void zio_write_override(zio_t *zio, blkptr_t *bp, int copies, int gang_copies, boolean_t nopwrite, boolean_t brtwrite); extern void zio_free(spa_t *spa, uint64_t txg, const blkptr_t *bp); extern zio_t *zio_claim(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, zio_done_func_t *done, void *priv, zio_flag_t flags); extern zio_t *zio_trim(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, zio_done_func_t *done, void *priv, zio_priority_t priority, zio_flag_t flags, enum trim_flag trim_flags); extern zio_t *zio_read_phys(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, struct abd *data, int checksum, zio_done_func_t *done, void *priv, zio_priority_t priority, zio_flag_t flags, boolean_t labels); extern zio_t *zio_write_phys(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, struct abd *data, int checksum, zio_done_func_t *done, void *priv, zio_priority_t priority, zio_flag_t flags, boolean_t labels); extern zio_t *zio_free_sync(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, zio_flag_t flags); extern int zio_alloc_zil(spa_t *spa, objset_t *os, uint64_t txg, blkptr_t *new_bp, uint64_t size, boolean_t *slog); extern void zio_flush(zio_t *zio, vdev_t *vd); extern void zio_shrink(zio_t *zio, uint64_t size); extern size_t zio_get_compression_max_size(enum zio_compress compress, uint64_t gcd_alloc, uint64_t min_alloc, size_t s_len); extern int zio_wait(zio_t *zio); extern void zio_nowait(zio_t *zio); extern void zio_execute(void *zio); extern void zio_interrupt(void *zio); extern void zio_delay_init(zio_t *zio); extern void zio_delay_interrupt(zio_t *zio); extern void zio_deadman(zio_t *zio, const char *tag); extern zio_t *zio_walk_parents(zio_t *cio, zio_link_t **); extern zio_t *zio_walk_children(zio_t *pio, zio_link_t **); extern zio_t *zio_unique_parent(zio_t *cio); extern void zio_add_child(zio_t *pio, zio_t *cio); extern void *zio_buf_alloc(size_t size); extern void zio_buf_free(void *buf, size_t size); extern void *zio_data_buf_alloc(size_t size); extern void zio_data_buf_free(void *buf, size_t size); extern void zio_push_transform(zio_t *zio, struct abd *abd, uint64_t size, uint64_t bufsize, zio_transform_func_t *transform); extern void zio_pop_transforms(zio_t *zio); extern void zio_resubmit_stage_async(void *); extern zio_t *zio_vdev_child_io(zio_t *zio, blkptr_t *bp, vdev_t *vd, uint64_t offset, struct abd *data, uint64_t size, int type, zio_priority_t priority, zio_flag_t flags, zio_done_func_t *done, void *priv); extern zio_t *zio_vdev_delegated_io(vdev_t *vd, uint64_t offset, struct abd *data, uint64_t size, zio_type_t type, zio_priority_t priority, zio_flag_t flags, zio_done_func_t *done, void *priv); extern void zio_vdev_io_bypass(zio_t *zio); extern void zio_vdev_io_reissue(zio_t *zio); extern void zio_vdev_io_redone(zio_t *zio); extern void zio_change_priority(zio_t *pio, zio_priority_t priority); extern void zio_checksum_verified(zio_t *zio); extern void zio_dio_chksum_verify_error_report(zio_t *zio); extern int zio_worst_error(int e1, int e2); extern enum zio_checksum zio_checksum_select(enum zio_checksum child, enum zio_checksum parent); extern enum zio_checksum zio_checksum_dedup_select(spa_t *spa, enum zio_checksum child, enum zio_checksum parent); extern enum zio_compress zio_compress_select(spa_t *spa, enum zio_compress child, enum zio_compress parent); extern uint8_t zio_complevel_select(spa_t *spa, enum zio_compress compress, uint8_t child, uint8_t parent); extern void zio_suspend(spa_t *spa, zio_t *zio, zio_suspend_reason_t); extern int zio_resume(spa_t *spa); extern void zio_resume_wait(spa_t *spa); extern int zfs_blkptr_verify(spa_t *spa, const blkptr_t *bp, enum blk_config_flag blk_config, enum blk_verify_flag blk_verify); /* * Initial setup and teardown. */ extern void zio_init(void); extern void zio_fini(void); /* * Fault injection */ struct zinject_record; extern uint32_t zio_injection_enabled; extern int zio_inject_fault(char *name, int flags, int *id, struct zinject_record *record); extern int zio_inject_list_next(int *id, char *name, size_t buflen, struct zinject_record *record); extern int zio_clear_fault(int id); extern void zio_handle_panic_injection(spa_t *spa, const char *tag, uint64_t type); extern int zio_handle_decrypt_injection(spa_t *spa, const zbookmark_phys_t *zb, uint64_t type, int error); extern int zio_handle_fault_injection(zio_t *zio, int error); extern int zio_handle_device_injection(vdev_t *vd, zio_t *zio, int error); extern int zio_handle_device_injections(vdev_t *vd, zio_t *zio, int err1, int err2); extern int zio_handle_label_injection(zio_t *zio, int error); extern void zio_handle_ignored_writes(zio_t *zio); extern hrtime_t zio_handle_io_delay(zio_t *zio); extern void zio_handle_import_delay(spa_t *spa, hrtime_t elapsed); extern void zio_handle_export_delay(spa_t *spa, hrtime_t elapsed); /* * Checksum ereport functions */ extern int zfs_ereport_start_checksum(spa_t *spa, vdev_t *vd, const zbookmark_phys_t *zb, struct zio *zio, uint64_t offset, uint64_t length, struct zio_bad_cksum *info); extern void zfs_ereport_finish_checksum(zio_cksum_report_t *report, const abd_t *good_data, const abd_t *bad_data, boolean_t drop_if_identical); extern void zfs_ereport_free_checksum(zio_cksum_report_t *report); /* If we have the good data in hand, this function can be used */ extern int zfs_ereport_post_checksum(spa_t *spa, vdev_t *vd, const zbookmark_phys_t *zb, struct zio *zio, uint64_t offset, uint64_t length, const abd_t *good_data, const abd_t *bad_data, struct zio_bad_cksum *info); void zio_vsd_default_cksum_report(zio_t *zio, zio_cksum_report_t *zcr); extern void zfs_ereport_snapshot_post(const char *subclass, spa_t *spa, const char *name); /* Called from spa_sync(), but primarily an injection handler */ extern void spa_handle_ignored_writes(spa_t *spa); /* zbookmark_phys functions */ boolean_t zbookmark_subtree_completed(const struct dnode_phys *dnp, const zbookmark_phys_t *subtree_root, const zbookmark_phys_t *last_block); boolean_t zbookmark_subtree_tbd(const struct dnode_phys *dnp, const zbookmark_phys_t *subtree_root, const zbookmark_phys_t *last_block); 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); #ifdef __cplusplus } #endif #endif /* _ZIO_H */ diff --git a/include/zfeature_common.h b/include/zfeature_common.h index 4877df4b114d..56382ca85b55 100644 --- a/include/zfeature_common.h +++ b/include/zfeature_common.h @@ -1,157 +1,158 @@ // SPDX-License-Identifier: CDDL-1.0 /* * 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 https://opensource.org/licenses/CDDL-1.0. * 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) 2011, 2018 by Delphix. All rights reserved. * Copyright (c) 2013 by Saso Kiselkov. All rights reserved. * Copyright (c) 2013, Joyent, Inc. All rights reserved. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2024, Klara, Inc. */ #ifndef _ZFEATURE_COMMON_H #define _ZFEATURE_COMMON_H extern __attribute__((visibility("default"))) #include #include #include #ifdef __cplusplus extern "C" { #endif struct zfeature_info; typedef enum spa_feature { SPA_FEATURE_NONE = -1, SPA_FEATURE_ASYNC_DESTROY, SPA_FEATURE_EMPTY_BPOBJ, SPA_FEATURE_LZ4_COMPRESS, SPA_FEATURE_MULTI_VDEV_CRASH_DUMP, SPA_FEATURE_SPACEMAP_HISTOGRAM, SPA_FEATURE_ENABLED_TXG, SPA_FEATURE_HOLE_BIRTH, SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_EMBEDDED_DATA, SPA_FEATURE_BOOKMARKS, SPA_FEATURE_FS_SS_LIMIT, SPA_FEATURE_LARGE_BLOCKS, SPA_FEATURE_LARGE_DNODE, SPA_FEATURE_SHA512, SPA_FEATURE_SKEIN, SPA_FEATURE_EDONR, SPA_FEATURE_USEROBJ_ACCOUNTING, SPA_FEATURE_ENCRYPTION, SPA_FEATURE_PROJECT_QUOTA, SPA_FEATURE_DEVICE_REMOVAL, SPA_FEATURE_OBSOLETE_COUNTS, SPA_FEATURE_POOL_CHECKPOINT, SPA_FEATURE_SPACEMAP_V2, SPA_FEATURE_ALLOCATION_CLASSES, SPA_FEATURE_RESILVER_DEFER, SPA_FEATURE_BOOKMARK_V2, SPA_FEATURE_REDACTION_BOOKMARKS, SPA_FEATURE_REDACTED_DATASETS, SPA_FEATURE_BOOKMARK_WRITTEN, SPA_FEATURE_LOG_SPACEMAP, SPA_FEATURE_LIVELIST, SPA_FEATURE_DEVICE_REBUILD, SPA_FEATURE_ZSTD_COMPRESS, SPA_FEATURE_DRAID, SPA_FEATURE_ZILSAXATTR, SPA_FEATURE_HEAD_ERRLOG, SPA_FEATURE_BLAKE3, SPA_FEATURE_BLOCK_CLONING, SPA_FEATURE_AVZ_V2, SPA_FEATURE_REDACTION_LIST_SPILL, SPA_FEATURE_RAIDZ_EXPANSION, SPA_FEATURE_FAST_DEDUP, SPA_FEATURE_LONGNAME, SPA_FEATURE_LARGE_MICROZAP, SPA_FEATURE_DYNAMIC_GANG_HEADER, SPA_FEATURE_BLOCK_CLONING_ENDIAN, + SPA_FEATURE_PHYSICAL_REWRITE, SPA_FEATURES } spa_feature_t; #define SPA_FEATURE_DISABLED (-1ULL) typedef enum zfeature_flags { /* Can open pool readonly even if this feature is not supported. */ ZFEATURE_FLAG_READONLY_COMPAT = (1 << 0), /* * Is this feature necessary to load the pool? i.e. do we need this * feature to read the full feature list out of the MOS? */ ZFEATURE_FLAG_MOS = (1 << 1), /* Activate this feature at the same time it is enabled. */ ZFEATURE_FLAG_ACTIVATE_ON_ENABLE = (1 << 2), /* Each dataset has a field set if it has ever used this feature. */ ZFEATURE_FLAG_PER_DATASET = (1 << 3), /* * This feature isn't enabled by zpool upgrade; it must be explicitly * listed to be enabled. It will also be applied if listed in an * explicitly provided compatibility list. This flag can be removed * from a given feature once support is sufficiently widespread, or * worries about backwards compatibility are no longer relevant. */ ZFEATURE_FLAG_NO_UPGRADE = (1 << 4) } zfeature_flags_t; typedef enum zfeature_type { ZFEATURE_TYPE_BOOLEAN, ZFEATURE_TYPE_UINT64_ARRAY, ZFEATURE_NUM_TYPES } zfeature_type_t; typedef struct zfeature_info { spa_feature_t fi_feature; const char *fi_uname; /* User-facing feature name */ const char *fi_guid; /* On-disk feature identifier */ const char *fi_desc; /* Feature description */ zfeature_flags_t fi_flags; boolean_t fi_zfs_mod_supported; /* supported by running zfs module */ zfeature_type_t fi_type; /* Only relevant for PER_DATASET features */ /* array of dependencies, terminated by SPA_FEATURE_NONE */ const spa_feature_t *fi_depends; } zfeature_info_t; typedef int (zfeature_func_t)(zfeature_info_t *, void *); #define ZFS_FEATURE_DEBUG _ZFEATURE_COMMON_H zfeature_info_t spa_feature_table[SPA_FEATURES]; _ZFEATURE_COMMON_H boolean_t zfeature_checks_disable; _ZFEATURE_COMMON_H boolean_t zfeature_is_valid_guid(const char *); _ZFEATURE_COMMON_H boolean_t zfeature_is_supported(const char *); _ZFEATURE_COMMON_H int zfeature_lookup_guid(const char *, spa_feature_t *); _ZFEATURE_COMMON_H int zfeature_lookup_name(const char *, spa_feature_t *); _ZFEATURE_COMMON_H boolean_t zfeature_depends_on(spa_feature_t, spa_feature_t); _ZFEATURE_COMMON_H void zpool_feature_init(void); #ifdef __cplusplus } #endif #endif /* _ZFEATURE_COMMON_H */ diff --git a/lib/libzfs/libzfs.abi b/lib/libzfs/libzfs.abi index 37d22402e777..ba161d1ef10f 100644 --- a/lib/libzfs/libzfs.abi +++ b/lib/libzfs/libzfs.abi @@ -1,10374 +1,10375 @@ - + - + + - - + + - + diff --git a/man/man7/zpool-features.7 b/man/man7/zpool-features.7 index 66aa100b7149..10dfd1f92936 100644 --- a/man/man7/zpool-features.7 +++ b/man/man7/zpool-features.7 @@ -1,1085 +1,1102 @@ .\" SPDX-License-Identifier: CDDL-1.0 .\" .\" 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 https://opensource.org/licenses/CDDL-1.0. .\" .\" 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, 2023, 2024, Klara, Inc. .\" Copyright (c) 2019, Allan Jude .\" Copyright (c) 2021, Colm Buckley .\" .Dd October 2, 2024 .Dt ZPOOL-FEATURES 7 .Os . .Sh NAME .Nm zpool-features .Nd description of ZFS pool features . .Sh DESCRIPTION ZFS pool on-disk format versions are specified via .Dq features which replace the old on-disk format numbers .Pq the last supported on-disk format number is 28 . To enable a feature on a pool use the .Nm zpool Cm upgrade , or set the .Sy feature Ns @ Ns Ar feature-name property to .Sy enabled . Please also see the .Sx Compatibility feature sets section for information on how sets of features may be enabled together. .Pp The pool format does not affect file system version compatibility or the ability to send file systems between pools. .Pp 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 .Sy active on the pool. If the pool was created by another software version this set may include unsupported features. . .Ss Identifying features Every feature has a GUID of the form .Ar com.example : Ns Ar feature-name . 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. .Pp Each supported feature also has a short name. By convention a feature's short name is the portion of its GUID which follows the .Sq \&: .Po i.e. .Ar com.example : Ns Ar feature-name would have the short name .Ar feature-name .Pc , however a feature's short name may differ across ZFS implementations if following the convention would result in name conflicts. . .Ss Feature states Features can be in one of three states: .Bl -tag -width "disabled" .It Sy active 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 .Pq see Sx Read-only compatibility . .It Sy enabled 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 .Sy active state. Some features may support returning to the .Sy enabled state after becoming .Sy active . See feature-specific documentation for details. .It Sy disabled This feature's on-disk format changes have not been made and will not be made unless an administrator moves the feature to the .Sy enabled state. Features cannot be disabled once they have been enabled. .El .Pp The state of supported features is exposed through pool properties of the form .Sy feature Ns @ Ns Ar short-name . . .Ss Read-only compatibility 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 .Dq 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 .Sy readonly property during import .Po see .Xr zpool-import 8 for details on importing pools .Pc . . .Ss Unsupported features For each unsupported feature enabled on an imported pool, a pool property named .Sy unsupported Ns @ Ns Ar feature-name will indicate why the import was allowed despite the unsupported feature. Possible values for this property are: .Bl -tag -width "readonly" .It Sy inactive The feature is in the .Sy enabled state and therefore the pool's on-disk format is still compatible with software that does not support this feature. .It Sy readonly The feature is read-only compatible and the pool has been imported in read-only mode. .El . .Ss Feature dependencies Some features depend on other features being enabled in order to function. Enabling a feature will automatically enable any features it depends on. . .Ss Compatibility feature sets It is sometimes necessary for a pool to maintain compatibility with a specific on-disk format, by enabling and disabling particular features. The .Sy compatibility feature facilitates this by allowing feature sets to be read from text files. When set to .Sy off .Pq the default , compatibility feature sets are disabled .Pq i.e. all features are enabled ; when set to .Sy legacy , no features are enabled. When set to a comma-separated list of filenames .Po each filename may either be an absolute path, or relative to .Pa /etc/zfs/compatibility.d or .Pa /usr/share/zfs/compatibility.d .Pc , the lists of requested features are read from those files, separated by whitespace and/or commas. Only features present in all files are enabled. .Pp Simple sanity checks are applied to the files: they must be between 1 B and 16 KiB in size, and must end with a newline character. .Pp The requested features are applied when a pool is created using .Nm zpool Cm create Fl o Sy compatibility Ns = Ns Ar … and controls which features are enabled when using .Nm zpool Cm upgrade . .Nm zpool Cm status will not show a warning about disabled features which are not part of the requested feature set. .Pp The special value .Sy legacy prevents any features from being enabled, either via .Nm zpool Cm upgrade or .Nm zpool Cm set Sy feature Ns @ Ns Ar feature-name Ns = Ns Sy enabled . This setting also prevents pools from being upgraded to newer on-disk versions. This is a safety measure to prevent new features from being accidentally enabled, breaking compatibility. .Pp By convention, compatibility files in .Pa /usr/share/zfs/compatibility.d are provided by the distribution, and include feature sets supported by important versions of popular distributions, and feature sets commonly supported at the start of each year. Compatibility files in .Pa /etc/zfs/compatibility.d , if present, will take precedence over files with the same name in .Pa /usr/share/zfs/compatibility.d . .Pp If an unrecognized feature is found in these files, an error message will be shown. If the unrecognized feature is in a file in .Pa /etc/zfs/compatibility.d , this is treated as an error and processing will stop. If the unrecognized feature is under .Pa /usr/share/zfs/compatibility.d , this is treated as a warning and processing will continue. This difference is to allow distributions to include features which might not be recognized by the currently-installed binaries. .Pp Compatibility files may include comments: any text from .Sq # to the end of the line is ignored. .Pp .Sy Example : .Bd -literal -compact -offset 4n .No example# Nm cat Pa /usr/share/zfs/compatibility.d/grub2 # Features which are supported by GRUB2 versions from v2.12 onwards. allocation_classes async_destroy block_cloning bookmarks device_rebuild embedded_data empty_bpobj enabled_txg extensible_dataset filesystem_limits hole_birth large_blocks livelist log_spacemap lz4_compress project_quota resilver_defer spacemap_histogram spacemap_v2 userobj_accounting zilsaxattr zpool_checkpoint .No example# Nm cat Pa /usr/share/zfs/compatibility.d/grub2-2.06 # Features which are supported by GRUB2 versions prior to v2.12. # # GRUB is not able to detect ZFS pool if snapshot of top level boot pool # is created. This issue is observed with GRUB versions before v2.12 if # extensible_dataset feature is enabled on ZFS boot pool. # # This file lists all read-only compatible features except # extensible_dataset and any other feature that depends on it. # allocation_classes async_destroy block_cloning device_rebuild embedded_data empty_bpobj enabled_txg hole_birth log_spacemap lz4_compress resilver_defer spacemap_histogram spacemap_v2 zpool_checkpoint .No example# Nm zpool Cm create Fl o Sy compatibility Ns = Ns Ar grub2 Ar bootpool Ar vdev .Ed .Pp See .Xr zpool-create 8 and .Xr zpool-upgrade 8 for more information on how these commands are affected by feature sets. . .de feature .It Sy \\$2 .Bl -tag -compact -width "READ-ONLY COMPATIBLE" .It GUID .Sy \\$1:\\$2 .if !"\\$4"" \{\ .It DEPENDENCIES \fB\\$4\fP\c .if !"\\$5"" , \fB\\$5\fP\c .if !"\\$6"" , \fB\\$6\fP\c .if !"\\$7"" , \fB\\$7\fP\c .if !"\\$8"" , \fB\\$8\fP\c .if !"\\$9"" , \fB\\$9\fP\c .\} .It READ-ONLY COMPATIBLE \\$3 .El .Pp .. . .ds instant-never \ .No This feature becomes Sy active No as soon as it is enabled \ and will never return to being Sy enabled . . .ds remount-upgrade \ .No Each filesystem will be upgraded automatically when remounted, \ or when a new file is created under that filesystem. \ The upgrade can also be triggered on filesystems via \ Nm zfs Cm set Sy version Ns = Ns Sy current Ar fs . \ No The upgrade process runs in the background and may take a while to complete \ for filesystems containing large amounts of files . . .de checksum-spiel When the .Sy \\$1 feature is set to .Sy enabled , the administrator can turn on the .Sy \\$1 checksum on any dataset using .Nm zfs Cm set Sy checksum Ns = Ns Sy \\$1 Ar dset .Po see Xr zfs-set 8 Pc . This feature becomes .Sy active once a .Sy checksum property has been set to .Sy \\$1 , and will return to being .Sy enabled once all filesystems that have ever had their checksum set to .Sy \\$1 are destroyed. .. . .Sh FEATURES The following features are supported on this system: .Bl -tag -width Ds .feature org.zfsonlinux allocation_classes yes This feature enables support for separate allocation classes. .Pp This feature becomes .Sy active when a dedicated allocation class vdev .Pq dedup or special is created with the .Nm zpool Cm create No or Nm zpool Cm add No commands . With device removal, it can be returned to the .Sy enabled state if all the dedicated allocation class vdevs are removed. . .feature com.delphix async_destroy yes Destroying a file system requires traversing all of its data in order to return its used space to the pool. Without .Sy async_destroy , 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. .Pp When .Sy async_destroy 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 .Sy freeing property. .Pp This feature is only .Sy active while .Sy freeing is non-zero. . .feature org.openzfs blake3 no extensible_dataset This feature enables the use of the BLAKE3 hash algorithm for checksum and dedup. BLAKE3 is a secure hash algorithm focused on high performance. .Pp .checksum-spiel blake3 . .feature com.fudosecurity block_cloning yes When this feature is enabled ZFS will use block cloning for operations like .Fn copy_file_range 2 . Block cloning allows to create multiple references to a single block. It is much faster than copying the data (as the actual data is neither read nor written) and takes no additional space. Blocks can be cloned across datasets under some conditions (like equal .Nm recordsize , the same master encryption key, etc.). ZFS tries its best to clone across datasets including encrypted ones. This is limited for various (nontrivial) reasons depending on the OS and/or ZFS internals. .Pp This feature becomes .Sy active when first block is cloned. When the last cloned block is freed, it goes back to the enabled state. .feature com.truenas block_cloning_endian yes This feature corrects ZAP entry endianness issues in the Block Reference Table (BRT) used by block cloning. During the original block cloning implementation, BRT ZAP entries were mistakenly stored as arrays of 8 single-byte entries instead of single 8-byte entries, making pools non-endian-safe. .Pp This feature is activated when the first BRT ZAP is created (that way ensuring compatibility with existing pools). When active, new BRT entries are stored in the correct endian-safe format. The feature becomes inactive when all BRT ZAPs are destroyed. .feature com.delphix bookmarks yes extensible_dataset This feature enables use of the .Nm zfs Cm bookmark command. .Pp This feature is .Sy active while any bookmarks exist in the pool. All bookmarks in the pool can be listed by running .Nm zfs Cm list Fl t Sy bookmark Fl r Ar poolname . . .feature com.datto bookmark_v2 no bookmark extensible_dataset This feature enables the creation and management of larger bookmarks which are needed for other features in ZFS. .Pp This feature becomes .Sy active when a v2 bookmark is created and will be returned to the .Sy enabled state when all v2 bookmarks are destroyed. . .feature com.delphix bookmark_written no bookmark extensible_dataset bookmark_v2 This feature enables additional bookmark accounting fields, enabling the .Sy written Ns # Ns Ar bookmark property .Pq space written since a bookmark and estimates of send stream sizes for incrementals from bookmarks. .Pp This feature becomes .Sy active when a bookmark is created and will be returned to the .Sy enabled state when all bookmarks with these fields are destroyed. . .feature org.openzfs device_rebuild yes This feature enables the ability for the .Nm zpool Cm attach and .Nm zpool Cm replace commands to perform sequential reconstruction .Pq instead of healing reconstruction when resilvering. .Pp 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 .Sy active while a sequential resilver is in progress, and returns to .Sy enabled when the resilver completes. . .feature com.delphix device_removal no This feature enables the .Nm zpool Cm remove command to remove top-level vdevs, evacuating them to reduce the total size of the pool. .Pp This feature becomes .Sy active when the .Nm zpool Cm remove command is used on a top-level vdev, and will never return to being .Sy enabled . . .feature org.openzfs draid no This feature enables use of the .Sy draid vdev type. dRAID is a variant of RAID-Z which provides integrated distributed hot spares that allow faster resilvering while retaining the benefits of RAID-Z. Data, parity, and spare space are organized in redundancy groups and distributed evenly over all of the devices. .Pp This feature becomes .Sy active when creating a pool which uses the .Sy draid vdev type, or when adding a new .Sy draid vdev to an existing pool. . .feature com.klarasystems dynamic_gang_header no This feature enables larger gang headers based on the sector size of the pool. When enabled, gang headers will use the entire space allocated for them, instead of always restricting themselves to 512 bytes. This can reduce the need for nested gang trees in extreme fragmentation scenarios. .Pp This feature becomes active when a gang header is written that is larger than 512 bytes. This feature is not enabled by .Xr zpool-upgrade 8 . Instead, it must be manually enabled, or be part of a compatibility file. . .feature org.illumos edonr no extensible_dataset This feature enables the use of the Edon-R hash algorithm for checksum, including for nopwrite .Po if compression is also enabled, an overwrite of a block whose checksum matches the data being written will be ignored .Pc . In an abundance of caution, Edon-R requires verification when used with dedup: .Nm zfs Cm set Sy dedup Ns = Ns Sy edonr , Ns Sy verify .Po see Xr zfs-set 8 Pc . .Pp Edon-R is a very high-performance hash algorithm that was part of the NIST SHA-3 competition. It provides extremely high hash performance .Pq 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 .Pq 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. .Pp .checksum-spiel edonr . .feature com.delphix embedded_data no 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. .Pp When this feature is enabled, the contents of highly-compressible blocks are stored in the block .Dq pointer itself .Po a misnomer in this case, as it contains the compressed data, rather than a pointer to its location on disk .Pc . Thus the space of the block .Pq one sector, typically 512 B or 4 KiB is saved, and no additional I/O is needed to read and write the data block. . \*[instant-never] . .feature com.delphix empty_bpobj yes 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. .Pp When there are many snapshots, each snapshot uses many Block Pointer Objects .Pq bpobjs to track blocks associated with that snapshot. However, in common use cases, most of these bpobjs are empty. This feature allows us to create each bpobj on-demand, thus eliminating the empty bpobjs. .Pp This feature is .Sy active while there are any filesystems, volumes, or snapshots which were created after enabling this feature. . .feature com.delphix enabled_txg yes 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. .Pp This feature becomes .Sy active as soon as it is enabled and will never return to being .Sy enabled . . .feature com.datto encryption no bookmark_v2 extensible_dataset This feature enables the creation and management of natively encrypted datasets. .Pp This feature becomes .Sy active when an encrypted dataset is created and will be returned to the .Sy enabled state when all datasets that use this feature are destroyed. . .feature com.klarasystems fast_dedup yes This feature allows more advanced deduplication features to be enabled on new dedup tables. .Pp This feature will be .Sy active when the first deduplicated block is written after a new dedup table is created (i.e. after a new pool creation, or new checksum used on a dataset with .Sy dedup enabled). It will be returned to the .Sy enabled state when all deduplicated blocks using it are freed. . .feature com.delphix extensible_dataset no This feature allows more flexible use of internal ZFS data structures, and exists for other features to depend on. .Pp This feature will be .Sy active when the first dependent feature uses it, and will be returned to the .Sy enabled state when all datasets that use this feature are destroyed. . .feature com.joyent filesystem_limits yes extensible_dataset 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. .Pp This feature is .Sy active once either of the limit properties has been set on a dataset and will never return to being .Sy enabled . . .feature com.delphix head_errlog no This feature enables the upgraded version of errlog, which required an on-disk error log format change. Now the error log of each head dataset is stored separately in the zap object and keyed by the head id. With this feature enabled, every dataset affected by an error block is listed in the output of .Nm zpool Cm status . In case of encrypted filesystems with unloaded keys we are unable to check their snapshots or clones for errors and these will not be reported. An "access denied" error will be reported. .Pp \*[instant-never] . .feature com.delphix hole_birth no enabled_txg This feature has/had bugs, the result of which is that, if you do a .Nm zfs Cm send Fl i .Pq or Fl R , No since it uses Fl i from an affected dataset, the receiving party will not see any checksum or other errors, but the resulting destination snapshot will not match the source. Its use by .Nm zfs Cm send Fl i has been disabled by default .Po see .Sy send_holes_without_birth_time in .Xr zfs 4 .Pc . .Pp This feature improves performance of incremental sends .Pq Nm zfs Cm send Fl i and receives for objects with many holes. The most common case of hole-filled objects is zvols. .Pp An incremental send stream from snapshot .Sy A No to snapshot Sy B contains information about every block that changed between .Sy A No and Sy B . Blocks which did not change between those snapshots can be identified and omitted from the stream using a piece of metadata called the .Dq block birth time , but birth times are not recorded for holes .Pq blocks filled only with zeroes . Since holes created after .Sy A No cannot be distinguished from holes created before Sy A , information about every hole in the entire filesystem or zvol is included in the send stream. .Pp For workloads where holes are rare this is not a problem. However, when incrementally replicating filesystems or zvols with many holes .Pq 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. .Pp Once the .Sy hole_birth 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. .Pp \*[instant-never] . .feature org.open-zfs large_blocks no extensible_dataset This feature allows the record size on a dataset to be set larger than 128 KiB. .Pp This feature becomes .Sy active once a dataset contains a file with a block size larger than 128 KiB, and will return to being .Sy enabled once all filesystems that have ever had their recordsize larger than 128 KiB are destroyed. . .feature org.zfsonlinux large_dnode no extensible_dataset This feature allows the size of dnodes in a dataset to be set larger than 512 B. . This feature becomes .Sy active once a dataset contains an object with a dnode larger than 512 B, which occurs as a result of setting the .Sy dnodesize dataset property to a value other than .Sy legacy . The feature will return to being .Sy enabled once all filesystems that have ever contained a dnode larger than 512 B 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. . .feature com.klarasystems large_microzap yes extensible_dataset large_blocks This feature allows "micro" ZAPs to grow larger than 128 KiB without being upgraded to "fat" ZAPs. .Pp This feature becomes .Sy active the first time a micro ZAP grows larger than 128KiB. It will only be returned to the .Sy enabled state when all datasets that ever had a large micro ZAP are destroyed. .Pp Note that even when this feature is enabled, micro ZAPs cannot grow larger than 128 KiB without also changing the .Sy zap_micro_max_size module parameter. See .Xr zfs 4 . . .feature com.delphix livelist yes extensible_dataset 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 .Sy active until all clones have been destroyed. . .feature com.delphix log_spacemap yes com.delphix:spacemap_v2 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. .Pp \*[instant-never] . .feature org.zfsonlinux longname no extensible_dataset This feature allows creating files and directories with name up to 1023 bytes in length. A new dataset property .Sy longname is also introduced to toggle longname support for each dataset individually. This property can be disabled even if it contains longname files. In such case, new file cannot be created with longname but existing longname files can still be looked up. .Pp This feature becomes .Sy active when a file name greater than 255 is created in a dataset, and returns to being .Sy enabled when all such datasets are destroyed. . .feature org.illumos lz4_compress no .Sy lz4 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 .Sy lzjb compression. Typically, .Sy lz4 compression is approximately 50% faster on compressible data and 200% faster on incompressible data than .Sy lzjb . It is also approximately 80% faster on decompression, while giving approximately a 10% better compression ratio. .Pp When the .Sy lz4_compress feature is set to .Sy enabled , the administrator can turn on .Sy lz4 compression on any dataset on the pool using the .Xr zfs-set 8 command. All newly written metadata will be compressed with the .Sy lz4 algorithm. .Pp \*[instant-never] . .feature com.joyent multi_vdev_crash_dump no 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. .Pp When the .Sy multi_vdev_crash_dump feature is set to .Sy enabled , the administrator can use .Xr dumpadm 8 to configure a dump device on a pool comprised of multiple vdevs. .Pp Under .Fx and Linux this feature is unused, but registered for compatibility. New pools created on these systems will have the feature .Sy enabled but will never transition to .Sy active , as this functionality is not required for crash dump support. Existing pools where this feature is .Sy active can be imported. . .feature com.delphix obsolete_counts yes device_removal This feature is an enhancement of .Sy 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 .Dq obsolete – no longer needed. .Pp This feature becomes .Sy active when the .Nm zpool Cm remove command is used on a top-level vdev, and will never return to being .Sy enabled . . +.feature com.truenas physical_rewrite yes extensible_dataset +This feature enables physical block rewriting that preserves logical birth +times, avoiding unnecessary inclusion of rewritten blocks in incremental +.Nm zfs Cm send +streams. +When enabled, the +.Nm zfs Cm rewrite Fl P +command can be used. +.Pp +This feature becomes +.Sy active +the first time +.Nm zfs Cm rewrite Fl P +is used on any dataset, and will return to being +.Sy enabled +once all datasets that have ever used physical rewrite are destroyed. +. .feature org.zfsonlinux project_quota yes extensible_dataset This feature allows administrators to account the spaces and objects usage information against the project identifier .Pq ID . .Pp The project ID is an object-based attribute. When upgrading an existing filesystem, objects without a project ID will be assigned a zero project ID. When this feature is enabled, newly created objects inherit their parent directories' project ID if the parent's inherit flag is set .Pq via Nm chattr Sy [+-]P No or Nm zfs Cm project Fl s Ns | Ns Fl C . Otherwise, the new object's project ID will be zero. An object's project ID can be changed at any time by the owner .Pq or privileged user via .Nm chattr Fl p Ar prjid or .Nm zfs Cm project Fl p Ar prjid . .Pp This feature will become .Sy active as soon as it is enabled and will never return to being .Sy disabled . \*[remount-upgrade] . .feature org.openzfs raidz_expansion no none This feature enables the .Nm zpool Cm attach subcommand to attach a new device to a RAID-Z group, expanding the total amount usable space in the pool. See .Xr zpool-attach 8 . . .feature com.delphix redaction_bookmarks no bookmarks extensible_dataset This feature enables the use of redacted .Nm zfs Cm send Ns s , which create redaction bookmarks storing the list of blocks redacted by the send that created them. For more information about redacted sends, see .Xr zfs-send 8 . . .feature com.delphix redacted_datasets no extensible_dataset This feature enables the receiving of redacted .Nm zfs Cm send streams, which 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 receives, see .Xr zfs-send 8 . . .feature com.delphix redaction_list_spill no redaction_bookmarks This feature enables the redaction list created by zfs redact to store many more entries. It becomes .Sy active when a redaction list is created with more than 36 entries, and returns to being .Sy enabled when no long redaction lists remain in the pool. For more information about redacted sends, see .Xr zfs-send 8 . . .feature com.datto resilver_defer yes 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. .Pp This feature becomes .Sy active once a resilver has been deferred, and returns to being .Sy enabled when the deferred resilver begins. . .feature org.illumos sha512 no extensible_dataset This feature enables the use of the SHA-512/256 truncated hash algorithm .Pq 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 .Sy skein No and Sy edonr algorithms. .Pp .checksum-spiel sha512 . .feature org.illumos skein no extensible_dataset 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 .Pq 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 .Pq 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. .Pp .checksum-spiel skein . .feature com.delphix spacemap_histogram yes This features allows ZFS to maintain more information about how free space is organized within the pool. If this feature is .Sy enabled , it will be activated when a new space map object is created, or an existing space map is upgraded to the new format, and never returns back to being .Sy enabled . . .feature com.delphix spacemap_v2 yes This feature enables the use of the new space map encoding which consists of two words .Pq 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. .Pp This feature becomes .Sy active once it is .Sy enabled , and never returns back to being .Sy enabled . . .feature org.zfsonlinux userobj_accounting yes extensible_dataset This feature allows administrators to account the object usage information by user and group. .Pp \*[instant-never] \*[remount-upgrade] . .feature com.klarasystems vdev_zaps_v2 no This feature creates a ZAP object for the root vdev. .Pp This feature becomes active after the next .Nm zpool Cm import or .Nm zpool reguid . . Properties can be retrieved or set on the root vdev using .Nm zpool Cm get and .Nm zpool Cm set with .Sy root as the vdev name which is an alias for .Sy root-0 . .feature org.openzfs zilsaxattr yes extensible_dataset This feature enables .Sy xattr Ns = Ns Sy sa extended attribute logging in the ZIL. If enabled, extended attribute changes .Pq both Sy xattrdir Ns = Ns Sy dir No and Sy xattr Ns = Ns Sy sa are guaranteed to be durable if either the dataset had .Sy sync Ns = Ns Sy always set at the time the changes were made, or .Xr sync 2 is called on the dataset after the changes were made. .Pp This feature becomes .Sy active when a ZIL is created for at least one dataset and will be returned to the .Sy enabled state when it is destroyed for all datasets that use this feature. . .feature com.delphix zpool_checkpoint yes This feature enables the .Nm zpool Cm checkpoint command that can checkpoint the state of the pool at the time it was issued and later rewind back to it or discard it. .Pp This feature becomes .Sy active when the .Nm zpool Cm checkpoint command is used to checkpoint the pool. The feature will only return back to being .Sy enabled when the pool is rewound or the checkpoint has been discarded. . .feature org.freebsd zstd_compress no extensible_dataset .Sy zstd is a high-performance compression algorithm that features a combination of high compression ratios and high speed. Compared to .Sy gzip , .Sy zstd offers slightly better compression at much higher speeds. Compared to .Sy lz4 , .Sy zstd offers much better compression while being only modestly slower. Typically, .Sy zstd compression speed ranges from 250 to 500 MB/s per thread and decompression speed is over 1 GB/s per thread. .Pp When the .Sy zstd feature is set to .Sy enabled , the administrator can turn on .Sy zstd compression of any dataset using .Nm zfs Cm set Sy compress Ns = Ns Sy zstd Ar dset .Po see Xr zfs-set 8 Pc . This feature becomes .Sy active once a .Sy compress property has been set to .Sy zstd , and will return to being .Sy enabled once all filesystems that have ever had their .Sy compress property set to .Sy zstd are destroyed. .El . .Sh SEE ALSO .Xr zfs 8 , .Xr zpool 8 diff --git a/man/man8/zfs-rewrite.8 b/man/man8/zfs-rewrite.8 index 423d6d439e28..a3a037f3794a 100644 --- a/man/man8/zfs-rewrite.8 +++ b/man/man8/zfs-rewrite.8 @@ -1,76 +1,90 @@ .\" SPDX-License-Identifier: CDDL-1.0 .\" .\" 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 https://opensource.org/licenses/CDDL-1.0. .\" 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) 2025 iXsystems, Inc. .\" .Dd May 6, 2025 .Dt ZFS-REWRITE 8 .Os . .Sh NAME .Nm zfs-rewrite .Nd rewrite specified files without modification .Sh SYNOPSIS .Nm zfs .Cm rewrite -.Oo Fl rvx Ns Oc +.Oo Fl Prvx Ns Oc .Op Fl l Ar length .Op Fl o Ar offset .Ar file Ns | Ns Ar directory Ns … . .Sh DESCRIPTION Rewrite blocks of specified .Ar file as is without modification at a new location and possibly with new properties, such as checksum, compression, dedup, copies, etc, as if they were atomically read and written back. .Bl -tag -width "-r" +.It Fl P +Perform physical rewrite, preserving logical birth time of blocks. +By default, rewrite updates logical birth times, making blocks appear +as modified in snapshots and incremental send streams. +Physical rewrite preserves logical birth times, avoiding unnecessary +inclusion in incremental streams. +Physical rewrite requires the +.Sy physical_rewrite +feature to be enabled on the pool. .It Fl l Ar length Rewrite at most this number of bytes. .It Fl o Ar offset Start at this offset in bytes. .It Fl r Recurse into directories. .It Fl v Print names of all successfully rewritten files. .It Fl x Don't cross file system mount points when recursing. .El .Sh NOTES Rewrite of cloned blocks and blocks that are part of any snapshots, same as some property changes may increase pool space usage. Holes that were never written or were previously zero-compressed are not rewritten and will remain holes even if compression is disabled. .Pp -Rewritten blocks will be seen as modified in next snapshot and as such -included into the incremental -.Nm zfs Cm send -stream. -.Pp If a .Fl l or .Fl o value request a rewrite to regions past the end of the file, then those regions are silently ignored, and no error is reported. +.Pp +By default, rewritten blocks update their logical birth time, +meaning they will be included in incremental +.Nm zfs Cm send +streams as modified data. +When the +.Fl P +flag is used, rewritten blocks preserve their logical birth time, since +there are no user data changes. . .Sh SEE ALSO -.Xr zfsprops 7 +.Xr zfsprops 7 , +.Xr zpool-features 7 diff --git a/module/zcommon/zfeature_common.c b/module/zcommon/zfeature_common.c index 0b37530b0e11..6ba9892eeb64 100644 --- a/module/zcommon/zfeature_common.c +++ b/module/zcommon/zfeature_common.c @@ -1,812 +1,824 @@ // SPDX-License-Identifier: CDDL-1.0 /* * 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 https://opensource.org/licenses/CDDL-1.0. * 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) 2011, 2018 by Delphix. All rights reserved. * Copyright (c) 2013 by Saso Kiselkov. All rights reserved. * Copyright (c) 2013, Joyent, Inc. All rights reserved. * Copyright (c) 2014, Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2019, 2024, Klara, Inc. * Copyright (c) 2019, Allan Jude */ #ifndef _KERNEL #include #include #include #include #include #endif #include #include #include #include #include #include #include "zfeature_common.h" /* * Set to disable all feature checks while opening pools, allowing pools with * unsupported features to be opened. Set for testing only. */ boolean_t zfeature_checks_disable = B_FALSE; zfeature_info_t spa_feature_table[SPA_FEATURES]; /* * Valid characters for feature guids. This list is mainly for aesthetic * purposes and could be expanded in the future. There are different allowed * characters in the guids reverse dns portion (before the colon) and its * short name (after the colon). */ static int valid_char(char c, boolean_t after_colon) { return ((c >= 'a' && c <= 'z') || (c >= '0' && c <= '9') || (after_colon && c == '_') || (!after_colon && (c == '.' || c == '-'))); } /* * Every feature guid must contain exactly one colon which separates a reverse * dns organization name from the feature's "short" name (e.g. * "com.company:feature_name"). */ boolean_t zfeature_is_valid_guid(const char *name) { int i; boolean_t has_colon = B_FALSE; i = 0; while (name[i] != '\0') { char c = name[i++]; if (c == ':') { if (has_colon) return (B_FALSE); has_colon = B_TRUE; continue; } if (!valid_char(c, has_colon)) return (B_FALSE); } return (has_colon); } boolean_t zfeature_is_supported(const char *guid) { if (zfeature_checks_disable) return (B_TRUE); for (spa_feature_t i = 0; i < SPA_FEATURES; i++) { zfeature_info_t *feature = &spa_feature_table[i]; if (!feature->fi_zfs_mod_supported) continue; if (strcmp(guid, feature->fi_guid) == 0) return (B_TRUE); } return (B_FALSE); } int zfeature_lookup_guid(const char *guid, spa_feature_t *res) { for (spa_feature_t i = 0; i < SPA_FEATURES; i++) { zfeature_info_t *feature = &spa_feature_table[i]; if (!feature->fi_zfs_mod_supported) continue; if (strcmp(guid, feature->fi_guid) == 0) { if (res != NULL) *res = i; return (0); } } return (ENOENT); } int zfeature_lookup_name(const char *name, spa_feature_t *res) { for (spa_feature_t i = 0; i < SPA_FEATURES; i++) { zfeature_info_t *feature = &spa_feature_table[i]; if (!feature->fi_zfs_mod_supported) continue; if (strcmp(name, feature->fi_uname) == 0) { if (res != NULL) *res = i; return (0); } } return (ENOENT); } boolean_t zfeature_depends_on(spa_feature_t fid, spa_feature_t check) { zfeature_info_t *feature = &spa_feature_table[fid]; for (int i = 0; feature->fi_depends[i] != SPA_FEATURE_NONE; i++) { if (feature->fi_depends[i] == check) return (B_TRUE); } return (B_FALSE); } static boolean_t deps_contains_feature(const spa_feature_t *deps, const spa_feature_t feature) { for (int i = 0; deps[i] != SPA_FEATURE_NONE; i++) if (deps[i] == feature) return (B_TRUE); return (B_FALSE); } #define STRCMP ((int(*)(const void *, const void *))&strcmp) struct zfs_mod_supported_features { void *tree; boolean_t all_features; }; struct zfs_mod_supported_features * zfs_mod_list_supported(const char *scope) { #if defined(__FreeBSD__) || defined(_KERNEL) || defined(LIB_ZPOOL_BUILD) (void) scope; return (NULL); #else struct zfs_mod_supported_features *ret = calloc(1, sizeof (*ret)); if (ret == NULL) return (NULL); DIR *sysfs_dir = NULL; char path[128]; if (snprintf(path, sizeof (path), "%s/%s", ZFS_SYSFS_DIR, scope) < sizeof (path)) sysfs_dir = opendir(path); if (sysfs_dir == NULL && errno == ENOENT) { if (snprintf(path, sizeof (path), "%s/%s", ZFS_SYSFS_ALT_DIR, scope) < sizeof (path)) sysfs_dir = opendir(path); } if (sysfs_dir == NULL) { ret->all_features = errno == ENOENT && (access(ZFS_SYSFS_DIR, F_OK) == 0 || access(ZFS_SYSFS_ALT_DIR, F_OK) == 0); return (ret); } struct dirent *node; while ((node = readdir(sysfs_dir)) != NULL) { if (strcmp(node->d_name, ".") == 0 || strcmp(node->d_name, "..") == 0) continue; char *name = strdup(node->d_name); if (name == NULL) { goto nomem; } if (tsearch(name, &ret->tree, STRCMP) == NULL) { /* * Don't bother checking for duplicate entries: * we're iterating a single directory. */ free(name); goto nomem; } } end: closedir(sysfs_dir); return (ret); nomem: zfs_mod_list_supported_free(ret); ret = NULL; goto end; #endif } void zfs_mod_list_supported_free(struct zfs_mod_supported_features *list) { #if !defined(__FreeBSD__) && !defined(_KERNEL) && !defined(LIB_ZPOOL_BUILD) if (list) { tdestroy(list->tree, free); free(list); } #else (void) list; #endif } #if !defined(_KERNEL) && !defined(LIB_ZPOOL_BUILD) static boolean_t zfs_mod_supported_impl(const char *scope, const char *name, const char *sysfs) { char path[128]; if (snprintf(path, sizeof (path), "%s%s%s%s%s", sysfs, scope == NULL ? "" : "/", scope ?: "", name == NULL ? "" : "/", name ?: "") < sizeof (path)) return (access(path, F_OK) == 0); else return (B_FALSE); } boolean_t zfs_mod_supported(const char *scope, const char *name, const struct zfs_mod_supported_features *sfeatures) { boolean_t supported; if (sfeatures != NULL) return (sfeatures->all_features || tfind(name, &sfeatures->tree, STRCMP)); /* * Check both the primary and alternate sysfs locations to determine * if the required functionality is supported. */ supported = (zfs_mod_supported_impl(scope, name, ZFS_SYSFS_DIR) || zfs_mod_supported_impl(scope, name, ZFS_SYSFS_ALT_DIR)); /* * For backwards compatibility with kernel modules that predate * supported feature/property checking. Report the feature/property * as supported if the kernel module is loaded but the requested * scope directory does not exist. */ if (supported == B_FALSE) { if ((access(ZFS_SYSFS_DIR, F_OK) == 0 && !zfs_mod_supported_impl(scope, NULL, ZFS_SYSFS_DIR)) || (access(ZFS_SYSFS_ALT_DIR, F_OK) == 0 && !zfs_mod_supported_impl(scope, NULL, ZFS_SYSFS_ALT_DIR))) { supported = B_TRUE; } } return (supported); } #endif static boolean_t zfs_mod_supported_feature(const char *name, const struct zfs_mod_supported_features *sfeatures) { /* * The zfs module spa_feature_table[], whether in-kernel or in * libzpool, always supports all the features. libzfs needs to * query the running module, via sysfs, to determine which * features are supported. * * The equivalent _can_ be done on FreeBSD by way of the sysctl * tree, but this has not been done yet. Therefore, we return * that all features are supported. */ #if defined(_KERNEL) || defined(LIB_ZPOOL_BUILD) || defined(__FreeBSD__) (void) name, (void) sfeatures; return (B_TRUE); #else return (zfs_mod_supported(ZFS_SYSFS_POOL_FEATURES, name, sfeatures)); #endif } static void zfeature_register(spa_feature_t fid, const char *guid, const char *name, const char *desc, zfeature_flags_t flags, zfeature_type_t type, const spa_feature_t *deps, const struct zfs_mod_supported_features *sfeatures) { zfeature_info_t *feature = &spa_feature_table[fid]; static const spa_feature_t nodeps[] = { SPA_FEATURE_NONE }; ASSERT(name != NULL); ASSERT(desc != NULL); ASSERT((flags & ZFEATURE_FLAG_READONLY_COMPAT) == 0 || (flags & ZFEATURE_FLAG_MOS) == 0); ASSERT3U(fid, <, SPA_FEATURES); ASSERT(zfeature_is_valid_guid(guid)); if (deps == NULL) deps = nodeps; VERIFY(((flags & ZFEATURE_FLAG_PER_DATASET) == 0) || (deps_contains_feature(deps, SPA_FEATURE_EXTENSIBLE_DATASET))); feature->fi_feature = fid; feature->fi_guid = guid; feature->fi_uname = name; feature->fi_desc = desc; feature->fi_flags = flags; feature->fi_type = type; feature->fi_depends = deps; feature->fi_zfs_mod_supported = zfs_mod_supported_feature(guid, sfeatures); } /* * Every feature has a GUID of the form com.example:feature_name. The * reversed DNS name ensures that the feature's GUID is unique across all ZFS * implementations. This allows companies to independently develop and * release features. Examples include org.delphix and org.datto. Previously, * features developed on one implementation have used that implementation's * domain name (e.g. org.illumos and org.zfsonlinux). Use of the org.openzfs * domain name is recommended for new features which are developed by the * OpenZFS community and its platforms. This domain may optionally be used by * companies developing features for initial release through an OpenZFS * implementation. Use of the org.openzfs domain requires reserving the * feature name in advance with the OpenZFS project. */ void zpool_feature_init(void) { struct zfs_mod_supported_features *sfeatures = zfs_mod_list_supported(ZFS_SYSFS_POOL_FEATURES); zfeature_register(SPA_FEATURE_ASYNC_DESTROY, "com.delphix:async_destroy", "async_destroy", "Destroy filesystems asynchronously.", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); zfeature_register(SPA_FEATURE_EMPTY_BPOBJ, "com.delphix:empty_bpobj", "empty_bpobj", "Snapshots use less space.", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); zfeature_register(SPA_FEATURE_LZ4_COMPRESS, "org.illumos:lz4_compress", "lz4_compress", "LZ4 compression algorithm support.", ZFEATURE_FLAG_ACTIVATE_ON_ENABLE, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); zfeature_register(SPA_FEATURE_MULTI_VDEV_CRASH_DUMP, "com.joyent:multi_vdev_crash_dump", "multi_vdev_crash_dump", "Crash dumps to multiple vdev pools.", 0, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); zfeature_register(SPA_FEATURE_SPACEMAP_HISTOGRAM, "com.delphix:spacemap_histogram", "spacemap_histogram", "Spacemaps maintain space histograms.", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); zfeature_register(SPA_FEATURE_ENABLED_TXG, "com.delphix:enabled_txg", "enabled_txg", "Record txg at which a feature is enabled", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); { static const spa_feature_t hole_birth_deps[] = { SPA_FEATURE_ENABLED_TXG, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_HOLE_BIRTH, "com.delphix:hole_birth", "hole_birth", "Retain hole birth txg for more precise zfs send", ZFEATURE_FLAG_MOS | ZFEATURE_FLAG_ACTIVATE_ON_ENABLE, ZFEATURE_TYPE_BOOLEAN, hole_birth_deps, sfeatures); } zfeature_register(SPA_FEATURE_POOL_CHECKPOINT, "com.delphix:zpool_checkpoint", "zpool_checkpoint", "Pool state can be checkpointed, allowing rewind later.", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); zfeature_register(SPA_FEATURE_SPACEMAP_V2, "com.delphix:spacemap_v2", "spacemap_v2", "Space maps representing large segments are more efficient.", ZFEATURE_FLAG_READONLY_COMPAT | ZFEATURE_FLAG_ACTIVATE_ON_ENABLE, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); zfeature_register(SPA_FEATURE_EXTENSIBLE_DATASET, "com.delphix:extensible_dataset", "extensible_dataset", "Enhanced dataset functionality, used by other features.", 0, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); { static const spa_feature_t bookmarks_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_BOOKMARKS, "com.delphix:bookmarks", "bookmarks", "\"zfs bookmark\" command", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, bookmarks_deps, sfeatures); } { static const spa_feature_t filesystem_limits_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_FS_SS_LIMIT, "com.joyent:filesystem_limits", "filesystem_limits", "Filesystem and snapshot limits.", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, filesystem_limits_deps, sfeatures); } zfeature_register(SPA_FEATURE_EMBEDDED_DATA, "com.delphix:embedded_data", "embedded_data", "Blocks which compress very well use even less space.", ZFEATURE_FLAG_MOS | ZFEATURE_FLAG_ACTIVATE_ON_ENABLE, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); { static const spa_feature_t livelist_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_LIVELIST, "com.delphix:livelist", "livelist", "Improved clone deletion performance.", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, livelist_deps, sfeatures); } { static const spa_feature_t log_spacemap_deps[] = { SPA_FEATURE_SPACEMAP_V2, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_LOG_SPACEMAP, "com.delphix:log_spacemap", "log_spacemap", "Log metaslab changes on a single spacemap and " "flush them periodically.", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, log_spacemap_deps, sfeatures); } { static const spa_feature_t large_blocks_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_LARGE_BLOCKS, "org.open-zfs:large_blocks", "large_blocks", "Support for blocks larger than 128KB.", ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN, large_blocks_deps, sfeatures); } { static const spa_feature_t large_dnode_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_LARGE_DNODE, "org.zfsonlinux:large_dnode", "large_dnode", "Variable on-disk size of dnodes.", ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN, large_dnode_deps, sfeatures); } { static const spa_feature_t sha512_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_SHA512, "org.illumos:sha512", "sha512", "SHA-512/256 hash algorithm.", ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN, sha512_deps, sfeatures); } { static const spa_feature_t skein_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_SKEIN, "org.illumos:skein", "skein", "Skein hash algorithm.", ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN, skein_deps, sfeatures); } { static const spa_feature_t edonr_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_EDONR, "org.illumos:edonr", "edonr", "Edon-R hash algorithm.", ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN, edonr_deps, sfeatures); } { static const spa_feature_t redact_books_deps[] = { SPA_FEATURE_BOOKMARK_V2, SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_BOOKMARKS, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_REDACTION_BOOKMARKS, "com.delphix:redaction_bookmarks", "redaction_bookmarks", "Support for bookmarks which store redaction lists for zfs " "redacted send/recv.", 0, ZFEATURE_TYPE_BOOLEAN, redact_books_deps, sfeatures); } { static const spa_feature_t redact_datasets_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_REDACTED_DATASETS, "com.delphix:redacted_datasets", "redacted_datasets", "Support for redacted datasets, produced by receiving " "a redacted zfs send stream.", ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_UINT64_ARRAY, redact_datasets_deps, sfeatures); } { static const spa_feature_t bookmark_written_deps[] = { SPA_FEATURE_BOOKMARK_V2, SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_BOOKMARKS, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_BOOKMARK_WRITTEN, "com.delphix:bookmark_written", "bookmark_written", "Additional accounting, enabling the written# " "property (space written since a bookmark), " "and estimates of send stream sizes for incrementals from " "bookmarks.", 0, ZFEATURE_TYPE_BOOLEAN, bookmark_written_deps, sfeatures); } zfeature_register(SPA_FEATURE_DEVICE_REMOVAL, "com.delphix:device_removal", "device_removal", "Top-level vdevs can be removed, reducing logical pool size.", ZFEATURE_FLAG_MOS, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); { static const spa_feature_t obsolete_counts_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_DEVICE_REMOVAL, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_OBSOLETE_COUNTS, "com.delphix:obsolete_counts", "obsolete_counts", "Reduce memory used by removed devices when their blocks " "are freed or remapped.", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, obsolete_counts_deps, sfeatures); } { static const spa_feature_t userobj_accounting_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_USEROBJ_ACCOUNTING, "org.zfsonlinux:userobj_accounting", "userobj_accounting", "User/Group object accounting.", ZFEATURE_FLAG_READONLY_COMPAT | ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN, userobj_accounting_deps, sfeatures); } { static const spa_feature_t bookmark_v2_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_BOOKMARKS, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_BOOKMARK_V2, "com.datto:bookmark_v2", "bookmark_v2", "Support for larger bookmarks", 0, ZFEATURE_TYPE_BOOLEAN, bookmark_v2_deps, sfeatures); } { static const spa_feature_t encryption_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_BOOKMARK_V2, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_ENCRYPTION, "com.datto:encryption", "encryption", "Support for dataset level encryption", ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN, encryption_deps, sfeatures); } { static const spa_feature_t project_quota_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_PROJECT_QUOTA, "org.zfsonlinux:project_quota", "project_quota", "space/object accounting based on project ID.", ZFEATURE_FLAG_READONLY_COMPAT | ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN, project_quota_deps, sfeatures); } zfeature_register(SPA_FEATURE_ALLOCATION_CLASSES, "org.zfsonlinux:allocation_classes", "allocation_classes", "Support for separate allocation classes.", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); zfeature_register(SPA_FEATURE_RESILVER_DEFER, "com.datto:resilver_defer", "resilver_defer", "Support for deferring new resilvers when one is already running.", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); zfeature_register(SPA_FEATURE_DEVICE_REBUILD, "org.openzfs:device_rebuild", "device_rebuild", "Support for sequential mirror/dRAID device rebuilds", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); { static const spa_feature_t zstd_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_ZSTD_COMPRESS, "org.freebsd:zstd_compress", "zstd_compress", "zstd compression algorithm support.", ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN, zstd_deps, sfeatures); } zfeature_register(SPA_FEATURE_DRAID, "org.openzfs:draid", "draid", "Support for distributed spare RAID", ZFEATURE_FLAG_MOS, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); { static const spa_feature_t zilsaxattr_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_ZILSAXATTR, "org.openzfs:zilsaxattr", "zilsaxattr", "Support for xattr=sa extended attribute logging in ZIL.", ZFEATURE_FLAG_PER_DATASET | ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, zilsaxattr_deps, sfeatures); } zfeature_register(SPA_FEATURE_HEAD_ERRLOG, "com.delphix:head_errlog", "head_errlog", "Support for per-dataset on-disk error logs.", ZFEATURE_FLAG_ACTIVATE_ON_ENABLE, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); { static const spa_feature_t blake3_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_BLAKE3, "org.openzfs:blake3", "blake3", "BLAKE3 hash algorithm.", ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN, blake3_deps, sfeatures); } zfeature_register(SPA_FEATURE_BLOCK_CLONING, "com.fudosecurity:block_cloning", "block_cloning", "Support for block cloning via Block Reference Table.", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); zfeature_register(SPA_FEATURE_BLOCK_CLONING_ENDIAN, "com.truenas:block_cloning_endian", "block_cloning_endian", "Fixes BRT ZAP endianness on new pools.", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); zfeature_register(SPA_FEATURE_AVZ_V2, "com.klarasystems:vdev_zaps_v2", "vdev_zaps_v2", "Support for root vdev ZAP.", ZFEATURE_FLAG_MOS, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); { static const spa_feature_t redact_list_spill_deps[] = { SPA_FEATURE_REDACTION_BOOKMARKS, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_REDACTION_LIST_SPILL, "com.delphix:redaction_list_spill", "redaction_list_spill", "Support for increased number of redaction_snapshot " "arguments in zfs redact.", 0, ZFEATURE_TYPE_BOOLEAN, redact_list_spill_deps, sfeatures); } zfeature_register(SPA_FEATURE_RAIDZ_EXPANSION, "org.openzfs:raidz_expansion", "raidz_expansion", "Support for raidz expansion", ZFEATURE_FLAG_MOS, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); zfeature_register(SPA_FEATURE_FAST_DEDUP, "com.klarasystems:fast_dedup", "fast_dedup", "Support for advanced deduplication", ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); { static const spa_feature_t longname_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_LONGNAME, "org.zfsonlinux:longname", "longname", "support filename up to 1024 bytes", ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN, longname_deps, sfeatures); } { static const spa_feature_t large_microzap_deps[] = { SPA_FEATURE_EXTENSIBLE_DATASET, SPA_FEATURE_LARGE_BLOCKS, SPA_FEATURE_NONE }; zfeature_register(SPA_FEATURE_LARGE_MICROZAP, "com.klarasystems:large_microzap", "large_microzap", "Support for microzaps larger than 128KB.", ZFEATURE_FLAG_PER_DATASET | ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, large_microzap_deps, sfeatures); } zfeature_register(SPA_FEATURE_DYNAMIC_GANG_HEADER, "com.klarasystems:dynamic_gang_header", "dynamic_gang_header", "Support for dynamically sized gang headers", ZFEATURE_FLAG_MOS | ZFEATURE_FLAG_NO_UPGRADE, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures); + { + static const spa_feature_t physical_rewrite_deps[] = { + SPA_FEATURE_EXTENSIBLE_DATASET, + SPA_FEATURE_NONE + }; + zfeature_register(SPA_FEATURE_PHYSICAL_REWRITE, + "com.truenas:physical_rewrite", "physical_rewrite", + "Support for preserving logical birth time during rewrite.", + ZFEATURE_FLAG_READONLY_COMPAT | ZFEATURE_FLAG_PER_DATASET, + ZFEATURE_TYPE_BOOLEAN, physical_rewrite_deps, sfeatures); + } + zfs_mod_list_supported_free(sfeatures); } #if defined(_KERNEL) EXPORT_SYMBOL(zfeature_lookup_guid); EXPORT_SYMBOL(zfeature_lookup_name); EXPORT_SYMBOL(zfeature_is_supported); EXPORT_SYMBOL(zfeature_is_valid_guid); EXPORT_SYMBOL(zfeature_depends_on); EXPORT_SYMBOL(zpool_feature_init); EXPORT_SYMBOL(spa_feature_table); #endif diff --git a/module/zfs/dbuf.c b/module/zfs/dbuf.c index a96666a4675f..432c99cec960 100644 --- a/module/zfs/dbuf.c +++ b/module/zfs/dbuf.c @@ -1,5454 +1,5511 @@ // SPDX-License-Identifier: CDDL-1.0 /* * 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 https://opensource.org/licenses/CDDL-1.0. * 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 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2012, 2020 by Delphix. All rights reserved. * Copyright (c) 2013 by Saso Kiselkov. All rights reserved. * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved. * Copyright (c) 2019, Klara Inc. * Copyright (c) 2019, Allan Jude * Copyright (c) 2021, 2022 by Pawel Jakub Dawidek */ #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 static kstat_t *dbuf_ksp; typedef struct dbuf_stats { /* * Various statistics about the size of the dbuf cache. */ kstat_named_t cache_count; kstat_named_t cache_size_bytes; kstat_named_t cache_size_bytes_max; /* * Statistics regarding the bounds on the dbuf cache size. */ kstat_named_t cache_target_bytes; kstat_named_t cache_lowater_bytes; kstat_named_t cache_hiwater_bytes; /* * Total number of dbuf cache evictions that have occurred. */ kstat_named_t cache_total_evicts; /* * The distribution of dbuf levels in the dbuf cache and * the total size of all dbufs at each level. */ kstat_named_t cache_levels[DN_MAX_LEVELS]; kstat_named_t cache_levels_bytes[DN_MAX_LEVELS]; /* * Statistics about the dbuf hash table. */ kstat_named_t hash_hits; kstat_named_t hash_misses; kstat_named_t hash_collisions; kstat_named_t hash_elements; /* * Number of sublists containing more than one dbuf in the dbuf * hash table. Keep track of the longest hash chain. */ kstat_named_t hash_chains; kstat_named_t hash_chain_max; /* * Number of times a dbuf_create() discovers that a dbuf was * already created and in the dbuf hash table. */ kstat_named_t hash_insert_race; /* * Number of entries in the hash table dbuf and mutex arrays. */ kstat_named_t hash_table_count; kstat_named_t hash_mutex_count; /* * Statistics about the size of the metadata dbuf cache. */ kstat_named_t metadata_cache_count; kstat_named_t metadata_cache_size_bytes; kstat_named_t metadata_cache_size_bytes_max; /* * For diagnostic purposes, this is incremented whenever we can't add * something to the metadata cache because it's full, and instead put * the data in the regular dbuf cache. */ kstat_named_t metadata_cache_overflow; } dbuf_stats_t; dbuf_stats_t dbuf_stats = { { "cache_count", KSTAT_DATA_UINT64 }, { "cache_size_bytes", KSTAT_DATA_UINT64 }, { "cache_size_bytes_max", KSTAT_DATA_UINT64 }, { "cache_target_bytes", KSTAT_DATA_UINT64 }, { "cache_lowater_bytes", KSTAT_DATA_UINT64 }, { "cache_hiwater_bytes", KSTAT_DATA_UINT64 }, { "cache_total_evicts", KSTAT_DATA_UINT64 }, { { "cache_levels_N", KSTAT_DATA_UINT64 } }, { { "cache_levels_bytes_N", KSTAT_DATA_UINT64 } }, { "hash_hits", KSTAT_DATA_UINT64 }, { "hash_misses", KSTAT_DATA_UINT64 }, { "hash_collisions", KSTAT_DATA_UINT64 }, { "hash_elements", KSTAT_DATA_UINT64 }, { "hash_chains", KSTAT_DATA_UINT64 }, { "hash_chain_max", KSTAT_DATA_UINT64 }, { "hash_insert_race", KSTAT_DATA_UINT64 }, { "hash_table_count", KSTAT_DATA_UINT64 }, { "hash_mutex_count", KSTAT_DATA_UINT64 }, { "metadata_cache_count", KSTAT_DATA_UINT64 }, { "metadata_cache_size_bytes", KSTAT_DATA_UINT64 }, { "metadata_cache_size_bytes_max", KSTAT_DATA_UINT64 }, { "metadata_cache_overflow", KSTAT_DATA_UINT64 } }; struct { wmsum_t cache_count; wmsum_t cache_total_evicts; wmsum_t cache_levels[DN_MAX_LEVELS]; wmsum_t cache_levels_bytes[DN_MAX_LEVELS]; wmsum_t hash_hits; wmsum_t hash_misses; wmsum_t hash_collisions; wmsum_t hash_elements; wmsum_t hash_chains; wmsum_t hash_insert_race; wmsum_t metadata_cache_count; wmsum_t metadata_cache_overflow; } dbuf_sums; #define DBUF_STAT_INCR(stat, val) \ wmsum_add(&dbuf_sums.stat, val) #define DBUF_STAT_DECR(stat, val) \ DBUF_STAT_INCR(stat, -(val)) #define DBUF_STAT_BUMP(stat) \ DBUF_STAT_INCR(stat, 1) #define DBUF_STAT_BUMPDOWN(stat) \ DBUF_STAT_INCR(stat, -1) #define DBUF_STAT_MAX(stat, v) { \ uint64_t _m; \ while ((v) > (_m = dbuf_stats.stat.value.ui64) && \ (_m != atomic_cas_64(&dbuf_stats.stat.value.ui64, _m, (v))))\ continue; \ } static void dbuf_write(dbuf_dirty_record_t *dr, arc_buf_t *data, dmu_tx_t *tx); static void dbuf_sync_leaf_verify_bonus_dnode(dbuf_dirty_record_t *dr); /* * Global data structures and functions for the dbuf cache. */ static kmem_cache_t *dbuf_kmem_cache; kmem_cache_t *dbuf_dirty_kmem_cache; static taskq_t *dbu_evict_taskq; static kthread_t *dbuf_cache_evict_thread; static kmutex_t dbuf_evict_lock; static kcondvar_t dbuf_evict_cv; static boolean_t dbuf_evict_thread_exit; /* * There are two dbuf caches; each dbuf can only be in one of them at a time. * * 1. Cache of metadata dbufs, to help make read-heavy administrative commands * from /sbin/zfs run faster. The "metadata cache" specifically stores dbufs * that represent the metadata that describes filesystems/snapshots/ * bookmarks/properties/etc. We only evict from this cache when we export a * pool, to short-circuit as much I/O as possible for all administrative * commands that need the metadata. There is no eviction policy for this * cache, because we try to only include types in it which would occupy a * very small amount of space per object but create a large impact on the * performance of these commands. Instead, after it reaches a maximum size * (which should only happen on very small memory systems with a very large * number of filesystem objects), we stop taking new dbufs into the * metadata cache, instead putting them in the normal dbuf cache. * * 2. LRU cache of dbufs. The dbuf cache maintains a list of dbufs that * are not currently held but have been recently released. These dbufs * are not eligible for arc eviction until they are aged out of the cache. * Dbufs that are aged out of the cache will be immediately destroyed and * become eligible for arc eviction. * * Dbufs are added to these caches once the last hold is released. If a dbuf is * later accessed and still exists in the dbuf cache, then it will be removed * from the cache and later re-added to the head of the cache. * * If a given dbuf meets the requirements for the metadata cache, it will go * there, otherwise it will be considered for the generic LRU dbuf cache. The * caches and the refcounts tracking their sizes are stored in an array indexed * by those caches' matching enum values (from dbuf_cached_state_t). */ typedef struct dbuf_cache { multilist_t cache; zfs_refcount_t size ____cacheline_aligned; } dbuf_cache_t; dbuf_cache_t dbuf_caches[DB_CACHE_MAX]; /* Size limits for the caches */ static uint64_t dbuf_cache_max_bytes = UINT64_MAX; static uint64_t dbuf_metadata_cache_max_bytes = UINT64_MAX; /* Set the default sizes of the caches to log2 fraction of arc size */ static uint_t dbuf_cache_shift = 5; static uint_t dbuf_metadata_cache_shift = 6; /* Set the dbuf hash mutex count as log2 shift (dynamic by default) */ static uint_t dbuf_mutex_cache_shift = 0; static unsigned long dbuf_cache_target_bytes(void); static unsigned long dbuf_metadata_cache_target_bytes(void); /* * The LRU dbuf cache uses a three-stage eviction policy: * - A low water marker designates when the dbuf eviction thread * should stop evicting from the dbuf cache. * - When we reach the maximum size (aka mid water mark), we * signal the eviction thread to run. * - The high water mark indicates when the eviction thread * is unable to keep up with the incoming load and eviction must * happen in the context of the calling thread. * * The dbuf cache: * (max size) * low water mid water hi water * +----------------------------------------+----------+----------+ * | | | | * | | | | * | | | | * | | | | * +----------------------------------------+----------+----------+ * stop signal evict * evicting eviction directly * thread * * The high and low water marks indicate the operating range for the eviction * thread. The low water mark is, by default, 90% of the total size of the * cache and the high water mark is at 110% (both of these percentages can be * changed by setting dbuf_cache_lowater_pct and dbuf_cache_hiwater_pct, * respectively). The eviction thread will try to ensure that the cache remains * within this range by waking up every second and checking if the cache is * above the low water mark. The thread can also be woken up by callers adding * elements into the cache if the cache is larger than the mid water (i.e max * cache size). Once the eviction thread is woken up and eviction is required, * it will continue evicting buffers until it's able to reduce the cache size * to the low water mark. If the cache size continues to grow and hits the high * water mark, then callers adding elements to the cache will begin to evict * directly from the cache until the cache is no longer above the high water * mark. */ /* * The percentage above and below the maximum cache size. */ static uint_t dbuf_cache_hiwater_pct = 10; static uint_t dbuf_cache_lowater_pct = 10; static int dbuf_cons(void *vdb, void *unused, int kmflag) { (void) unused, (void) kmflag; dmu_buf_impl_t *db = vdb; memset(db, 0, sizeof (dmu_buf_impl_t)); mutex_init(&db->db_mtx, NULL, MUTEX_NOLOCKDEP, NULL); rw_init(&db->db_rwlock, NULL, RW_NOLOCKDEP, NULL); cv_init(&db->db_changed, NULL, CV_DEFAULT, NULL); multilist_link_init(&db->db_cache_link); zfs_refcount_create(&db->db_holds); return (0); } static void dbuf_dest(void *vdb, void *unused) { (void) unused; dmu_buf_impl_t *db = vdb; mutex_destroy(&db->db_mtx); rw_destroy(&db->db_rwlock); cv_destroy(&db->db_changed); ASSERT(!multilist_link_active(&db->db_cache_link)); zfs_refcount_destroy(&db->db_holds); } /* * dbuf hash table routines */ static dbuf_hash_table_t dbuf_hash_table; /* * We use Cityhash for this. It's fast, and has good hash properties without * requiring any large static buffers. */ static uint64_t dbuf_hash(void *os, uint64_t obj, uint8_t lvl, uint64_t blkid) { return (cityhash4((uintptr_t)os, obj, (uint64_t)lvl, blkid)); } #define DTRACE_SET_STATE(db, why) \ DTRACE_PROBE2(dbuf__state_change, dmu_buf_impl_t *, db, \ const char *, why) #define DBUF_EQUAL(dbuf, os, obj, level, blkid) \ ((dbuf)->db.db_object == (obj) && \ (dbuf)->db_objset == (os) && \ (dbuf)->db_level == (level) && \ (dbuf)->db_blkid == (blkid)) dmu_buf_impl_t * dbuf_find(objset_t *os, uint64_t obj, uint8_t level, uint64_t blkid, uint64_t *hash_out) { dbuf_hash_table_t *h = &dbuf_hash_table; uint64_t hv; uint64_t idx; dmu_buf_impl_t *db; hv = dbuf_hash(os, obj, level, blkid); idx = hv & h->hash_table_mask; mutex_enter(DBUF_HASH_MUTEX(h, idx)); for (db = h->hash_table[idx]; db != NULL; db = db->db_hash_next) { if (DBUF_EQUAL(db, os, obj, level, blkid)) { mutex_enter(&db->db_mtx); if (db->db_state != DB_EVICTING) { mutex_exit(DBUF_HASH_MUTEX(h, idx)); return (db); } mutex_exit(&db->db_mtx); } } mutex_exit(DBUF_HASH_MUTEX(h, idx)); if (hash_out != NULL) *hash_out = hv; return (NULL); } static dmu_buf_impl_t * dbuf_find_bonus(objset_t *os, uint64_t object) { dnode_t *dn; dmu_buf_impl_t *db = NULL; if (dnode_hold(os, object, FTAG, &dn) == 0) { rw_enter(&dn->dn_struct_rwlock, RW_READER); if (dn->dn_bonus != NULL) { db = dn->dn_bonus; mutex_enter(&db->db_mtx); } rw_exit(&dn->dn_struct_rwlock); dnode_rele(dn, FTAG); } return (db); } /* * Insert an entry into the hash table. If there is already an element * equal to elem in the hash table, then the already existing element * will be returned and the new element will not be inserted. * Otherwise returns NULL. */ static dmu_buf_impl_t * dbuf_hash_insert(dmu_buf_impl_t *db) { dbuf_hash_table_t *h = &dbuf_hash_table; objset_t *os = db->db_objset; uint64_t obj = db->db.db_object; int level = db->db_level; uint64_t blkid, idx; dmu_buf_impl_t *dbf; uint32_t i; blkid = db->db_blkid; ASSERT3U(dbuf_hash(os, obj, level, blkid), ==, db->db_hash); idx = db->db_hash & h->hash_table_mask; mutex_enter(DBUF_HASH_MUTEX(h, idx)); for (dbf = h->hash_table[idx], i = 0; dbf != NULL; dbf = dbf->db_hash_next, i++) { if (DBUF_EQUAL(dbf, os, obj, level, blkid)) { mutex_enter(&dbf->db_mtx); if (dbf->db_state != DB_EVICTING) { mutex_exit(DBUF_HASH_MUTEX(h, idx)); return (dbf); } mutex_exit(&dbf->db_mtx); } } if (i > 0) { DBUF_STAT_BUMP(hash_collisions); if (i == 1) DBUF_STAT_BUMP(hash_chains); DBUF_STAT_MAX(hash_chain_max, i); } mutex_enter(&db->db_mtx); db->db_hash_next = h->hash_table[idx]; h->hash_table[idx] = db; mutex_exit(DBUF_HASH_MUTEX(h, idx)); DBUF_STAT_BUMP(hash_elements); return (NULL); } /* * This returns whether this dbuf should be stored in the metadata cache, which * is based on whether it's from one of the dnode types that store data related * to traversing dataset hierarchies. */ static boolean_t dbuf_include_in_metadata_cache(dmu_buf_impl_t *db) { DB_DNODE_ENTER(db); dmu_object_type_t type = DB_DNODE(db)->dn_type; DB_DNODE_EXIT(db); /* Check if this dbuf is one of the types we care about */ if (DMU_OT_IS_METADATA_CACHED(type)) { /* If we hit this, then we set something up wrong in dmu_ot */ ASSERT(DMU_OT_IS_METADATA(type)); /* * Sanity check for small-memory systems: don't allocate too * much memory for this purpose. */ if (zfs_refcount_count( &dbuf_caches[DB_DBUF_METADATA_CACHE].size) > dbuf_metadata_cache_target_bytes()) { DBUF_STAT_BUMP(metadata_cache_overflow); return (B_FALSE); } return (B_TRUE); } return (B_FALSE); } /* * Remove an entry from the hash table. It must be in the EVICTING state. */ static void dbuf_hash_remove(dmu_buf_impl_t *db) { dbuf_hash_table_t *h = &dbuf_hash_table; uint64_t idx; dmu_buf_impl_t *dbf, **dbp; ASSERT3U(dbuf_hash(db->db_objset, db->db.db_object, db->db_level, db->db_blkid), ==, db->db_hash); idx = db->db_hash & h->hash_table_mask; /* * We mustn't hold db_mtx to maintain lock ordering: * DBUF_HASH_MUTEX > db_mtx. */ ASSERT(zfs_refcount_is_zero(&db->db_holds)); ASSERT(db->db_state == DB_EVICTING); ASSERT(!MUTEX_HELD(&db->db_mtx)); mutex_enter(DBUF_HASH_MUTEX(h, idx)); dbp = &h->hash_table[idx]; while ((dbf = *dbp) != db) { dbp = &dbf->db_hash_next; ASSERT(dbf != NULL); } *dbp = db->db_hash_next; db->db_hash_next = NULL; if (h->hash_table[idx] && h->hash_table[idx]->db_hash_next == NULL) DBUF_STAT_BUMPDOWN(hash_chains); mutex_exit(DBUF_HASH_MUTEX(h, idx)); DBUF_STAT_BUMPDOWN(hash_elements); } typedef enum { DBVU_EVICTING, DBVU_NOT_EVICTING } dbvu_verify_type_t; static void dbuf_verify_user(dmu_buf_impl_t *db, dbvu_verify_type_t verify_type) { #ifdef ZFS_DEBUG int64_t holds; if (db->db_user == NULL) return; /* Only data blocks support the attachment of user data. */ ASSERT(db->db_level == 0); /* Clients must resolve a dbuf before attaching user data. */ ASSERT(db->db.db_data != NULL); ASSERT3U(db->db_state, ==, DB_CACHED); holds = zfs_refcount_count(&db->db_holds); if (verify_type == DBVU_EVICTING) { /* * Immediate eviction occurs when holds == dirtycnt. * For normal eviction buffers, holds is zero on * eviction, except when dbuf_fix_old_data() calls * dbuf_clear_data(). However, the hold count can grow * during eviction even though db_mtx is held (see * dmu_bonus_hold() for an example), so we can only * test the generic invariant that holds >= dirtycnt. */ ASSERT3U(holds, >=, db->db_dirtycnt); } else { if (db->db_user_immediate_evict == TRUE) ASSERT3U(holds, >=, db->db_dirtycnt); else ASSERT3U(holds, >, 0); } #endif } static void dbuf_evict_user(dmu_buf_impl_t *db) { dmu_buf_user_t *dbu = db->db_user; ASSERT(MUTEX_HELD(&db->db_mtx)); if (dbu == NULL) return; dbuf_verify_user(db, DBVU_EVICTING); db->db_user = NULL; #ifdef ZFS_DEBUG if (dbu->dbu_clear_on_evict_dbufp != NULL) *dbu->dbu_clear_on_evict_dbufp = NULL; #endif if (db->db_caching_status != DB_NO_CACHE) { /* * This is a cached dbuf, so the size of the user data is * included in its cached amount. We adjust it here because the * user data has already been detached from the dbuf, and the * sync functions are not supposed to touch it (the dbuf might * not exist anymore by the time the sync functions run. */ uint64_t size = dbu->dbu_size; (void) zfs_refcount_remove_many( &dbuf_caches[db->db_caching_status].size, size, dbu); if (db->db_caching_status == DB_DBUF_CACHE) DBUF_STAT_DECR(cache_levels_bytes[db->db_level], size); } /* * There are two eviction callbacks - one that we call synchronously * and one that we invoke via a taskq. The async one is useful for * avoiding lock order reversals and limiting stack depth. * * Note that if we have a sync callback but no async callback, * it's likely that the sync callback will free the structure * containing the dbu. In that case we need to take care to not * dereference dbu after calling the sync evict func. */ boolean_t has_async = (dbu->dbu_evict_func_async != NULL); if (dbu->dbu_evict_func_sync != NULL) dbu->dbu_evict_func_sync(dbu); if (has_async) { taskq_dispatch_ent(dbu_evict_taskq, dbu->dbu_evict_func_async, dbu, 0, &dbu->dbu_tqent); } } boolean_t dbuf_is_metadata(dmu_buf_impl_t *db) { /* * Consider indirect blocks and spill blocks to be meta data. */ if (db->db_level > 0 || db->db_blkid == DMU_SPILL_BLKID) { return (B_TRUE); } else { boolean_t is_metadata; DB_DNODE_ENTER(db); is_metadata = DMU_OT_IS_METADATA(DB_DNODE(db)->dn_type); DB_DNODE_EXIT(db); return (is_metadata); } } /* * We want to exclude buffers that are on a special allocation class from * L2ARC. */ boolean_t dbuf_is_l2cacheable(dmu_buf_impl_t *db, blkptr_t *bp) { if (db->db_objset->os_secondary_cache == ZFS_CACHE_ALL || (db->db_objset->os_secondary_cache == ZFS_CACHE_METADATA && dbuf_is_metadata(db))) { if (l2arc_exclude_special == 0) return (B_TRUE); /* * bp must be checked in the event it was passed from * dbuf_read_impl() as the result of a the BP being set from * a Direct I/O write in dbuf_read(). See comments in * dbuf_read(). */ blkptr_t *db_bp = bp == NULL ? db->db_blkptr : bp; if (db_bp == NULL || BP_IS_HOLE(db_bp)) return (B_FALSE); uint64_t vdev = DVA_GET_VDEV(db_bp->blk_dva); vdev_t *rvd = db->db_objset->os_spa->spa_root_vdev; vdev_t *vd = NULL; if (vdev < rvd->vdev_children) vd = rvd->vdev_child[vdev]; if (vd == NULL) return (B_TRUE); if (vd->vdev_alloc_bias != VDEV_BIAS_SPECIAL && vd->vdev_alloc_bias != VDEV_BIAS_DEDUP) return (B_TRUE); } return (B_FALSE); } static inline boolean_t dnode_level_is_l2cacheable(blkptr_t *bp, dnode_t *dn, int64_t level) { if (dn->dn_objset->os_secondary_cache == ZFS_CACHE_ALL || (dn->dn_objset->os_secondary_cache == ZFS_CACHE_METADATA && (level > 0 || DMU_OT_IS_METADATA(dn->dn_handle->dnh_dnode->dn_type)))) { if (l2arc_exclude_special == 0) return (B_TRUE); if (bp == NULL || BP_IS_HOLE(bp)) return (B_FALSE); uint64_t vdev = DVA_GET_VDEV(bp->blk_dva); vdev_t *rvd = dn->dn_objset->os_spa->spa_root_vdev; vdev_t *vd = NULL; if (vdev < rvd->vdev_children) vd = rvd->vdev_child[vdev]; if (vd == NULL) return (B_TRUE); if (vd->vdev_alloc_bias != VDEV_BIAS_SPECIAL && vd->vdev_alloc_bias != VDEV_BIAS_DEDUP) return (B_TRUE); } return (B_FALSE); } /* * This function *must* return indices evenly distributed between all * sublists of the multilist. This is needed due to how the dbuf eviction * code is laid out; dbuf_evict_thread() assumes dbufs are evenly * distributed between all sublists and uses this assumption when * deciding which sublist to evict from and how much to evict from it. */ static unsigned int dbuf_cache_multilist_index_func(multilist_t *ml, void *obj) { dmu_buf_impl_t *db = obj; /* * The assumption here, is the hash value for a given * dmu_buf_impl_t will remain constant throughout it's lifetime * (i.e. it's objset, object, level and blkid fields don't change). * Thus, we don't need to store the dbuf's sublist index * on insertion, as this index can be recalculated on removal. * * Also, the low order bits of the hash value are thought to be * distributed evenly. Otherwise, in the case that the multilist * has a power of two number of sublists, each sublists' usage * would not be evenly distributed. In this context full 64bit * division would be a waste of time, so limit it to 32 bits. */ return ((unsigned int)dbuf_hash(db->db_objset, db->db.db_object, db->db_level, db->db_blkid) % multilist_get_num_sublists(ml)); } /* * The target size of the dbuf cache can grow with the ARC target, * unless limited by the tunable dbuf_cache_max_bytes. */ static inline unsigned long dbuf_cache_target_bytes(void) { return (MIN(dbuf_cache_max_bytes, arc_target_bytes() >> dbuf_cache_shift)); } /* * The target size of the dbuf metadata cache can grow with the ARC target, * unless limited by the tunable dbuf_metadata_cache_max_bytes. */ static inline unsigned long dbuf_metadata_cache_target_bytes(void) { return (MIN(dbuf_metadata_cache_max_bytes, arc_target_bytes() >> dbuf_metadata_cache_shift)); } static inline uint64_t dbuf_cache_hiwater_bytes(void) { uint64_t dbuf_cache_target = dbuf_cache_target_bytes(); return (dbuf_cache_target + (dbuf_cache_target * dbuf_cache_hiwater_pct) / 100); } static inline uint64_t dbuf_cache_lowater_bytes(void) { uint64_t dbuf_cache_target = dbuf_cache_target_bytes(); return (dbuf_cache_target - (dbuf_cache_target * dbuf_cache_lowater_pct) / 100); } static inline boolean_t dbuf_cache_above_lowater(void) { return (zfs_refcount_count(&dbuf_caches[DB_DBUF_CACHE].size) > dbuf_cache_lowater_bytes()); } /* * Evict the oldest eligible dbuf from the dbuf cache. */ static void dbuf_evict_one(void) { int idx = multilist_get_random_index(&dbuf_caches[DB_DBUF_CACHE].cache); multilist_sublist_t *mls = multilist_sublist_lock_idx( &dbuf_caches[DB_DBUF_CACHE].cache, idx); ASSERT(!MUTEX_HELD(&dbuf_evict_lock)); dmu_buf_impl_t *db = multilist_sublist_tail(mls); while (db != NULL && mutex_tryenter(&db->db_mtx) == 0) { db = multilist_sublist_prev(mls, db); } DTRACE_PROBE2(dbuf__evict__one, dmu_buf_impl_t *, db, multilist_sublist_t *, mls); if (db != NULL) { multilist_sublist_remove(mls, db); multilist_sublist_unlock(mls); uint64_t size = db->db.db_size; uint64_t usize = dmu_buf_user_size(&db->db); (void) zfs_refcount_remove_many( &dbuf_caches[DB_DBUF_CACHE].size, size, db); (void) zfs_refcount_remove_many( &dbuf_caches[DB_DBUF_CACHE].size, usize, db->db_user); DBUF_STAT_BUMPDOWN(cache_levels[db->db_level]); DBUF_STAT_BUMPDOWN(cache_count); DBUF_STAT_DECR(cache_levels_bytes[db->db_level], size + usize); ASSERT3U(db->db_caching_status, ==, DB_DBUF_CACHE); db->db_caching_status = DB_NO_CACHE; dbuf_destroy(db); DBUF_STAT_BUMP(cache_total_evicts); } else { multilist_sublist_unlock(mls); } } /* * The dbuf evict thread is responsible for aging out dbufs from the * cache. Once the cache has reached it's maximum size, dbufs are removed * and destroyed. The eviction thread will continue running until the size * of the dbuf cache is at or below the maximum size. Once the dbuf is aged * out of the cache it is destroyed and becomes eligible for arc eviction. */ static __attribute__((noreturn)) void dbuf_evict_thread(void *unused) { (void) unused; callb_cpr_t cpr; CALLB_CPR_INIT(&cpr, &dbuf_evict_lock, callb_generic_cpr, FTAG); mutex_enter(&dbuf_evict_lock); while (!dbuf_evict_thread_exit) { while (!dbuf_cache_above_lowater() && !dbuf_evict_thread_exit) { CALLB_CPR_SAFE_BEGIN(&cpr); (void) cv_timedwait_idle_hires(&dbuf_evict_cv, &dbuf_evict_lock, SEC2NSEC(1), MSEC2NSEC(1), 0); CALLB_CPR_SAFE_END(&cpr, &dbuf_evict_lock); } mutex_exit(&dbuf_evict_lock); /* * Keep evicting as long as we're above the low water mark * for the cache. We do this without holding the locks to * minimize lock contention. */ while (dbuf_cache_above_lowater() && !dbuf_evict_thread_exit) { dbuf_evict_one(); } mutex_enter(&dbuf_evict_lock); } dbuf_evict_thread_exit = B_FALSE; cv_broadcast(&dbuf_evict_cv); CALLB_CPR_EXIT(&cpr); /* drops dbuf_evict_lock */ thread_exit(); } /* * Wake up the dbuf eviction thread if the dbuf cache is at its max size. * If the dbuf cache is at its high water mark, then evict a dbuf from the * dbuf cache using the caller's context. */ static void dbuf_evict_notify(uint64_t size) { /* * We check if we should evict without holding the dbuf_evict_lock, * because it's OK to occasionally make the wrong decision here, * and grabbing the lock results in massive lock contention. */ if (size > dbuf_cache_target_bytes()) { /* * Avoid calling dbuf_evict_one() from memory reclaim context * (e.g. Linux kswapd, FreeBSD pagedaemon) to prevent deadlocks. * Memory reclaim threads can get stuck waiting for the dbuf * hash lock. */ if (size > dbuf_cache_hiwater_bytes() && !current_is_reclaim_thread()) { dbuf_evict_one(); } cv_signal(&dbuf_evict_cv); } } /* * Since dbuf cache size is a fraction of target ARC size, ARC calls this when * its target size is reduced due to memory pressure. */ void dbuf_cache_reduce_target_size(void) { uint64_t size = zfs_refcount_count(&dbuf_caches[DB_DBUF_CACHE].size); if (size > dbuf_cache_target_bytes()) cv_signal(&dbuf_evict_cv); } static int dbuf_kstat_update(kstat_t *ksp, int rw) { dbuf_stats_t *ds = ksp->ks_data; dbuf_hash_table_t *h = &dbuf_hash_table; if (rw == KSTAT_WRITE) return (SET_ERROR(EACCES)); ds->cache_count.value.ui64 = wmsum_value(&dbuf_sums.cache_count); ds->cache_size_bytes.value.ui64 = zfs_refcount_count(&dbuf_caches[DB_DBUF_CACHE].size); ds->cache_target_bytes.value.ui64 = dbuf_cache_target_bytes(); ds->cache_hiwater_bytes.value.ui64 = dbuf_cache_hiwater_bytes(); ds->cache_lowater_bytes.value.ui64 = dbuf_cache_lowater_bytes(); ds->cache_total_evicts.value.ui64 = wmsum_value(&dbuf_sums.cache_total_evicts); for (int i = 0; i < DN_MAX_LEVELS; i++) { ds->cache_levels[i].value.ui64 = wmsum_value(&dbuf_sums.cache_levels[i]); ds->cache_levels_bytes[i].value.ui64 = wmsum_value(&dbuf_sums.cache_levels_bytes[i]); } ds->hash_hits.value.ui64 = wmsum_value(&dbuf_sums.hash_hits); ds->hash_misses.value.ui64 = wmsum_value(&dbuf_sums.hash_misses); ds->hash_collisions.value.ui64 = wmsum_value(&dbuf_sums.hash_collisions); ds->hash_elements.value.ui64 = wmsum_value(&dbuf_sums.hash_elements); ds->hash_chains.value.ui64 = wmsum_value(&dbuf_sums.hash_chains); ds->hash_insert_race.value.ui64 = wmsum_value(&dbuf_sums.hash_insert_race); ds->hash_table_count.value.ui64 = h->hash_table_mask + 1; ds->hash_mutex_count.value.ui64 = h->hash_mutex_mask + 1; ds->metadata_cache_count.value.ui64 = wmsum_value(&dbuf_sums.metadata_cache_count); ds->metadata_cache_size_bytes.value.ui64 = zfs_refcount_count( &dbuf_caches[DB_DBUF_METADATA_CACHE].size); ds->metadata_cache_overflow.value.ui64 = wmsum_value(&dbuf_sums.metadata_cache_overflow); return (0); } void dbuf_init(void) { uint64_t hmsize, hsize = 1ULL << 16; dbuf_hash_table_t *h = &dbuf_hash_table; /* * The hash table is big enough to fill one eighth of physical memory * with an average block size of zfs_arc_average_blocksize (default 8K). * By default, the table will take up * totalmem * sizeof(void*) / 8K (1MB per GB with 8-byte pointers). */ while (hsize * zfs_arc_average_blocksize < arc_all_memory() / 8) hsize <<= 1; h->hash_table = NULL; while (h->hash_table == NULL) { h->hash_table_mask = hsize - 1; h->hash_table = vmem_zalloc(hsize * sizeof (void *), KM_SLEEP); if (h->hash_table == NULL) hsize >>= 1; ASSERT3U(hsize, >=, 1ULL << 10); } /* * The hash table buckets are protected by an array of mutexes where * each mutex is reponsible for protecting 128 buckets. A minimum * array size of 8192 is targeted to avoid contention. */ if (dbuf_mutex_cache_shift == 0) hmsize = MAX(hsize >> 7, 1ULL << 13); else hmsize = 1ULL << MIN(dbuf_mutex_cache_shift, 24); h->hash_mutexes = NULL; while (h->hash_mutexes == NULL) { h->hash_mutex_mask = hmsize - 1; h->hash_mutexes = vmem_zalloc(hmsize * sizeof (kmutex_t), KM_SLEEP); if (h->hash_mutexes == NULL) hmsize >>= 1; } dbuf_kmem_cache = kmem_cache_create("dmu_buf_impl_t", sizeof (dmu_buf_impl_t), 0, dbuf_cons, dbuf_dest, NULL, NULL, NULL, 0); dbuf_dirty_kmem_cache = kmem_cache_create("dbuf_dirty_record_t", sizeof (dbuf_dirty_record_t), 0, NULL, NULL, NULL, NULL, NULL, 0); for (int i = 0; i < hmsize; i++) mutex_init(&h->hash_mutexes[i], NULL, MUTEX_NOLOCKDEP, NULL); dbuf_stats_init(h); /* * All entries are queued via taskq_dispatch_ent(), so min/maxalloc * configuration is not required. */ dbu_evict_taskq = taskq_create("dbu_evict", 1, defclsyspri, 0, 0, 0); for (dbuf_cached_state_t dcs = 0; dcs < DB_CACHE_MAX; dcs++) { multilist_create(&dbuf_caches[dcs].cache, sizeof (dmu_buf_impl_t), offsetof(dmu_buf_impl_t, db_cache_link), dbuf_cache_multilist_index_func); zfs_refcount_create(&dbuf_caches[dcs].size); } dbuf_evict_thread_exit = B_FALSE; mutex_init(&dbuf_evict_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&dbuf_evict_cv, NULL, CV_DEFAULT, NULL); dbuf_cache_evict_thread = thread_create(NULL, 0, dbuf_evict_thread, NULL, 0, &p0, TS_RUN, minclsyspri); wmsum_init(&dbuf_sums.cache_count, 0); wmsum_init(&dbuf_sums.cache_total_evicts, 0); for (int i = 0; i < DN_MAX_LEVELS; i++) { wmsum_init(&dbuf_sums.cache_levels[i], 0); wmsum_init(&dbuf_sums.cache_levels_bytes[i], 0); } wmsum_init(&dbuf_sums.hash_hits, 0); wmsum_init(&dbuf_sums.hash_misses, 0); wmsum_init(&dbuf_sums.hash_collisions, 0); wmsum_init(&dbuf_sums.hash_elements, 0); wmsum_init(&dbuf_sums.hash_chains, 0); wmsum_init(&dbuf_sums.hash_insert_race, 0); wmsum_init(&dbuf_sums.metadata_cache_count, 0); wmsum_init(&dbuf_sums.metadata_cache_overflow, 0); dbuf_ksp = kstat_create("zfs", 0, "dbufstats", "misc", KSTAT_TYPE_NAMED, sizeof (dbuf_stats) / sizeof (kstat_named_t), KSTAT_FLAG_VIRTUAL); if (dbuf_ksp != NULL) { for (int i = 0; i < DN_MAX_LEVELS; i++) { snprintf(dbuf_stats.cache_levels[i].name, KSTAT_STRLEN, "cache_level_%d", i); dbuf_stats.cache_levels[i].data_type = KSTAT_DATA_UINT64; snprintf(dbuf_stats.cache_levels_bytes[i].name, KSTAT_STRLEN, "cache_level_%d_bytes", i); dbuf_stats.cache_levels_bytes[i].data_type = KSTAT_DATA_UINT64; } dbuf_ksp->ks_data = &dbuf_stats; dbuf_ksp->ks_update = dbuf_kstat_update; kstat_install(dbuf_ksp); } } void dbuf_fini(void) { dbuf_hash_table_t *h = &dbuf_hash_table; dbuf_stats_destroy(); for (int i = 0; i < (h->hash_mutex_mask + 1); i++) mutex_destroy(&h->hash_mutexes[i]); vmem_free(h->hash_table, (h->hash_table_mask + 1) * sizeof (void *)); vmem_free(h->hash_mutexes, (h->hash_mutex_mask + 1) * sizeof (kmutex_t)); kmem_cache_destroy(dbuf_kmem_cache); kmem_cache_destroy(dbuf_dirty_kmem_cache); taskq_destroy(dbu_evict_taskq); mutex_enter(&dbuf_evict_lock); dbuf_evict_thread_exit = B_TRUE; while (dbuf_evict_thread_exit) { cv_signal(&dbuf_evict_cv); cv_wait(&dbuf_evict_cv, &dbuf_evict_lock); } mutex_exit(&dbuf_evict_lock); mutex_destroy(&dbuf_evict_lock); cv_destroy(&dbuf_evict_cv); for (dbuf_cached_state_t dcs = 0; dcs < DB_CACHE_MAX; dcs++) { zfs_refcount_destroy(&dbuf_caches[dcs].size); multilist_destroy(&dbuf_caches[dcs].cache); } if (dbuf_ksp != NULL) { kstat_delete(dbuf_ksp); dbuf_ksp = NULL; } wmsum_fini(&dbuf_sums.cache_count); wmsum_fini(&dbuf_sums.cache_total_evicts); for (int i = 0; i < DN_MAX_LEVELS; i++) { wmsum_fini(&dbuf_sums.cache_levels[i]); wmsum_fini(&dbuf_sums.cache_levels_bytes[i]); } wmsum_fini(&dbuf_sums.hash_hits); wmsum_fini(&dbuf_sums.hash_misses); wmsum_fini(&dbuf_sums.hash_collisions); wmsum_fini(&dbuf_sums.hash_elements); wmsum_fini(&dbuf_sums.hash_chains); wmsum_fini(&dbuf_sums.hash_insert_race); wmsum_fini(&dbuf_sums.metadata_cache_count); wmsum_fini(&dbuf_sums.metadata_cache_overflow); } /* * Other stuff. */ #ifdef ZFS_DEBUG static void dbuf_verify(dmu_buf_impl_t *db) { dnode_t *dn; dbuf_dirty_record_t *dr; uint32_t txg_prev; ASSERT(MUTEX_HELD(&db->db_mtx)); if (!(zfs_flags & ZFS_DEBUG_DBUF_VERIFY)) return; ASSERT(db->db_objset != NULL); DB_DNODE_ENTER(db); dn = DB_DNODE(db); if (dn == NULL) { ASSERT(db->db_parent == NULL); ASSERT(db->db_blkptr == NULL); } else { ASSERT3U(db->db.db_object, ==, dn->dn_object); ASSERT3P(db->db_objset, ==, dn->dn_objset); ASSERT3U(db->db_level, <, dn->dn_nlevels); ASSERT(db->db_blkid == DMU_BONUS_BLKID || db->db_blkid == DMU_SPILL_BLKID || !avl_is_empty(&dn->dn_dbufs)); } if (db->db_blkid == DMU_BONUS_BLKID) { ASSERT(dn != NULL); ASSERT3U(db->db.db_size, >=, dn->dn_bonuslen); ASSERT3U(db->db.db_offset, ==, DMU_BONUS_BLKID); } else if (db->db_blkid == DMU_SPILL_BLKID) { ASSERT(dn != NULL); ASSERT0(db->db.db_offset); } else { ASSERT3U(db->db.db_offset, ==, db->db_blkid * db->db.db_size); } if ((dr = list_head(&db->db_dirty_records)) != NULL) { ASSERT(dr->dr_dbuf == db); txg_prev = dr->dr_txg; for (dr = list_next(&db->db_dirty_records, dr); dr != NULL; dr = list_next(&db->db_dirty_records, dr)) { ASSERT(dr->dr_dbuf == db); ASSERT(txg_prev > dr->dr_txg); txg_prev = dr->dr_txg; } } /* * We can't assert that db_size matches dn_datablksz because it * can be momentarily different when another thread is doing * dnode_set_blksz(). */ if (db->db_level == 0 && db->db.db_object == DMU_META_DNODE_OBJECT) { dr = db->db_data_pending; /* * It should only be modified in syncing context, so * make sure we only have one copy of the data. */ ASSERT(dr == NULL || dr->dt.dl.dr_data == db->db_buf); } /* verify db->db_blkptr */ if (db->db_blkptr) { if (db->db_parent == dn->dn_dbuf) { /* db is pointed to by the dnode */ /* ASSERT3U(db->db_blkid, <, dn->dn_nblkptr); */ if (DMU_OBJECT_IS_SPECIAL(db->db.db_object)) ASSERT(db->db_parent == NULL); else ASSERT(db->db_parent != NULL); if (db->db_blkid != DMU_SPILL_BLKID) ASSERT3P(db->db_blkptr, ==, &dn->dn_phys->dn_blkptr[db->db_blkid]); } else { /* db is pointed to by an indirect block */ int epb __maybe_unused = db->db_parent->db.db_size >> SPA_BLKPTRSHIFT; ASSERT3U(db->db_parent->db_level, ==, db->db_level+1); ASSERT3U(db->db_parent->db.db_object, ==, db->db.db_object); ASSERT3P(db->db_blkptr, ==, ((blkptr_t *)db->db_parent->db.db_data + db->db_blkid % epb)); } } if ((db->db_blkptr == NULL || BP_IS_HOLE(db->db_blkptr)) && (db->db_buf == NULL || db->db_buf->b_data) && db->db.db_data && db->db_blkid != DMU_BONUS_BLKID && db->db_state != DB_FILL && (dn == NULL || !dn->dn_free_txg)) { /* * If the blkptr isn't set but they have nonzero data, * it had better be dirty, otherwise we'll lose that * data when we evict this buffer. * * There is an exception to this rule for indirect blocks; in * this case, if the indirect block is a hole, we fill in a few * fields on each of the child blocks (importantly, birth time) * to prevent hole birth times from being lost when you * partially fill in a hole. */ if (db->db_dirtycnt == 0) { if (db->db_level == 0) { uint64_t *buf = db->db.db_data; int i; for (i = 0; i < db->db.db_size >> 3; i++) { ASSERT(buf[i] == 0); } } else { blkptr_t *bps = db->db.db_data; ASSERT3U(1 << DB_DNODE(db)->dn_indblkshift, ==, db->db.db_size); /* * We want to verify that all the blkptrs in the * indirect block are holes, but we may have * automatically set up a few fields for them. * We iterate through each blkptr and verify * they only have those fields set. */ for (int i = 0; i < db->db.db_size / sizeof (blkptr_t); i++) { blkptr_t *bp = &bps[i]; ASSERT(ZIO_CHECKSUM_IS_ZERO( &bp->blk_cksum)); ASSERT( DVA_IS_EMPTY(&bp->blk_dva[0]) && DVA_IS_EMPTY(&bp->blk_dva[1]) && DVA_IS_EMPTY(&bp->blk_dva[2])); ASSERT0(bp->blk_fill); ASSERT(!BP_IS_EMBEDDED(bp)); ASSERT(BP_IS_HOLE(bp)); ASSERT0(BP_GET_RAW_PHYSICAL_BIRTH(bp)); } } } } DB_DNODE_EXIT(db); } #endif static void dbuf_clear_data(dmu_buf_impl_t *db) { ASSERT(MUTEX_HELD(&db->db_mtx)); dbuf_evict_user(db); ASSERT3P(db->db_buf, ==, NULL); db->db.db_data = NULL; if (db->db_state != DB_NOFILL) { db->db_state = DB_UNCACHED; DTRACE_SET_STATE(db, "clear data"); } } static void dbuf_set_data(dmu_buf_impl_t *db, arc_buf_t *buf) { ASSERT(MUTEX_HELD(&db->db_mtx)); ASSERT(buf != NULL); db->db_buf = buf; ASSERT(buf->b_data != NULL); db->db.db_data = buf->b_data; } static arc_buf_t * dbuf_alloc_arcbuf(dmu_buf_impl_t *db) { spa_t *spa = db->db_objset->os_spa; return (arc_alloc_buf(spa, db, DBUF_GET_BUFC_TYPE(db), db->db.db_size)); } /* * Calculate which level n block references the data at the level 0 offset * provided. */ uint64_t dbuf_whichblock(const dnode_t *dn, const int64_t level, const uint64_t offset) { if (dn->dn_datablkshift != 0 && dn->dn_indblkshift != 0) { /* * The level n blkid is equal to the level 0 blkid divided by * the number of level 0s in a level n block. * * The level 0 blkid is offset >> datablkshift = * offset / 2^datablkshift. * * The number of level 0s in a level n is the number of block * pointers in an indirect block, raised to the power of level. * This is 2^(indblkshift - SPA_BLKPTRSHIFT)^level = * 2^(level*(indblkshift - SPA_BLKPTRSHIFT)). * * Thus, the level n blkid is: offset / * ((2^datablkshift)*(2^(level*(indblkshift-SPA_BLKPTRSHIFT)))) * = offset / 2^(datablkshift + level * * (indblkshift - SPA_BLKPTRSHIFT)) * = offset >> (datablkshift + level * * (indblkshift - SPA_BLKPTRSHIFT)) */ const unsigned exp = dn->dn_datablkshift + level * (dn->dn_indblkshift - SPA_BLKPTRSHIFT); if (exp >= 8 * sizeof (offset)) { /* This only happens on the highest indirection level */ ASSERT3U(level, ==, dn->dn_nlevels - 1); return (0); } ASSERT3U(exp, <, 8 * sizeof (offset)); return (offset >> exp); } else { ASSERT3U(offset, <, dn->dn_datablksz); return (0); } } /* * This function is used to lock the parent of the provided dbuf. This should be * used when modifying or reading db_blkptr. */ db_lock_type_t dmu_buf_lock_parent(dmu_buf_impl_t *db, krw_t rw, const void *tag) { enum db_lock_type ret = DLT_NONE; if (db->db_parent != NULL) { rw_enter(&db->db_parent->db_rwlock, rw); ret = DLT_PARENT; } else if (dmu_objset_ds(db->db_objset) != NULL) { rrw_enter(&dmu_objset_ds(db->db_objset)->ds_bp_rwlock, rw, tag); ret = DLT_OBJSET; } /* * We only return a DLT_NONE lock when it's the top-most indirect block * of the meta-dnode of the MOS. */ return (ret); } /* * We need to pass the lock type in because it's possible that the block will * move from being the topmost indirect block in a dnode (and thus, have no * parent) to not the top-most via an indirection increase. This would cause a * panic if we didn't pass the lock type in. */ void dmu_buf_unlock_parent(dmu_buf_impl_t *db, db_lock_type_t type, const void *tag) { if (type == DLT_PARENT) rw_exit(&db->db_parent->db_rwlock); else if (type == DLT_OBJSET) rrw_exit(&dmu_objset_ds(db->db_objset)->ds_bp_rwlock, tag); } static void dbuf_read_done(zio_t *zio, const zbookmark_phys_t *zb, const blkptr_t *bp, arc_buf_t *buf, void *vdb) { (void) zb, (void) bp; dmu_buf_impl_t *db = vdb; mutex_enter(&db->db_mtx); ASSERT3U(db->db_state, ==, DB_READ); /* * All reads are synchronous, so we must have a hold on the dbuf */ ASSERT(zfs_refcount_count(&db->db_holds) > 0); ASSERT(db->db_buf == NULL); ASSERT(db->db.db_data == NULL); if (buf == NULL) { /* i/o error */ ASSERT(zio == NULL || zio->io_error != 0); ASSERT(db->db_blkid != DMU_BONUS_BLKID); ASSERT3P(db->db_buf, ==, NULL); db->db_state = DB_UNCACHED; DTRACE_SET_STATE(db, "i/o error"); } else if (db->db_level == 0 && db->db_freed_in_flight) { /* freed in flight */ ASSERT(zio == NULL || zio->io_error == 0); arc_release(buf, db); memset(buf->b_data, 0, db->db.db_size); arc_buf_freeze(buf); db->db_freed_in_flight = FALSE; dbuf_set_data(db, buf); db->db_state = DB_CACHED; DTRACE_SET_STATE(db, "freed in flight"); } else { /* success */ ASSERT(zio == NULL || zio->io_error == 0); dbuf_set_data(db, buf); db->db_state = DB_CACHED; DTRACE_SET_STATE(db, "successful read"); } cv_broadcast(&db->db_changed); dbuf_rele_and_unlock(db, NULL, B_FALSE); } /* * Shortcut for performing reads on bonus dbufs. Returns * an error if we fail to verify the dnode associated with * a decrypted block. Otherwise success. */ static int dbuf_read_bonus(dmu_buf_impl_t *db, dnode_t *dn) { void* db_data; int bonuslen, max_bonuslen; bonuslen = MIN(dn->dn_bonuslen, dn->dn_phys->dn_bonuslen); max_bonuslen = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots); ASSERT(MUTEX_HELD(&db->db_mtx)); ASSERT(DB_DNODE_HELD(db)); ASSERT3U(bonuslen, <=, db->db.db_size); db_data = kmem_alloc(max_bonuslen, KM_SLEEP); arc_space_consume(max_bonuslen, ARC_SPACE_BONUS); if (bonuslen < max_bonuslen) memset(db_data, 0, max_bonuslen); if (bonuslen) memcpy(db_data, DN_BONUS(dn->dn_phys), bonuslen); db->db.db_data = db_data; db->db_state = DB_CACHED; DTRACE_SET_STATE(db, "bonus buffer filled"); return (0); } static void dbuf_handle_indirect_hole(void *data, dnode_t *dn, blkptr_t *dbbp) { blkptr_t *bps = data; uint32_t indbs = 1ULL << dn->dn_indblkshift; int n_bps = indbs >> SPA_BLKPTRSHIFT; for (int i = 0; i < n_bps; i++) { blkptr_t *bp = &bps[i]; ASSERT3U(BP_GET_LSIZE(dbbp), ==, indbs); BP_SET_LSIZE(bp, BP_GET_LEVEL(dbbp) == 1 ? dn->dn_datablksz : BP_GET_LSIZE(dbbp)); BP_SET_TYPE(bp, BP_GET_TYPE(dbbp)); BP_SET_LEVEL(bp, BP_GET_LEVEL(dbbp) - 1); BP_SET_BIRTH(bp, BP_GET_LOGICAL_BIRTH(dbbp), 0); } } /* * Handle reads on dbufs that are holes, if necessary. This function * requires that the dbuf's mutex is held. Returns success (0) if action * was taken, ENOENT if no action was taken. */ static int dbuf_read_hole(dmu_buf_impl_t *db, dnode_t *dn, blkptr_t *bp) { ASSERT(MUTEX_HELD(&db->db_mtx)); arc_buf_t *db_data; int is_hole = bp == NULL || BP_IS_HOLE(bp); /* * For level 0 blocks only, if the above check fails: * Recheck BP_IS_HOLE() after dnode_block_freed() in case dnode_sync() * processes the delete record and clears the bp while we are waiting * for the dn_mtx (resulting in a "no" from block_freed). */ if (!is_hole && db->db_level == 0) is_hole = dnode_block_freed(dn, db->db_blkid) || BP_IS_HOLE(bp); if (is_hole) { db_data = dbuf_alloc_arcbuf(db); memset(db_data->b_data, 0, db->db.db_size); if (bp != NULL && db->db_level > 0 && BP_IS_HOLE(bp) && BP_GET_LOGICAL_BIRTH(bp) != 0) { dbuf_handle_indirect_hole(db_data->b_data, dn, bp); } dbuf_set_data(db, db_data); db->db_state = DB_CACHED; DTRACE_SET_STATE(db, "hole read satisfied"); return (0); } return (ENOENT); } /* * This function ensures that, when doing a decrypting read of a block, * we make sure we have decrypted the dnode associated with it. We must do * this so that we ensure we are fully authenticating the checksum-of-MACs * tree from the root of the objset down to this block. Indirect blocks are * always verified against their secure checksum-of-MACs assuming that the * dnode containing them is correct. Now that we are doing a decrypting read, * we can be sure that the key is loaded and verify that assumption. This is * especially important considering that we always read encrypted dnode * blocks as raw data (without verifying their MACs) to start, and * decrypt / authenticate them when we need to read an encrypted bonus buffer. */ static int dbuf_read_verify_dnode_crypt(dmu_buf_impl_t *db, dnode_t *dn, dmu_flags_t flags) { objset_t *os = db->db_objset; dmu_buf_impl_t *dndb; arc_buf_t *dnbuf; zbookmark_phys_t zb; int err; if ((flags & DMU_READ_NO_DECRYPT) != 0 || !os->os_encrypted || os->os_raw_receive || (dndb = dn->dn_dbuf) == NULL) return (0); dnbuf = dndb->db_buf; if (!arc_is_encrypted(dnbuf)) return (0); mutex_enter(&dndb->db_mtx); /* * Since dnode buffer is modified by sync process, there can be only * one copy of it. It means we can not modify (decrypt) it while it * is being written. I don't see how this may happen now, since * encrypted dnode writes by receive should be completed before any * plain-text reads due to txg wait, but better be safe than sorry. */ while (1) { if (!arc_is_encrypted(dnbuf)) { mutex_exit(&dndb->db_mtx); return (0); } dbuf_dirty_record_t *dr = dndb->db_data_pending; if (dr == NULL || dr->dt.dl.dr_data != dnbuf) break; cv_wait(&dndb->db_changed, &dndb->db_mtx); }; SET_BOOKMARK(&zb, dmu_objset_id(os), DMU_META_DNODE_OBJECT, 0, dndb->db_blkid); err = arc_untransform(dnbuf, os->os_spa, &zb, B_TRUE); /* * An error code of EACCES tells us that the key is still not * available. This is ok if we are only reading authenticated * (and therefore non-encrypted) blocks. */ if (err == EACCES && ((db->db_blkid != DMU_BONUS_BLKID && !DMU_OT_IS_ENCRYPTED(dn->dn_type)) || (db->db_blkid == DMU_BONUS_BLKID && !DMU_OT_IS_ENCRYPTED(dn->dn_bonustype)))) err = 0; mutex_exit(&dndb->db_mtx); return (err); } /* * Drops db_mtx and the parent lock specified by dblt and tag before * returning. */ static int dbuf_read_impl(dmu_buf_impl_t *db, dnode_t *dn, zio_t *zio, dmu_flags_t flags, db_lock_type_t dblt, blkptr_t *bp, const void *tag) { zbookmark_phys_t zb; uint32_t aflags = ARC_FLAG_NOWAIT; int err, zio_flags; ASSERT(!zfs_refcount_is_zero(&db->db_holds)); ASSERT(MUTEX_HELD(&db->db_mtx)); ASSERT(db->db_state == DB_UNCACHED || db->db_state == DB_NOFILL); ASSERT(db->db_buf == NULL); ASSERT(db->db_parent == NULL || RW_LOCK_HELD(&db->db_parent->db_rwlock)); if (db->db_blkid == DMU_BONUS_BLKID) { err = dbuf_read_bonus(db, dn); goto early_unlock; } err = dbuf_read_hole(db, dn, bp); if (err == 0) goto early_unlock; ASSERT(bp != NULL); /* * Any attempt to read a redacted block should result in an error. This * will never happen under normal conditions, but can be useful for * debugging purposes. */ if (BP_IS_REDACTED(bp)) { ASSERT(dsl_dataset_feature_is_active( db->db_objset->os_dsl_dataset, SPA_FEATURE_REDACTED_DATASETS)); err = SET_ERROR(EIO); goto early_unlock; } SET_BOOKMARK(&zb, dmu_objset_id(db->db_objset), db->db.db_object, db->db_level, db->db_blkid); /* * All bps of an encrypted os should have the encryption bit set. * If this is not true it indicates tampering and we report an error. */ if (db->db_objset->os_encrypted && !BP_USES_CRYPT(bp)) { spa_log_error(db->db_objset->os_spa, &zb, BP_GET_PHYSICAL_BIRTH(bp)); err = SET_ERROR(EIO); goto early_unlock; } db->db_state = DB_READ; DTRACE_SET_STATE(db, "read issued"); mutex_exit(&db->db_mtx); if (!DBUF_IS_CACHEABLE(db)) aflags |= ARC_FLAG_UNCACHED; else if (dbuf_is_l2cacheable(db, bp)) aflags |= ARC_FLAG_L2CACHE; dbuf_add_ref(db, NULL); zio_flags = (flags & DB_RF_CANFAIL) ? ZIO_FLAG_CANFAIL : ZIO_FLAG_MUSTSUCCEED; if ((flags & DMU_READ_NO_DECRYPT) && BP_IS_PROTECTED(bp)) zio_flags |= ZIO_FLAG_RAW; /* * The zio layer will copy the provided blkptr later, but we need to * do this now so that we can release the parent's rwlock. We have to * do that now so that if dbuf_read_done is called synchronously (on * an l1 cache hit) we don't acquire the db_mtx while holding the * parent's rwlock, which would be a lock ordering violation. */ blkptr_t copy = *bp; dmu_buf_unlock_parent(db, dblt, tag); return (arc_read(zio, db->db_objset->os_spa, ©, dbuf_read_done, db, ZIO_PRIORITY_SYNC_READ, zio_flags, &aflags, &zb)); early_unlock: mutex_exit(&db->db_mtx); dmu_buf_unlock_parent(db, dblt, tag); return (err); } /* * This is our just-in-time copy function. It makes a copy of buffers that * have been modified in a previous transaction group before we access them in * the current active group. * * This function is used in three places: when we are dirtying a buffer for the * first time in a txg, when we are freeing a range in a dnode that includes * this buffer, and when we are accessing a buffer which was received compressed * and later referenced in a WRITE_BYREF record. * * Note that when we are called from dbuf_free_range() we do not put a hold on * the buffer, we just traverse the active dbuf list for the dnode. */ static void dbuf_fix_old_data(dmu_buf_impl_t *db, uint64_t txg) { dbuf_dirty_record_t *dr = list_head(&db->db_dirty_records); ASSERT(MUTEX_HELD(&db->db_mtx)); ASSERT(db->db.db_data != NULL); ASSERT(db->db_level == 0); ASSERT(db->db.db_object != DMU_META_DNODE_OBJECT); if (dr == NULL || (dr->dt.dl.dr_data != ((db->db_blkid == DMU_BONUS_BLKID) ? db->db.db_data : db->db_buf))) return; /* * If the last dirty record for this dbuf has not yet synced * and its referencing the dbuf data, either: * reset the reference to point to a new copy, * or (if there a no active holders) * just null out the current db_data pointer. */ ASSERT3U(dr->dr_txg, >=, txg - 2); if (db->db_blkid == DMU_BONUS_BLKID) { dnode_t *dn = DB_DNODE(db); int bonuslen = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots); dr->dt.dl.dr_data = kmem_alloc(bonuslen, KM_SLEEP); arc_space_consume(bonuslen, ARC_SPACE_BONUS); memcpy(dr->dt.dl.dr_data, db->db.db_data, bonuslen); } else if (zfs_refcount_count(&db->db_holds) > db->db_dirtycnt) { dnode_t *dn = DB_DNODE(db); int size = arc_buf_size(db->db_buf); arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db); spa_t *spa = db->db_objset->os_spa; enum zio_compress compress_type = arc_get_compression(db->db_buf); uint8_t complevel = arc_get_complevel(db->db_buf); if (arc_is_encrypted(db->db_buf)) { boolean_t byteorder; uint8_t salt[ZIO_DATA_SALT_LEN]; uint8_t iv[ZIO_DATA_IV_LEN]; uint8_t mac[ZIO_DATA_MAC_LEN]; arc_get_raw_params(db->db_buf, &byteorder, salt, iv, mac); dr->dt.dl.dr_data = arc_alloc_raw_buf(spa, db, dmu_objset_id(dn->dn_objset), byteorder, salt, iv, mac, dn->dn_type, size, arc_buf_lsize(db->db_buf), compress_type, complevel); } else if (compress_type != ZIO_COMPRESS_OFF) { ASSERT3U(type, ==, ARC_BUFC_DATA); dr->dt.dl.dr_data = arc_alloc_compressed_buf(spa, db, size, arc_buf_lsize(db->db_buf), compress_type, complevel); } else { dr->dt.dl.dr_data = arc_alloc_buf(spa, db, type, size); } memcpy(dr->dt.dl.dr_data->b_data, db->db.db_data, size); } else { db->db_buf = NULL; dbuf_clear_data(db); } } int dbuf_read(dmu_buf_impl_t *db, zio_t *pio, dmu_flags_t flags) { dnode_t *dn; boolean_t miss = B_TRUE, need_wait = B_FALSE, prefetch; int err; ASSERT(!zfs_refcount_is_zero(&db->db_holds)); DB_DNODE_ENTER(db); dn = DB_DNODE(db); /* * Ensure that this block's dnode has been decrypted if the caller * has requested decrypted data. */ err = dbuf_read_verify_dnode_crypt(db, dn, flags); if (err != 0) goto done; prefetch = db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID && (flags & DMU_READ_NO_PREFETCH) == 0; mutex_enter(&db->db_mtx); if (!(flags & (DMU_UNCACHEDIO | DMU_KEEP_CACHING))) db->db_pending_evict = B_FALSE; if (flags & DMU_PARTIAL_FIRST) db->db_partial_read = B_TRUE; else if (!(flags & (DMU_PARTIAL_MORE | DMU_KEEP_CACHING))) db->db_partial_read = B_FALSE; miss = (db->db_state != DB_CACHED); if (db->db_state == DB_READ || db->db_state == DB_FILL) { /* * Another reader came in while the dbuf was in flight between * UNCACHED and CACHED. Either a writer will finish filling * the buffer, sending the dbuf to CACHED, or the first reader's * request will reach the read_done callback and send the dbuf * to CACHED. Otherwise, a failure occurred and the dbuf will * be sent to UNCACHED. */ if (flags & DB_RF_NEVERWAIT) { mutex_exit(&db->db_mtx); DB_DNODE_EXIT(db); goto done; } do { ASSERT(db->db_state == DB_READ || (flags & DB_RF_HAVESTRUCT) == 0); DTRACE_PROBE2(blocked__read, dmu_buf_impl_t *, db, zio_t *, pio); cv_wait(&db->db_changed, &db->db_mtx); } while (db->db_state == DB_READ || db->db_state == DB_FILL); if (db->db_state == DB_UNCACHED) { err = SET_ERROR(EIO); mutex_exit(&db->db_mtx); DB_DNODE_EXIT(db); goto done; } } if (db->db_state == DB_CACHED) { /* * If the arc buf is compressed or encrypted and the caller * requested uncompressed data, we need to untransform it * before returning. We also call arc_untransform() on any * unauthenticated blocks, which will verify their MAC if * the key is now available. */ if ((flags & DMU_READ_NO_DECRYPT) == 0 && db->db_buf != NULL && (arc_is_encrypted(db->db_buf) || arc_is_unauthenticated(db->db_buf) || arc_get_compression(db->db_buf) != ZIO_COMPRESS_OFF)) { spa_t *spa = dn->dn_objset->os_spa; zbookmark_phys_t zb; SET_BOOKMARK(&zb, dmu_objset_id(db->db_objset), db->db.db_object, db->db_level, db->db_blkid); dbuf_fix_old_data(db, spa_syncing_txg(spa)); err = arc_untransform(db->db_buf, spa, &zb, B_FALSE); dbuf_set_data(db, db->db_buf); } mutex_exit(&db->db_mtx); } else { ASSERT(db->db_state == DB_UNCACHED || db->db_state == DB_NOFILL); db_lock_type_t dblt = dmu_buf_lock_parent(db, RW_READER, FTAG); blkptr_t *bp; /* * If a block clone or Direct I/O write has occurred we will * get the dirty records overridden BP so we get the most * recent data. */ err = dmu_buf_get_bp_from_dbuf(db, &bp); if (!err) { if (pio == NULL && (db->db_state == DB_NOFILL || (bp != NULL && !BP_IS_HOLE(bp)))) { spa_t *spa = dn->dn_objset->os_spa; pio = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL); need_wait = B_TRUE; } err = dbuf_read_impl(db, dn, pio, flags, dblt, bp, FTAG); } else { mutex_exit(&db->db_mtx); dmu_buf_unlock_parent(db, dblt, FTAG); } /* dbuf_read_impl drops db_mtx and parent's rwlock. */ miss = (db->db_state != DB_CACHED); } if (err == 0 && prefetch) { dmu_zfetch(&dn->dn_zfetch, db->db_blkid, 1, B_TRUE, miss, flags & DB_RF_HAVESTRUCT, (flags & DMU_UNCACHEDIO) || db->db_pending_evict); } DB_DNODE_EXIT(db); /* * If we created a zio we must execute it to avoid leaking it, even if * it isn't attached to any work due to an error in dbuf_read_impl(). */ if (need_wait) { if (err == 0) err = zio_wait(pio); else (void) zio_wait(pio); pio = NULL; } done: if (miss) DBUF_STAT_BUMP(hash_misses); else DBUF_STAT_BUMP(hash_hits); if (pio && err != 0) { zio_t *zio = zio_null(pio, pio->io_spa, NULL, NULL, NULL, ZIO_FLAG_CANFAIL); zio->io_error = err; zio_nowait(zio); } return (err); } static void dbuf_noread(dmu_buf_impl_t *db, dmu_flags_t flags) { ASSERT(!zfs_refcount_is_zero(&db->db_holds)); ASSERT(db->db_blkid != DMU_BONUS_BLKID); mutex_enter(&db->db_mtx); if (!(flags & (DMU_UNCACHEDIO | DMU_KEEP_CACHING))) db->db_pending_evict = B_FALSE; db->db_partial_read = B_FALSE; while (db->db_state == DB_READ || db->db_state == DB_FILL) cv_wait(&db->db_changed, &db->db_mtx); if (db->db_state == DB_UNCACHED) { ASSERT(db->db_buf == NULL); ASSERT(db->db.db_data == NULL); dbuf_set_data(db, dbuf_alloc_arcbuf(db)); db->db_state = DB_FILL; DTRACE_SET_STATE(db, "assigning filled buffer"); } else if (db->db_state == DB_NOFILL) { dbuf_clear_data(db); } else { ASSERT3U(db->db_state, ==, DB_CACHED); } mutex_exit(&db->db_mtx); } void dbuf_unoverride(dbuf_dirty_record_t *dr) { dmu_buf_impl_t *db = dr->dr_dbuf; blkptr_t *bp = &dr->dt.dl.dr_overridden_by; uint64_t txg = dr->dr_txg; ASSERT(MUTEX_HELD(&db->db_mtx)); /* * This assert is valid because dmu_sync() expects to be called by * a zilog's get_data while holding a range lock. This call only * comes from dbuf_dirty() callers who must also hold a range lock. */ ASSERT(dr->dt.dl.dr_override_state != DR_IN_DMU_SYNC); ASSERT(db->db_level == 0); if (db->db_blkid == DMU_BONUS_BLKID || dr->dt.dl.dr_override_state == DR_NOT_OVERRIDDEN) return; ASSERT(db->db_data_pending != dr); /* free this block */ if (!BP_IS_HOLE(bp) && !dr->dt.dl.dr_nopwrite) zio_free(db->db_objset->os_spa, txg, bp); if (dr->dt.dl.dr_brtwrite || dr->dt.dl.dr_diowrite) { ASSERT0P(dr->dt.dl.dr_data); dr->dt.dl.dr_data = db->db_buf; } dr->dt.dl.dr_override_state = DR_NOT_OVERRIDDEN; dr->dt.dl.dr_nopwrite = B_FALSE; dr->dt.dl.dr_brtwrite = B_FALSE; dr->dt.dl.dr_diowrite = B_FALSE; dr->dt.dl.dr_has_raw_params = B_FALSE; /* * In the event that Direct I/O was used, we do not * need to release the buffer from the ARC. * * Release the already-written buffer, so we leave it in * a consistent dirty state. Note that all callers are * modifying the buffer, so they will immediately do * another (redundant) arc_release(). Therefore, leave * the buf thawed to save the effort of freezing & * immediately re-thawing it. */ if (dr->dt.dl.dr_data) arc_release(dr->dt.dl.dr_data, db); } /* * Evict (if its unreferenced) or clear (if its referenced) any level-0 * data blocks in the free range, so that any future readers will find * empty blocks. */ void dbuf_free_range(dnode_t *dn, uint64_t start_blkid, uint64_t end_blkid, dmu_tx_t *tx) { dmu_buf_impl_t *db_search; dmu_buf_impl_t *db, *db_next; uint64_t txg = tx->tx_txg; avl_index_t where; dbuf_dirty_record_t *dr; if (end_blkid > dn->dn_maxblkid && !(start_blkid == DMU_SPILL_BLKID || end_blkid == DMU_SPILL_BLKID)) end_blkid = dn->dn_maxblkid; dprintf_dnode(dn, "start=%llu end=%llu\n", (u_longlong_t)start_blkid, (u_longlong_t)end_blkid); db_search = kmem_alloc(sizeof (dmu_buf_impl_t), KM_SLEEP); db_search->db_level = 0; db_search->db_blkid = start_blkid; db_search->db_state = DB_SEARCH; mutex_enter(&dn->dn_dbufs_mtx); db = avl_find(&dn->dn_dbufs, db_search, &where); ASSERT3P(db, ==, NULL); db = avl_nearest(&dn->dn_dbufs, where, AVL_AFTER); for (; db != NULL; db = db_next) { db_next = AVL_NEXT(&dn->dn_dbufs, db); ASSERT(db->db_blkid != DMU_BONUS_BLKID); if (db->db_level != 0 || db->db_blkid > end_blkid) { break; } ASSERT3U(db->db_blkid, >=, start_blkid); /* found a level 0 buffer in the range */ mutex_enter(&db->db_mtx); if (dbuf_undirty(db, tx)) { /* mutex has been dropped and dbuf destroyed */ continue; } if (db->db_state == DB_UNCACHED || db->db_state == DB_NOFILL || db->db_state == DB_EVICTING) { ASSERT(db->db.db_data == NULL); mutex_exit(&db->db_mtx); continue; } if (db->db_state == DB_READ || db->db_state == DB_FILL) { /* will be handled in dbuf_read_done or dbuf_rele */ db->db_freed_in_flight = TRUE; mutex_exit(&db->db_mtx); continue; } if (zfs_refcount_count(&db->db_holds) == 0) { ASSERT(db->db_buf); dbuf_destroy(db); continue; } /* The dbuf is referenced */ dr = list_head(&db->db_dirty_records); if (dr != NULL) { if (dr->dr_txg == txg) { /* * This buffer is "in-use", re-adjust the file * size to reflect that this buffer may * contain new data when we sync. */ if (db->db_blkid != DMU_SPILL_BLKID && db->db_blkid > dn->dn_maxblkid) dn->dn_maxblkid = db->db_blkid; dbuf_unoverride(dr); } else { /* * This dbuf is not dirty in the open context. * Either uncache it (if its not referenced in * the open context) or reset its contents to * empty. */ dbuf_fix_old_data(db, txg); } } /* clear the contents if its cached */ if (db->db_state == DB_CACHED) { ASSERT(db->db.db_data != NULL); arc_release(db->db_buf, db); rw_enter(&db->db_rwlock, RW_WRITER); memset(db->db.db_data, 0, db->db.db_size); rw_exit(&db->db_rwlock); arc_buf_freeze(db->db_buf); } mutex_exit(&db->db_mtx); } mutex_exit(&dn->dn_dbufs_mtx); kmem_free(db_search, sizeof (dmu_buf_impl_t)); } void dbuf_new_size(dmu_buf_impl_t *db, int size, dmu_tx_t *tx) { arc_buf_t *buf, *old_buf; dbuf_dirty_record_t *dr; int osize = db->db.db_size; arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db); dnode_t *dn; ASSERT(db->db_blkid != DMU_BONUS_BLKID); DB_DNODE_ENTER(db); dn = DB_DNODE(db); /* * XXX we should be doing a dbuf_read, checking the return * value and returning that up to our callers */ dmu_buf_will_dirty(&db->db, tx); VERIFY3P(db->db_buf, !=, NULL); /* create the data buffer for the new block */ buf = arc_alloc_buf(dn->dn_objset->os_spa, db, type, size); /* copy old block data to the new block */ old_buf = db->db_buf; memcpy(buf->b_data, old_buf->b_data, MIN(osize, size)); /* zero the remainder */ if (size > osize) memset((uint8_t *)buf->b_data + osize, 0, size - osize); mutex_enter(&db->db_mtx); dbuf_set_data(db, buf); arc_buf_destroy(old_buf, db); db->db.db_size = size; dr = list_head(&db->db_dirty_records); /* dirty record added by dmu_buf_will_dirty() */ VERIFY(dr != NULL); if (db->db_level == 0) dr->dt.dl.dr_data = buf; ASSERT3U(dr->dr_txg, ==, tx->tx_txg); ASSERT3U(dr->dr_accounted, ==, osize); dr->dr_accounted = size; mutex_exit(&db->db_mtx); dmu_objset_willuse_space(dn->dn_objset, size - osize, tx); DB_DNODE_EXIT(db); } void dbuf_release_bp(dmu_buf_impl_t *db) { objset_t *os __maybe_unused = db->db_objset; ASSERT(dsl_pool_sync_context(dmu_objset_pool(os))); ASSERT(arc_released(os->os_phys_buf) || list_link_active(&os->os_dsl_dataset->ds_synced_link)); ASSERT(db->db_parent == NULL || arc_released(db->db_parent->db_buf)); (void) arc_release(db->db_buf, db); } /* * We already have a dirty record for this TXG, and we are being * dirtied again. */ static void dbuf_redirty(dbuf_dirty_record_t *dr) { dmu_buf_impl_t *db = dr->dr_dbuf; ASSERT(MUTEX_HELD(&db->db_mtx)); if (db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID) { /* * If this buffer has already been written out, * we now need to reset its state. */ dbuf_unoverride(dr); if (db->db.db_object != DMU_META_DNODE_OBJECT && db->db_state != DB_NOFILL) { /* Already released on initial dirty, so just thaw. */ ASSERT(arc_released(db->db_buf)); arc_buf_thaw(db->db_buf); } + + /* + * Clear the rewrite flag since this is now a logical + * modification. + */ + dr->dt.dl.dr_rewrite = B_FALSE; } } dbuf_dirty_record_t * dbuf_dirty_lightweight(dnode_t *dn, uint64_t blkid, dmu_tx_t *tx) { rw_enter(&dn->dn_struct_rwlock, RW_READER); IMPLY(dn->dn_objset->os_raw_receive, dn->dn_maxblkid >= blkid); dnode_new_blkid(dn, blkid, tx, B_TRUE, B_FALSE); ASSERT(dn->dn_maxblkid >= blkid); dbuf_dirty_record_t *dr = kmem_zalloc(sizeof (*dr), KM_SLEEP); list_link_init(&dr->dr_dirty_node); list_link_init(&dr->dr_dbuf_node); dr->dr_dnode = dn; dr->dr_txg = tx->tx_txg; dr->dt.dll.dr_blkid = blkid; dr->dr_accounted = dn->dn_datablksz; /* * There should not be any dbuf for the block that we're dirtying. * Otherwise the buffer contents could be inconsistent between the * dbuf and the lightweight dirty record. */ ASSERT3P(NULL, ==, dbuf_find(dn->dn_objset, dn->dn_object, 0, blkid, NULL)); mutex_enter(&dn->dn_mtx); int txgoff = tx->tx_txg & TXG_MASK; if (dn->dn_free_ranges[txgoff] != NULL) { zfs_range_tree_clear(dn->dn_free_ranges[txgoff], blkid, 1); } if (dn->dn_nlevels == 1) { ASSERT3U(blkid, <, dn->dn_nblkptr); list_insert_tail(&dn->dn_dirty_records[txgoff], dr); mutex_exit(&dn->dn_mtx); rw_exit(&dn->dn_struct_rwlock); dnode_setdirty(dn, tx); } else { mutex_exit(&dn->dn_mtx); int epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT; dmu_buf_impl_t *parent_db = dbuf_hold_level(dn, 1, blkid >> epbs, FTAG); rw_exit(&dn->dn_struct_rwlock); if (parent_db == NULL) { kmem_free(dr, sizeof (*dr)); return (NULL); } int err = dbuf_read(parent_db, NULL, DB_RF_CANFAIL | DMU_READ_NO_PREFETCH); if (err != 0) { dbuf_rele(parent_db, FTAG); kmem_free(dr, sizeof (*dr)); return (NULL); } dbuf_dirty_record_t *parent_dr = dbuf_dirty(parent_db, tx); dbuf_rele(parent_db, FTAG); mutex_enter(&parent_dr->dt.di.dr_mtx); ASSERT3U(parent_dr->dr_txg, ==, tx->tx_txg); list_insert_tail(&parent_dr->dt.di.dr_children, dr); mutex_exit(&parent_dr->dt.di.dr_mtx); dr->dr_parent = parent_dr; } dmu_objset_willuse_space(dn->dn_objset, dr->dr_accounted, tx); return (dr); } dbuf_dirty_record_t * dbuf_dirty(dmu_buf_impl_t *db, dmu_tx_t *tx) { dnode_t *dn; objset_t *os; dbuf_dirty_record_t *dr, *dr_next, *dr_head; int txgoff = tx->tx_txg & TXG_MASK; boolean_t drop_struct_rwlock = B_FALSE; ASSERT(tx->tx_txg != 0); ASSERT(!zfs_refcount_is_zero(&db->db_holds)); DMU_TX_DIRTY_BUF(tx, db); DB_DNODE_ENTER(db); dn = DB_DNODE(db); /* * Shouldn't dirty a regular buffer in syncing context. Private * objects may be dirtied in syncing context, but only if they * were already pre-dirtied in open context. */ #ifdef ZFS_DEBUG if (dn->dn_objset->os_dsl_dataset != NULL) { rrw_enter(&dn->dn_objset->os_dsl_dataset->ds_bp_rwlock, RW_READER, FTAG); } ASSERT(!dmu_tx_is_syncing(tx) || BP_IS_HOLE(dn->dn_objset->os_rootbp) || DMU_OBJECT_IS_SPECIAL(dn->dn_object) || dn->dn_objset->os_dsl_dataset == NULL); if (dn->dn_objset->os_dsl_dataset != NULL) rrw_exit(&dn->dn_objset->os_dsl_dataset->ds_bp_rwlock, FTAG); #endif /* * We make this assert for private objects as well, but after we * check if we're already dirty. They are allowed to re-dirty * in syncing context. */ ASSERT(dn->dn_object == DMU_META_DNODE_OBJECT || dn->dn_dirtyctx == DN_UNDIRTIED || dn->dn_dirtyctx == (dmu_tx_is_syncing(tx) ? DN_DIRTY_SYNC : DN_DIRTY_OPEN)); mutex_enter(&db->db_mtx); /* * XXX make this true for indirects too? The problem is that * transactions created with dmu_tx_create_assigned() from * syncing context don't bother holding ahead. */ ASSERT(db->db_level != 0 || db->db_state == DB_CACHED || db->db_state == DB_FILL || db->db_state == DB_NOFILL); mutex_enter(&dn->dn_mtx); dnode_set_dirtyctx(dn, tx, db); if (tx->tx_txg > dn->dn_dirty_txg) dn->dn_dirty_txg = tx->tx_txg; mutex_exit(&dn->dn_mtx); if (db->db_blkid == DMU_SPILL_BLKID) dn->dn_have_spill = B_TRUE; /* * If this buffer is already dirty, we're done. */ dr_head = list_head(&db->db_dirty_records); ASSERT(dr_head == NULL || dr_head->dr_txg <= tx->tx_txg || db->db.db_object == DMU_META_DNODE_OBJECT); dr_next = dbuf_find_dirty_lte(db, tx->tx_txg); if (dr_next && dr_next->dr_txg == tx->tx_txg) { DB_DNODE_EXIT(db); dbuf_redirty(dr_next); mutex_exit(&db->db_mtx); return (dr_next); } /* * Only valid if not already dirty. */ ASSERT(dn->dn_object == 0 || dn->dn_dirtyctx == DN_UNDIRTIED || dn->dn_dirtyctx == (dmu_tx_is_syncing(tx) ? DN_DIRTY_SYNC : DN_DIRTY_OPEN)); ASSERT3U(dn->dn_nlevels, >, db->db_level); /* * We should only be dirtying in syncing context if it's the * mos or we're initializing the os or it's a special object. * However, we are allowed to dirty in syncing context provided * we already dirtied it in open context. Hence we must make * this assertion only if we're not already dirty. */ os = dn->dn_objset; VERIFY3U(tx->tx_txg, <=, spa_final_dirty_txg(os->os_spa)); #ifdef ZFS_DEBUG if (dn->dn_objset->os_dsl_dataset != NULL) rrw_enter(&os->os_dsl_dataset->ds_bp_rwlock, RW_READER, FTAG); ASSERT(!dmu_tx_is_syncing(tx) || DMU_OBJECT_IS_SPECIAL(dn->dn_object) || os->os_dsl_dataset == NULL || BP_IS_HOLE(os->os_rootbp)); if (dn->dn_objset->os_dsl_dataset != NULL) rrw_exit(&os->os_dsl_dataset->ds_bp_rwlock, FTAG); #endif ASSERT(db->db.db_size != 0); dprintf_dbuf(db, "size=%llx\n", (u_longlong_t)db->db.db_size); if (db->db_blkid != DMU_BONUS_BLKID && db->db_state != DB_NOFILL) { dmu_objset_willuse_space(os, db->db.db_size, tx); } /* * If this buffer is dirty in an old transaction group we need * to make a copy of it so that the changes we make in this * transaction group won't leak out when we sync the older txg. */ dr = kmem_cache_alloc(dbuf_dirty_kmem_cache, KM_SLEEP); memset(dr, 0, sizeof (*dr)); list_link_init(&dr->dr_dirty_node); list_link_init(&dr->dr_dbuf_node); dr->dr_dnode = dn; if (db->db_level == 0) { void *data_old = db->db_buf; if (db->db_state != DB_NOFILL) { if (db->db_blkid == DMU_BONUS_BLKID) { dbuf_fix_old_data(db, tx->tx_txg); data_old = db->db.db_data; } else if (db->db.db_object != DMU_META_DNODE_OBJECT) { /* * Release the data buffer from the cache so * that we can modify it without impacting * possible other users of this cached data * block. Note that indirect blocks and * private objects are not released until the * syncing state (since they are only modified * then). */ arc_release(db->db_buf, db); dbuf_fix_old_data(db, tx->tx_txg); data_old = db->db_buf; } ASSERT(data_old != NULL); } dr->dt.dl.dr_data = data_old; } else { mutex_init(&dr->dt.di.dr_mtx, NULL, MUTEX_NOLOCKDEP, NULL); list_create(&dr->dt.di.dr_children, sizeof (dbuf_dirty_record_t), offsetof(dbuf_dirty_record_t, dr_dirty_node)); } if (db->db_blkid != DMU_BONUS_BLKID && db->db_state != DB_NOFILL) { dr->dr_accounted = db->db.db_size; } dr->dr_dbuf = db; dr->dr_txg = tx->tx_txg; list_insert_before(&db->db_dirty_records, dr_next, dr); /* * We could have been freed_in_flight between the dbuf_noread * and dbuf_dirty. We win, as though the dbuf_noread() had * happened after the free. */ if (db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID && db->db_blkid != DMU_SPILL_BLKID) { mutex_enter(&dn->dn_mtx); if (dn->dn_free_ranges[txgoff] != NULL) { zfs_range_tree_clear(dn->dn_free_ranges[txgoff], db->db_blkid, 1); } mutex_exit(&dn->dn_mtx); db->db_freed_in_flight = FALSE; } /* * This buffer is now part of this txg */ dbuf_add_ref(db, (void *)(uintptr_t)tx->tx_txg); db->db_dirtycnt += 1; ASSERT3U(db->db_dirtycnt, <=, 3); mutex_exit(&db->db_mtx); if (db->db_blkid == DMU_BONUS_BLKID || db->db_blkid == DMU_SPILL_BLKID) { mutex_enter(&dn->dn_mtx); ASSERT(!list_link_active(&dr->dr_dirty_node)); list_insert_tail(&dn->dn_dirty_records[txgoff], dr); mutex_exit(&dn->dn_mtx); dnode_setdirty(dn, tx); DB_DNODE_EXIT(db); return (dr); } if (!RW_WRITE_HELD(&dn->dn_struct_rwlock)) { rw_enter(&dn->dn_struct_rwlock, RW_READER); drop_struct_rwlock = B_TRUE; } /* * If we are overwriting a dedup BP, then unless it is snapshotted, * when we get to syncing context we will need to decrement its * refcount in the DDT. Prefetch the relevant DDT block so that * syncing context won't have to wait for the i/o. */ if (db->db_blkptr != NULL) { db_lock_type_t dblt = dmu_buf_lock_parent(db, RW_READER, FTAG); ddt_prefetch(os->os_spa, db->db_blkptr); dmu_buf_unlock_parent(db, dblt, FTAG); } /* * We need to hold the dn_struct_rwlock to make this assertion, * because it protects dn_phys / dn_next_nlevels from changing. */ ASSERT((dn->dn_phys->dn_nlevels == 0 && db->db_level == 0) || dn->dn_phys->dn_nlevels > db->db_level || dn->dn_next_nlevels[txgoff] > db->db_level || dn->dn_next_nlevels[(tx->tx_txg-1) & TXG_MASK] > db->db_level || dn->dn_next_nlevels[(tx->tx_txg-2) & TXG_MASK] > db->db_level); if (db->db_level == 0) { ASSERT(!db->db_objset->os_raw_receive || dn->dn_maxblkid >= db->db_blkid); dnode_new_blkid(dn, db->db_blkid, tx, drop_struct_rwlock, B_FALSE); ASSERT(dn->dn_maxblkid >= db->db_blkid); } if (db->db_level+1 < dn->dn_nlevels) { dmu_buf_impl_t *parent = db->db_parent; dbuf_dirty_record_t *di; int parent_held = FALSE; if (db->db_parent == NULL || db->db_parent == dn->dn_dbuf) { int epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT; parent = dbuf_hold_level(dn, db->db_level + 1, db->db_blkid >> epbs, FTAG); ASSERT(parent != NULL); parent_held = TRUE; } if (drop_struct_rwlock) rw_exit(&dn->dn_struct_rwlock); ASSERT3U(db->db_level + 1, ==, parent->db_level); di = dbuf_dirty(parent, tx); if (parent_held) dbuf_rele(parent, FTAG); mutex_enter(&db->db_mtx); /* * Since we've dropped the mutex, it's possible that * dbuf_undirty() might have changed this out from under us. */ if (list_head(&db->db_dirty_records) == dr || dn->dn_object == DMU_META_DNODE_OBJECT) { mutex_enter(&di->dt.di.dr_mtx); ASSERT3U(di->dr_txg, ==, tx->tx_txg); ASSERT(!list_link_active(&dr->dr_dirty_node)); list_insert_tail(&di->dt.di.dr_children, dr); mutex_exit(&di->dt.di.dr_mtx); dr->dr_parent = di; } mutex_exit(&db->db_mtx); } else { ASSERT(db->db_level + 1 == dn->dn_nlevels); ASSERT(db->db_blkid < dn->dn_nblkptr); ASSERT(db->db_parent == NULL || db->db_parent == dn->dn_dbuf); mutex_enter(&dn->dn_mtx); ASSERT(!list_link_active(&dr->dr_dirty_node)); list_insert_tail(&dn->dn_dirty_records[txgoff], dr); mutex_exit(&dn->dn_mtx); if (drop_struct_rwlock) rw_exit(&dn->dn_struct_rwlock); } dnode_setdirty(dn, tx); DB_DNODE_EXIT(db); return (dr); } static void dbuf_undirty_bonus(dbuf_dirty_record_t *dr) { dmu_buf_impl_t *db = dr->dr_dbuf; ASSERT(MUTEX_HELD(&db->db_mtx)); if (dr->dt.dl.dr_data != db->db.db_data) { struct dnode *dn = dr->dr_dnode; int max_bonuslen = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots); kmem_free(dr->dt.dl.dr_data, max_bonuslen); arc_space_return(max_bonuslen, ARC_SPACE_BONUS); } db->db_data_pending = NULL; ASSERT(list_next(&db->db_dirty_records, dr) == NULL); list_remove(&db->db_dirty_records, dr); if (dr->dr_dbuf->db_level != 0) { mutex_destroy(&dr->dt.di.dr_mtx); list_destroy(&dr->dt.di.dr_children); } kmem_cache_free(dbuf_dirty_kmem_cache, dr); ASSERT3U(db->db_dirtycnt, >, 0); db->db_dirtycnt -= 1; } /* * Undirty a buffer in the transaction group referenced by the given * transaction. Return whether this evicted the dbuf. */ boolean_t dbuf_undirty(dmu_buf_impl_t *db, dmu_tx_t *tx) { uint64_t txg = tx->tx_txg; boolean_t brtwrite; boolean_t diowrite; ASSERT(txg != 0); /* * Due to our use of dn_nlevels below, this can only be called * in open context, unless we are operating on the MOS. * From syncing context, dn_nlevels may be different from the * dn_nlevels used when dbuf was dirtied. */ ASSERT(db->db_objset == dmu_objset_pool(db->db_objset)->dp_meta_objset || txg != spa_syncing_txg(dmu_objset_spa(db->db_objset))); ASSERT(db->db_blkid != DMU_BONUS_BLKID); ASSERT0(db->db_level); ASSERT(MUTEX_HELD(&db->db_mtx)); /* * If this buffer is not dirty, we're done. */ dbuf_dirty_record_t *dr = dbuf_find_dirty_eq(db, txg); if (dr == NULL) return (B_FALSE); ASSERT(dr->dr_dbuf == db); brtwrite = dr->dt.dl.dr_brtwrite; diowrite = dr->dt.dl.dr_diowrite; if (brtwrite) { ASSERT3B(diowrite, ==, B_FALSE); /* * We are freeing a block that we cloned in the same * transaction group. */ blkptr_t *bp = &dr->dt.dl.dr_overridden_by; if (!BP_IS_HOLE(bp) && !BP_IS_EMBEDDED(bp)) { brt_pending_remove(dmu_objset_spa(db->db_objset), bp, tx); } } dnode_t *dn = dr->dr_dnode; dprintf_dbuf(db, "size=%llx\n", (u_longlong_t)db->db.db_size); ASSERT(db->db.db_size != 0); dsl_pool_undirty_space(dmu_objset_pool(dn->dn_objset), dr->dr_accounted, txg); list_remove(&db->db_dirty_records, dr); /* * Note that there are three places in dbuf_dirty() * where this dirty record may be put on a list. * Make sure to do a list_remove corresponding to * every one of those list_insert calls. */ if (dr->dr_parent) { mutex_enter(&dr->dr_parent->dt.di.dr_mtx); list_remove(&dr->dr_parent->dt.di.dr_children, dr); mutex_exit(&dr->dr_parent->dt.di.dr_mtx); } else if (db->db_blkid == DMU_SPILL_BLKID || db->db_level + 1 == dn->dn_nlevels) { ASSERT(db->db_blkptr == NULL || db->db_parent == dn->dn_dbuf); mutex_enter(&dn->dn_mtx); list_remove(&dn->dn_dirty_records[txg & TXG_MASK], dr); mutex_exit(&dn->dn_mtx); } if (db->db_state != DB_NOFILL && !brtwrite) { dbuf_unoverride(dr); if (dr->dt.dl.dr_data != db->db_buf) { ASSERT(db->db_buf != NULL); ASSERT(dr->dt.dl.dr_data != NULL); arc_buf_destroy(dr->dt.dl.dr_data, db); } } kmem_cache_free(dbuf_dirty_kmem_cache, dr); ASSERT(db->db_dirtycnt > 0); db->db_dirtycnt -= 1; if (zfs_refcount_remove(&db->db_holds, (void *)(uintptr_t)txg) == 0) { ASSERT(db->db_state == DB_NOFILL || brtwrite || diowrite || arc_released(db->db_buf)); dbuf_destroy(db); return (B_TRUE); } return (B_FALSE); } void dmu_buf_will_dirty_flags(dmu_buf_t *db_fake, dmu_tx_t *tx, dmu_flags_t flags) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; boolean_t undirty = B_FALSE; ASSERT(tx->tx_txg != 0); ASSERT(!zfs_refcount_is_zero(&db->db_holds)); /* * Quick check for dirtiness to improve performance for some workloads * (e.g. file deletion with indirect blocks cached). */ mutex_enter(&db->db_mtx); if (db->db_state == DB_CACHED || db->db_state == DB_NOFILL) { /* * It's possible that the dbuf is already dirty but not cached, * because there are some calls to dbuf_dirty() that don't * go through dmu_buf_will_dirty(). */ dbuf_dirty_record_t *dr = dbuf_find_dirty_eq(db, tx->tx_txg); if (dr != NULL) { if (db->db_level == 0 && dr->dt.dl.dr_brtwrite) { /* * Block cloning: If we are dirtying a cloned * level 0 block, we cannot simply redirty it, * because this dr has no associated data. * We will go through a full undirtying below, * before dirtying it again. */ undirty = B_TRUE; } else { /* This dbuf is already dirty and cached. */ dbuf_redirty(dr); mutex_exit(&db->db_mtx); return; } } } mutex_exit(&db->db_mtx); DB_DNODE_ENTER(db); if (RW_WRITE_HELD(&DB_DNODE(db)->dn_struct_rwlock)) flags |= DB_RF_HAVESTRUCT; DB_DNODE_EXIT(db); /* * Block cloning: Do the dbuf_read() before undirtying the dbuf, as we * want to make sure dbuf_read() will read the pending cloned block and * not the uderlying block that is being replaced. dbuf_undirty() will * do brt_pending_remove() before removing the dirty record. */ (void) dbuf_read(db, NULL, flags | DB_RF_MUST_SUCCEED); if (undirty) { mutex_enter(&db->db_mtx); VERIFY(!dbuf_undirty(db, tx)); mutex_exit(&db->db_mtx); } (void) dbuf_dirty(db, tx); } void dmu_buf_will_dirty(dmu_buf_t *db_fake, dmu_tx_t *tx) { dmu_buf_will_dirty_flags(db_fake, tx, DMU_READ_NO_PREFETCH); } +void +dmu_buf_will_rewrite(dmu_buf_t *db_fake, dmu_tx_t *tx) +{ + dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; + + ASSERT(tx->tx_txg != 0); + ASSERT(!zfs_refcount_is_zero(&db->db_holds)); + + /* + * If the dbuf is already dirty in this txg, it will be written + * anyway, so there's nothing to do. + */ + mutex_enter(&db->db_mtx); + if (dbuf_find_dirty_eq(db, tx->tx_txg) != NULL) { + mutex_exit(&db->db_mtx); + return; + } + mutex_exit(&db->db_mtx); + + /* + * The dbuf is not dirty, so we need to make it dirty and + * mark it for rewrite (preserve logical birth time). + */ + dmu_buf_will_dirty_flags(db_fake, tx, DMU_READ_NO_PREFETCH); + + mutex_enter(&db->db_mtx); + dbuf_dirty_record_t *dr = dbuf_find_dirty_eq(db, tx->tx_txg); + if (dr != NULL && db->db_level == 0) + dr->dt.dl.dr_rewrite = B_TRUE; + mutex_exit(&db->db_mtx); +} + boolean_t dmu_buf_is_dirty(dmu_buf_t *db_fake, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dbuf_dirty_record_t *dr; mutex_enter(&db->db_mtx); dr = dbuf_find_dirty_eq(db, tx->tx_txg); mutex_exit(&db->db_mtx); return (dr != NULL); } /* * Normally the db_blkptr points to the most recent on-disk content for the * dbuf (and anything newer will be cached in the dbuf). However, a pending * block clone or not yet synced Direct I/O write will have a dirty record BP * pointing to the most recent data. */ int dmu_buf_get_bp_from_dbuf(dmu_buf_impl_t *db, blkptr_t **bp) { ASSERT(MUTEX_HELD(&db->db_mtx)); int error = 0; if (db->db_level != 0) { *bp = db->db_blkptr; return (0); } *bp = db->db_blkptr; dbuf_dirty_record_t *dr = list_head(&db->db_dirty_records); if (dr && db->db_state == DB_NOFILL) { /* Block clone */ if (!dr->dt.dl.dr_brtwrite) error = EIO; else *bp = &dr->dt.dl.dr_overridden_by; } else if (dr && db->db_state == DB_UNCACHED) { /* Direct I/O write */ if (dr->dt.dl.dr_diowrite) *bp = &dr->dt.dl.dr_overridden_by; } return (error); } /* * Direct I/O reads can read directly from the ARC, but the data has * to be untransformed in order to copy it over into user pages. */ int dmu_buf_untransform_direct(dmu_buf_impl_t *db, spa_t *spa) { int err = 0; DB_DNODE_ENTER(db); dnode_t *dn = DB_DNODE(db); ASSERT3S(db->db_state, ==, DB_CACHED); ASSERT(MUTEX_HELD(&db->db_mtx)); /* * Ensure that this block's dnode has been decrypted if * the caller has requested decrypted data. */ err = dbuf_read_verify_dnode_crypt(db, dn, 0); /* * If the arc buf is compressed or encrypted and the caller * requested uncompressed data, we need to untransform it * before returning. We also call arc_untransform() on any * unauthenticated blocks, which will verify their MAC if * the key is now available. */ if (err == 0 && db->db_buf != NULL && (arc_is_encrypted(db->db_buf) || arc_is_unauthenticated(db->db_buf) || arc_get_compression(db->db_buf) != ZIO_COMPRESS_OFF)) { zbookmark_phys_t zb; SET_BOOKMARK(&zb, dmu_objset_id(db->db_objset), db->db.db_object, db->db_level, db->db_blkid); dbuf_fix_old_data(db, spa_syncing_txg(spa)); err = arc_untransform(db->db_buf, spa, &zb, B_FALSE); dbuf_set_data(db, db->db_buf); } DB_DNODE_EXIT(db); DBUF_STAT_BUMP(hash_hits); return (err); } void dmu_buf_will_clone_or_dio(dmu_buf_t *db_fake, dmu_tx_t *tx) { /* * Block clones and Direct I/O writes always happen in open-context. */ dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; ASSERT0(db->db_level); ASSERT(!dmu_tx_is_syncing(tx)); ASSERT0(db->db_level); ASSERT(db->db_blkid != DMU_BONUS_BLKID); ASSERT(db->db.db_object != DMU_META_DNODE_OBJECT); mutex_enter(&db->db_mtx); DBUF_VERIFY(db); /* * We are going to clone or issue a Direct I/O write on this block, so * undirty modifications done to this block so far in this txg. This * includes writes and clones into this block. * * If there dirty record associated with this txg from a previous Direct * I/O write then space accounting cleanup takes place. It is important * to go ahead free up the space accounting through dbuf_undirty() -> * dbuf_unoverride() -> zio_free(). Space accountiung for determining * if a write can occur in zfs_write() happens through dmu_tx_assign(). * This can cause an issue with Direct I/O writes in the case of * overwriting the same block, because all DVA allocations are being * done in open-context. Constantly allowing Direct I/O overwrites to * the same block can exhaust the pools available space leading to * ENOSPC errors at the DVA allocation part of the ZIO pipeline, which * will eventually suspend the pool. By cleaning up sapce acccounting * now, the ENOSPC error can be avoided. * * Since we are undirtying the record in open-context, we must have a * hold on the db, so it should never be evicted after calling * dbuf_undirty(). */ VERIFY3B(dbuf_undirty(db, tx), ==, B_FALSE); ASSERT0P(dbuf_find_dirty_eq(db, tx->tx_txg)); if (db->db_buf != NULL) { /* * If there is an associated ARC buffer with this dbuf we can * only destroy it if the previous dirty record does not * reference it. */ dbuf_dirty_record_t *dr = list_head(&db->db_dirty_records); if (dr == NULL || dr->dt.dl.dr_data != db->db_buf) arc_buf_destroy(db->db_buf, db); /* * Setting the dbuf's data pointers to NULL will force all * future reads down to the devices to get the most up to date * version of the data after a Direct I/O write has completed. */ db->db_buf = NULL; dbuf_clear_data(db); } ASSERT3P(db->db_buf, ==, NULL); ASSERT3P(db->db.db_data, ==, NULL); db->db_state = DB_NOFILL; DTRACE_SET_STATE(db, "allocating NOFILL buffer for clone or direct I/O write"); DBUF_VERIFY(db); mutex_exit(&db->db_mtx); dbuf_noread(db, DMU_KEEP_CACHING); (void) dbuf_dirty(db, tx); } void dmu_buf_will_not_fill(dmu_buf_t *db_fake, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; mutex_enter(&db->db_mtx); db->db_state = DB_NOFILL; DTRACE_SET_STATE(db, "allocating NOFILL buffer"); mutex_exit(&db->db_mtx); dbuf_noread(db, DMU_KEEP_CACHING); (void) dbuf_dirty(db, tx); } void dmu_buf_will_fill_flags(dmu_buf_t *db_fake, dmu_tx_t *tx, boolean_t canfail, dmu_flags_t flags) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; ASSERT(db->db_blkid != DMU_BONUS_BLKID); ASSERT(tx->tx_txg != 0); ASSERT(db->db_level == 0); ASSERT(!zfs_refcount_is_zero(&db->db_holds)); ASSERT(db->db.db_object != DMU_META_DNODE_OBJECT || dmu_tx_private_ok(tx)); mutex_enter(&db->db_mtx); dbuf_dirty_record_t *dr = dbuf_find_dirty_eq(db, tx->tx_txg); if (db->db_state == DB_NOFILL || (db->db_state == DB_UNCACHED && dr && dr->dt.dl.dr_diowrite)) { /* * If the fill can fail we should have a way to return back to * the cloned or Direct I/O write data. */ if (canfail && dr) { mutex_exit(&db->db_mtx); dmu_buf_will_dirty_flags(db_fake, tx, flags); return; } /* * Block cloning: We will be completely overwriting a block * cloned in this transaction group, so let's undirty the * pending clone and mark the block as uncached. This will be * as if the clone was never done. */ if (db->db_state == DB_NOFILL) { VERIFY(!dbuf_undirty(db, tx)); db->db_state = DB_UNCACHED; } } mutex_exit(&db->db_mtx); dbuf_noread(db, flags); (void) dbuf_dirty(db, tx); } void dmu_buf_will_fill(dmu_buf_t *db_fake, dmu_tx_t *tx, boolean_t canfail) { dmu_buf_will_fill_flags(db_fake, tx, canfail, DMU_READ_NO_PREFETCH); } /* * This function is effectively the same as dmu_buf_will_dirty(), but * indicates the caller expects raw encrypted data in the db, and provides * the crypt params (byteorder, salt, iv, mac) which should be stored in the * blkptr_t when this dbuf is written. This is only used for blocks of * dnodes, during raw receive. */ void dmu_buf_set_crypt_params(dmu_buf_t *db_fake, boolean_t byteorder, const uint8_t *salt, const uint8_t *iv, const uint8_t *mac, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dbuf_dirty_record_t *dr; /* * dr_has_raw_params is only processed for blocks of dnodes * (see dbuf_sync_dnode_leaf_crypt()). */ ASSERT3U(db->db.db_object, ==, DMU_META_DNODE_OBJECT); ASSERT0(db->db_level); ASSERT(db->db_objset->os_raw_receive); dmu_buf_will_dirty_flags(db_fake, tx, DMU_READ_NO_PREFETCH | DMU_READ_NO_DECRYPT); dr = dbuf_find_dirty_eq(db, tx->tx_txg); ASSERT3P(dr, !=, NULL); ASSERT3U(dr->dt.dl.dr_override_state, ==, DR_NOT_OVERRIDDEN); dr->dt.dl.dr_has_raw_params = B_TRUE; dr->dt.dl.dr_byteorder = byteorder; memcpy(dr->dt.dl.dr_salt, salt, ZIO_DATA_SALT_LEN); memcpy(dr->dt.dl.dr_iv, iv, ZIO_DATA_IV_LEN); memcpy(dr->dt.dl.dr_mac, mac, ZIO_DATA_MAC_LEN); } static void dbuf_override_impl(dmu_buf_impl_t *db, const blkptr_t *bp, dmu_tx_t *tx) { struct dirty_leaf *dl; dbuf_dirty_record_t *dr; ASSERT3U(db->db.db_object, !=, DMU_META_DNODE_OBJECT); ASSERT0(db->db_level); dr = list_head(&db->db_dirty_records); ASSERT3P(dr, !=, NULL); ASSERT3U(dr->dr_txg, ==, tx->tx_txg); dl = &dr->dt.dl; ASSERT0(dl->dr_has_raw_params); dl->dr_overridden_by = *bp; dl->dr_override_state = DR_OVERRIDDEN; BP_SET_LOGICAL_BIRTH(&dl->dr_overridden_by, dr->dr_txg); } boolean_t dmu_buf_fill_done(dmu_buf_t *dbuf, dmu_tx_t *tx, boolean_t failed) { (void) tx; dmu_buf_impl_t *db = (dmu_buf_impl_t *)dbuf; mutex_enter(&db->db_mtx); DBUF_VERIFY(db); if (db->db_state == DB_FILL) { if (db->db_level == 0 && db->db_freed_in_flight) { ASSERT(db->db_blkid != DMU_BONUS_BLKID); /* we were freed while filling */ /* XXX dbuf_undirty? */ memset(db->db.db_data, 0, db->db.db_size); db->db_freed_in_flight = FALSE; db->db_state = DB_CACHED; DTRACE_SET_STATE(db, "fill done handling freed in flight"); failed = B_FALSE; } else if (failed) { VERIFY(!dbuf_undirty(db, tx)); arc_buf_destroy(db->db_buf, db); db->db_buf = NULL; dbuf_clear_data(db); DTRACE_SET_STATE(db, "fill failed"); } else { db->db_state = DB_CACHED; DTRACE_SET_STATE(db, "fill done"); } cv_broadcast(&db->db_changed); } else { db->db_state = DB_CACHED; failed = B_FALSE; } mutex_exit(&db->db_mtx); return (failed); } void dmu_buf_write_embedded(dmu_buf_t *dbuf, void *data, bp_embedded_type_t etype, enum zio_compress comp, int uncompressed_size, int compressed_size, int byteorder, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)dbuf; struct dirty_leaf *dl; dmu_object_type_t type; dbuf_dirty_record_t *dr; if (etype == BP_EMBEDDED_TYPE_DATA) { ASSERT(spa_feature_is_active(dmu_objset_spa(db->db_objset), SPA_FEATURE_EMBEDDED_DATA)); } DB_DNODE_ENTER(db); type = DB_DNODE(db)->dn_type; DB_DNODE_EXIT(db); ASSERT0(db->db_level); ASSERT(db->db_blkid != DMU_BONUS_BLKID); dmu_buf_will_not_fill(dbuf, tx); dr = list_head(&db->db_dirty_records); ASSERT3P(dr, !=, NULL); ASSERT3U(dr->dr_txg, ==, tx->tx_txg); dl = &dr->dt.dl; ASSERT0(dl->dr_has_raw_params); encode_embedded_bp_compressed(&dl->dr_overridden_by, data, comp, uncompressed_size, compressed_size); BPE_SET_ETYPE(&dl->dr_overridden_by, etype); BP_SET_TYPE(&dl->dr_overridden_by, type); BP_SET_LEVEL(&dl->dr_overridden_by, 0); BP_SET_BYTEORDER(&dl->dr_overridden_by, byteorder); dl->dr_override_state = DR_OVERRIDDEN; BP_SET_LOGICAL_BIRTH(&dl->dr_overridden_by, dr->dr_txg); } void dmu_buf_redact(dmu_buf_t *dbuf, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)dbuf; dmu_object_type_t type; ASSERT(dsl_dataset_feature_is_active(db->db_objset->os_dsl_dataset, SPA_FEATURE_REDACTED_DATASETS)); DB_DNODE_ENTER(db); type = DB_DNODE(db)->dn_type; DB_DNODE_EXIT(db); ASSERT0(db->db_level); dmu_buf_will_not_fill(dbuf, tx); blkptr_t bp = { { { {0} } } }; BP_SET_TYPE(&bp, type); BP_SET_LEVEL(&bp, 0); BP_SET_BIRTH(&bp, tx->tx_txg, 0); BP_SET_REDACTED(&bp); BPE_SET_LSIZE(&bp, dbuf->db_size); dbuf_override_impl(db, &bp, tx); } /* * Directly assign a provided arc buf to a given dbuf if it's not referenced * by anybody except our caller. Otherwise copy arcbuf's contents to dbuf. */ void dbuf_assign_arcbuf(dmu_buf_impl_t *db, arc_buf_t *buf, dmu_tx_t *tx, dmu_flags_t flags) { ASSERT(!zfs_refcount_is_zero(&db->db_holds)); ASSERT(db->db_blkid != DMU_BONUS_BLKID); ASSERT(db->db_level == 0); ASSERT3U(dbuf_is_metadata(db), ==, arc_is_metadata(buf)); ASSERT(buf != NULL); ASSERT3U(arc_buf_lsize(buf), ==, db->db.db_size); ASSERT(tx->tx_txg != 0); arc_return_buf(buf, db); ASSERT(arc_released(buf)); mutex_enter(&db->db_mtx); if (!(flags & (DMU_UNCACHEDIO | DMU_KEEP_CACHING))) db->db_pending_evict = B_FALSE; db->db_partial_read = B_FALSE; while (db->db_state == DB_READ || db->db_state == DB_FILL) cv_wait(&db->db_changed, &db->db_mtx); ASSERT(db->db_state == DB_CACHED || db->db_state == DB_UNCACHED || db->db_state == DB_NOFILL); if (db->db_state == DB_CACHED && zfs_refcount_count(&db->db_holds) - 1 > db->db_dirtycnt) { /* * In practice, we will never have a case where we have an * encrypted arc buffer while additional holds exist on the * dbuf. We don't handle this here so we simply assert that * fact instead. */ ASSERT(!arc_is_encrypted(buf)); mutex_exit(&db->db_mtx); (void) dbuf_dirty(db, tx); memcpy(db->db.db_data, buf->b_data, db->db.db_size); arc_buf_destroy(buf, db); return; } if (db->db_state == DB_CACHED) { dbuf_dirty_record_t *dr = list_head(&db->db_dirty_records); ASSERT(db->db_buf != NULL); if (dr != NULL && dr->dr_txg == tx->tx_txg) { ASSERT(dr->dt.dl.dr_data == db->db_buf); if (!arc_released(db->db_buf)) { ASSERT(dr->dt.dl.dr_override_state == DR_OVERRIDDEN); arc_release(db->db_buf, db); } dr->dt.dl.dr_data = buf; arc_buf_destroy(db->db_buf, db); } else if (dr == NULL || dr->dt.dl.dr_data != db->db_buf) { arc_release(db->db_buf, db); arc_buf_destroy(db->db_buf, db); } db->db_buf = NULL; } else if (db->db_state == DB_NOFILL) { /* * We will be completely replacing the cloned block. In case * it was cloned in this transaction group, let's undirty the * pending clone and mark the block as uncached. This will be * as if the clone was never done. */ VERIFY(!dbuf_undirty(db, tx)); db->db_state = DB_UNCACHED; } ASSERT(db->db_buf == NULL); dbuf_set_data(db, buf); db->db_state = DB_FILL; DTRACE_SET_STATE(db, "filling assigned arcbuf"); mutex_exit(&db->db_mtx); (void) dbuf_dirty(db, tx); dmu_buf_fill_done(&db->db, tx, B_FALSE); } void dbuf_destroy(dmu_buf_impl_t *db) { dnode_t *dn; dmu_buf_impl_t *parent = db->db_parent; dmu_buf_impl_t *dndb; ASSERT(MUTEX_HELD(&db->db_mtx)); ASSERT(zfs_refcount_is_zero(&db->db_holds)); if (db->db_buf != NULL) { arc_buf_destroy(db->db_buf, db); db->db_buf = NULL; } if (db->db_blkid == DMU_BONUS_BLKID) { int slots = DB_DNODE(db)->dn_num_slots; int bonuslen = DN_SLOTS_TO_BONUSLEN(slots); if (db->db.db_data != NULL) { kmem_free(db->db.db_data, bonuslen); arc_space_return(bonuslen, ARC_SPACE_BONUS); db->db_state = DB_UNCACHED; DTRACE_SET_STATE(db, "buffer cleared"); } } dbuf_clear_data(db); if (multilist_link_active(&db->db_cache_link)) { ASSERT(db->db_caching_status == DB_DBUF_CACHE || db->db_caching_status == DB_DBUF_METADATA_CACHE); multilist_remove(&dbuf_caches[db->db_caching_status].cache, db); ASSERT0(dmu_buf_user_size(&db->db)); (void) zfs_refcount_remove_many( &dbuf_caches[db->db_caching_status].size, db->db.db_size, db); if (db->db_caching_status == DB_DBUF_METADATA_CACHE) { DBUF_STAT_BUMPDOWN(metadata_cache_count); } else { DBUF_STAT_BUMPDOWN(cache_levels[db->db_level]); DBUF_STAT_BUMPDOWN(cache_count); DBUF_STAT_DECR(cache_levels_bytes[db->db_level], db->db.db_size); } db->db_caching_status = DB_NO_CACHE; } ASSERT(db->db_state == DB_UNCACHED || db->db_state == DB_NOFILL); ASSERT(db->db_data_pending == NULL); ASSERT(list_is_empty(&db->db_dirty_records)); db->db_state = DB_EVICTING; DTRACE_SET_STATE(db, "buffer eviction started"); db->db_blkptr = NULL; /* * Now that db_state is DB_EVICTING, nobody else can find this via * the hash table. We can now drop db_mtx, which allows us to * acquire the dn_dbufs_mtx. */ mutex_exit(&db->db_mtx); DB_DNODE_ENTER(db); dn = DB_DNODE(db); dndb = dn->dn_dbuf; if (db->db_blkid != DMU_BONUS_BLKID) { boolean_t needlock = !MUTEX_HELD(&dn->dn_dbufs_mtx); if (needlock) mutex_enter_nested(&dn->dn_dbufs_mtx, NESTED_SINGLE); avl_remove(&dn->dn_dbufs, db); membar_producer(); DB_DNODE_EXIT(db); if (needlock) mutex_exit(&dn->dn_dbufs_mtx); /* * Decrementing the dbuf count means that the hold corresponding * to the removed dbuf is no longer discounted in dnode_move(), * so the dnode cannot be moved until after we release the hold. * The membar_producer() ensures visibility of the decremented * value in dnode_move(), since DB_DNODE_EXIT doesn't actually * release any lock. */ mutex_enter(&dn->dn_mtx); dnode_rele_and_unlock(dn, db, B_TRUE); #ifdef USE_DNODE_HANDLE db->db_dnode_handle = NULL; #else db->db_dnode = NULL; #endif dbuf_hash_remove(db); } else { DB_DNODE_EXIT(db); } ASSERT(zfs_refcount_is_zero(&db->db_holds)); db->db_parent = NULL; ASSERT(db->db_buf == NULL); ASSERT(db->db.db_data == NULL); ASSERT(db->db_hash_next == NULL); ASSERT(db->db_blkptr == NULL); ASSERT(db->db_data_pending == NULL); ASSERT3U(db->db_caching_status, ==, DB_NO_CACHE); ASSERT(!multilist_link_active(&db->db_cache_link)); /* * If this dbuf is referenced from an indirect dbuf, * decrement the ref count on the indirect dbuf. */ if (parent && parent != dndb) { mutex_enter(&parent->db_mtx); dbuf_rele_and_unlock(parent, db, B_TRUE); } kmem_cache_free(dbuf_kmem_cache, db); arc_space_return(sizeof (dmu_buf_impl_t), ARC_SPACE_DBUF); } /* * Note: While bpp will always be updated if the function returns success, * parentp will not be updated if the dnode does not have dn_dbuf filled in; * this happens when the dnode is the meta-dnode, or {user|group|project}used * object. */ __attribute__((always_inline)) static inline int dbuf_findbp(dnode_t *dn, int level, uint64_t blkid, int fail_sparse, dmu_buf_impl_t **parentp, blkptr_t **bpp) { *parentp = NULL; *bpp = NULL; ASSERT(blkid != DMU_BONUS_BLKID); if (blkid == DMU_SPILL_BLKID) { mutex_enter(&dn->dn_mtx); if (dn->dn_have_spill && (dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR)) *bpp = DN_SPILL_BLKPTR(dn->dn_phys); else *bpp = NULL; dbuf_add_ref(dn->dn_dbuf, NULL); *parentp = dn->dn_dbuf; mutex_exit(&dn->dn_mtx); return (0); } int nlevels = (dn->dn_phys->dn_nlevels == 0) ? 1 : dn->dn_phys->dn_nlevels; int epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT; ASSERT3U(level * epbs, <, 64); ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock)); /* * This assertion shouldn't trip as long as the max indirect block size * is less than 1M. The reason for this is that up to that point, * the number of levels required to address an entire object with blocks * of size SPA_MINBLOCKSIZE satisfies nlevels * epbs + 1 <= 64. In * other words, if N * epbs + 1 > 64, then if (N-1) * epbs + 1 > 55 * (i.e. we can address the entire object), objects will all use at most * N-1 levels and the assertion won't overflow. However, once epbs is * 13, 4 * 13 + 1 = 53, but 5 * 13 + 1 = 66. Then, 4 levels will not be * enough to address an entire object, so objects will have 5 levels, * but then this assertion will overflow. * * All this is to say that if we ever increase DN_MAX_INDBLKSHIFT, we * need to redo this logic to handle overflows. */ ASSERT(level >= nlevels || ((nlevels - level - 1) * epbs) + highbit64(dn->dn_phys->dn_nblkptr) <= 64); if (level >= nlevels || blkid >= ((uint64_t)dn->dn_phys->dn_nblkptr << ((nlevels - level - 1) * epbs)) || (fail_sparse && blkid > (dn->dn_phys->dn_maxblkid >> (level * epbs)))) { /* the buffer has no parent yet */ return (SET_ERROR(ENOENT)); } else if (level < nlevels-1) { /* this block is referenced from an indirect block */ int err; err = dbuf_hold_impl(dn, level + 1, blkid >> epbs, fail_sparse, FALSE, NULL, parentp); if (err) return (err); err = dbuf_read(*parentp, NULL, DB_RF_CANFAIL | DB_RF_HAVESTRUCT | DMU_READ_NO_PREFETCH); if (err) { dbuf_rele(*parentp, NULL); *parentp = NULL; return (err); } *bpp = ((blkptr_t *)(*parentp)->db.db_data) + (blkid & ((1ULL << epbs) - 1)); return (0); } else { /* the block is referenced from the dnode */ ASSERT3U(level, ==, nlevels-1); ASSERT(dn->dn_phys->dn_nblkptr == 0 || blkid < dn->dn_phys->dn_nblkptr); if (dn->dn_dbuf) { dbuf_add_ref(dn->dn_dbuf, NULL); *parentp = dn->dn_dbuf; } *bpp = &dn->dn_phys->dn_blkptr[blkid]; return (0); } } static dmu_buf_impl_t * dbuf_create(dnode_t *dn, uint8_t level, uint64_t blkid, dmu_buf_impl_t *parent, blkptr_t *blkptr, uint64_t hash) { objset_t *os = dn->dn_objset; dmu_buf_impl_t *db, *odb; ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock)); ASSERT(dn->dn_type != DMU_OT_NONE); db = kmem_cache_alloc(dbuf_kmem_cache, KM_SLEEP); list_create(&db->db_dirty_records, sizeof (dbuf_dirty_record_t), offsetof(dbuf_dirty_record_t, dr_dbuf_node)); db->db_objset = os; db->db.db_object = dn->dn_object; db->db_level = level; db->db_blkid = blkid; db->db_dirtycnt = 0; #ifdef USE_DNODE_HANDLE db->db_dnode_handle = dn->dn_handle; #else db->db_dnode = dn; #endif db->db_parent = parent; db->db_blkptr = blkptr; db->db_hash = hash; db->db_user = NULL; db->db_user_immediate_evict = FALSE; db->db_freed_in_flight = FALSE; db->db_pending_evict = TRUE; db->db_partial_read = FALSE; if (blkid == DMU_BONUS_BLKID) { ASSERT3P(parent, ==, dn->dn_dbuf); db->db.db_size = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots) - (dn->dn_nblkptr-1) * sizeof (blkptr_t); ASSERT3U(db->db.db_size, >=, dn->dn_bonuslen); db->db.db_offset = DMU_BONUS_BLKID; db->db_state = DB_UNCACHED; DTRACE_SET_STATE(db, "bonus buffer created"); db->db_caching_status = DB_NO_CACHE; /* the bonus dbuf is not placed in the hash table */ arc_space_consume(sizeof (dmu_buf_impl_t), ARC_SPACE_DBUF); return (db); } else if (blkid == DMU_SPILL_BLKID) { db->db.db_size = (blkptr != NULL) ? BP_GET_LSIZE(blkptr) : SPA_MINBLOCKSIZE; db->db.db_offset = 0; } else { int blocksize = db->db_level ? 1 << dn->dn_indblkshift : dn->dn_datablksz; db->db.db_size = blocksize; db->db.db_offset = db->db_blkid * blocksize; } /* * Hold the dn_dbufs_mtx while we get the new dbuf * in the hash table *and* added to the dbufs list. * This prevents a possible deadlock with someone * trying to look up this dbuf before it's added to the * dn_dbufs list. */ mutex_enter(&dn->dn_dbufs_mtx); db->db_state = DB_EVICTING; /* not worth logging this state change */ if ((odb = dbuf_hash_insert(db)) != NULL) { /* someone else inserted it first */ mutex_exit(&dn->dn_dbufs_mtx); kmem_cache_free(dbuf_kmem_cache, db); DBUF_STAT_BUMP(hash_insert_race); return (odb); } avl_add(&dn->dn_dbufs, db); db->db_state = DB_UNCACHED; DTRACE_SET_STATE(db, "regular buffer created"); db->db_caching_status = DB_NO_CACHE; mutex_exit(&dn->dn_dbufs_mtx); arc_space_consume(sizeof (dmu_buf_impl_t), ARC_SPACE_DBUF); if (parent && parent != dn->dn_dbuf) dbuf_add_ref(parent, db); ASSERT(dn->dn_object == DMU_META_DNODE_OBJECT || zfs_refcount_count(&dn->dn_holds) > 0); (void) zfs_refcount_add(&dn->dn_holds, db); dprintf_dbuf(db, "db=%p\n", db); return (db); } /* * This function returns a block pointer and information about the object, * given a dnode and a block. This is a publicly accessible version of * dbuf_findbp that only returns some information, rather than the * dbuf. Note that the dnode passed in must be held, and the dn_struct_rwlock * should be locked as (at least) a reader. */ int dbuf_dnode_findbp(dnode_t *dn, uint64_t level, uint64_t blkid, blkptr_t *bp, uint16_t *datablkszsec, uint8_t *indblkshift) { dmu_buf_impl_t *dbp = NULL; blkptr_t *bp2; int err = 0; ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock)); err = dbuf_findbp(dn, level, blkid, B_FALSE, &dbp, &bp2); if (err == 0) { ASSERT3P(bp2, !=, NULL); *bp = *bp2; if (dbp != NULL) dbuf_rele(dbp, NULL); if (datablkszsec != NULL) *datablkszsec = dn->dn_phys->dn_datablkszsec; if (indblkshift != NULL) *indblkshift = dn->dn_phys->dn_indblkshift; } return (err); } typedef struct dbuf_prefetch_arg { spa_t *dpa_spa; /* The spa to issue the prefetch in. */ zbookmark_phys_t dpa_zb; /* The target block to prefetch. */ int dpa_epbs; /* Entries (blkptr_t's) Per Block Shift. */ int dpa_curlevel; /* The current level that we're reading */ dnode_t *dpa_dnode; /* The dnode associated with the prefetch */ zio_priority_t dpa_prio; /* The priority I/Os should be issued at. */ zio_t *dpa_zio; /* The parent zio_t for all prefetches. */ arc_flags_t dpa_aflags; /* Flags to pass to the final prefetch. */ dbuf_prefetch_fn dpa_cb; /* prefetch completion callback */ void *dpa_arg; /* prefetch completion arg */ } dbuf_prefetch_arg_t; static void dbuf_prefetch_fini(dbuf_prefetch_arg_t *dpa, boolean_t io_done) { if (dpa->dpa_cb != NULL) { dpa->dpa_cb(dpa->dpa_arg, dpa->dpa_zb.zb_level, dpa->dpa_zb.zb_blkid, io_done); } kmem_free(dpa, sizeof (*dpa)); } static void dbuf_issue_final_prefetch_done(zio_t *zio, const zbookmark_phys_t *zb, const blkptr_t *iobp, arc_buf_t *abuf, void *private) { (void) zio, (void) zb, (void) iobp; dbuf_prefetch_arg_t *dpa = private; if (abuf != NULL) arc_buf_destroy(abuf, private); dbuf_prefetch_fini(dpa, B_TRUE); } /* * Actually issue the prefetch read for the block given. */ static void dbuf_issue_final_prefetch(dbuf_prefetch_arg_t *dpa, blkptr_t *bp) { ASSERT(!BP_IS_HOLE(bp)); ASSERT(!BP_IS_REDACTED(bp)); if (BP_IS_EMBEDDED(bp)) return (dbuf_prefetch_fini(dpa, B_FALSE)); int zio_flags = ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE; arc_flags_t aflags = dpa->dpa_aflags | ARC_FLAG_NOWAIT | ARC_FLAG_PREFETCH | ARC_FLAG_NO_BUF; /* dnodes are always read as raw and then converted later */ if (BP_GET_TYPE(bp) == DMU_OT_DNODE && BP_IS_PROTECTED(bp) && dpa->dpa_curlevel == 0) zio_flags |= ZIO_FLAG_RAW; ASSERT3U(dpa->dpa_curlevel, ==, BP_GET_LEVEL(bp)); ASSERT3U(dpa->dpa_curlevel, ==, dpa->dpa_zb.zb_level); ASSERT(dpa->dpa_zio != NULL); (void) arc_read(dpa->dpa_zio, dpa->dpa_spa, bp, dbuf_issue_final_prefetch_done, dpa, dpa->dpa_prio, zio_flags, &aflags, &dpa->dpa_zb); } /* * Called when an indirect block above our prefetch target is read in. This * will either read in the next indirect block down the tree or issue the actual * prefetch if the next block down is our target. */ static void dbuf_prefetch_indirect_done(zio_t *zio, const zbookmark_phys_t *zb, const blkptr_t *iobp, arc_buf_t *abuf, void *private) { (void) zb, (void) iobp; dbuf_prefetch_arg_t *dpa = private; ASSERT3S(dpa->dpa_zb.zb_level, <, dpa->dpa_curlevel); ASSERT3S(dpa->dpa_curlevel, >, 0); if (abuf == NULL) { ASSERT(zio == NULL || zio->io_error != 0); dbuf_prefetch_fini(dpa, B_TRUE); return; } ASSERT(zio == NULL || zio->io_error == 0); /* * The dpa_dnode is only valid if we are called with a NULL * zio. This indicates that the arc_read() returned without * first calling zio_read() to issue a physical read. Once * a physical read is made the dpa_dnode must be invalidated * as the locks guarding it may have been dropped. If the * dpa_dnode is still valid, then we want to add it to the dbuf * cache. To do so, we must hold the dbuf associated with the block * we just prefetched, read its contents so that we associate it * with an arc_buf_t, and then release it. */ if (zio != NULL) { ASSERT3S(BP_GET_LEVEL(zio->io_bp), ==, dpa->dpa_curlevel); if (zio->io_flags & ZIO_FLAG_RAW_COMPRESS) { ASSERT3U(BP_GET_PSIZE(zio->io_bp), ==, zio->io_size); } else { ASSERT3U(BP_GET_LSIZE(zio->io_bp), ==, zio->io_size); } ASSERT3P(zio->io_spa, ==, dpa->dpa_spa); dpa->dpa_dnode = NULL; } else if (dpa->dpa_dnode != NULL) { uint64_t curblkid = dpa->dpa_zb.zb_blkid >> (dpa->dpa_epbs * (dpa->dpa_curlevel - dpa->dpa_zb.zb_level)); dmu_buf_impl_t *db = dbuf_hold_level(dpa->dpa_dnode, dpa->dpa_curlevel, curblkid, FTAG); if (db == NULL) { arc_buf_destroy(abuf, private); dbuf_prefetch_fini(dpa, B_TRUE); return; } (void) dbuf_read(db, NULL, DB_RF_CANFAIL | DB_RF_HAVESTRUCT | DMU_READ_NO_PREFETCH); dbuf_rele(db, FTAG); } dpa->dpa_curlevel--; uint64_t nextblkid = dpa->dpa_zb.zb_blkid >> (dpa->dpa_epbs * (dpa->dpa_curlevel - dpa->dpa_zb.zb_level)); blkptr_t *bp = ((blkptr_t *)abuf->b_data) + P2PHASE(nextblkid, 1ULL << dpa->dpa_epbs); ASSERT(!BP_IS_REDACTED(bp) || dpa->dpa_dnode == NULL || dsl_dataset_feature_is_active( dpa->dpa_dnode->dn_objset->os_dsl_dataset, SPA_FEATURE_REDACTED_DATASETS)); if (BP_IS_HOLE(bp) || BP_IS_REDACTED(bp)) { arc_buf_destroy(abuf, private); dbuf_prefetch_fini(dpa, B_TRUE); return; } else if (dpa->dpa_curlevel == dpa->dpa_zb.zb_level) { ASSERT3U(nextblkid, ==, dpa->dpa_zb.zb_blkid); dbuf_issue_final_prefetch(dpa, bp); } else { arc_flags_t iter_aflags = ARC_FLAG_NOWAIT; zbookmark_phys_t zb; /* flag if L2ARC eligible, l2arc_noprefetch then decides */ if (dpa->dpa_dnode) { if (dnode_level_is_l2cacheable(bp, dpa->dpa_dnode, dpa->dpa_curlevel)) iter_aflags |= ARC_FLAG_L2CACHE; } else { if (dpa->dpa_aflags & ARC_FLAG_L2CACHE) iter_aflags |= ARC_FLAG_L2CACHE; } ASSERT3U(dpa->dpa_curlevel, ==, BP_GET_LEVEL(bp)); SET_BOOKMARK(&zb, dpa->dpa_zb.zb_objset, dpa->dpa_zb.zb_object, dpa->dpa_curlevel, nextblkid); (void) arc_read(dpa->dpa_zio, dpa->dpa_spa, bp, dbuf_prefetch_indirect_done, dpa, ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &iter_aflags, &zb); } arc_buf_destroy(abuf, private); } /* * Issue prefetch reads for the given block on the given level. If the indirect * blocks above that block are not in memory, we will read them in * asynchronously. As a result, this call never blocks waiting for a read to * complete. Note that the prefetch might fail if the dataset is encrypted and * the encryption key is unmapped before the IO completes. */ int dbuf_prefetch_impl(dnode_t *dn, int64_t level, uint64_t blkid, zio_priority_t prio, arc_flags_t aflags, dbuf_prefetch_fn cb, void *arg) { blkptr_t bp; int epbs, nlevels, curlevel; uint64_t curblkid; ASSERT(blkid != DMU_BONUS_BLKID); ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock)); if (blkid > dn->dn_maxblkid) goto no_issue; if (level == 0 && dnode_block_freed(dn, blkid)) goto no_issue; /* * This dnode hasn't been written to disk yet, so there's nothing to * prefetch. */ nlevels = dn->dn_phys->dn_nlevels; if (level >= nlevels || dn->dn_phys->dn_nblkptr == 0) goto no_issue; epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT; if (dn->dn_phys->dn_maxblkid < blkid << (epbs * level)) goto no_issue; dmu_buf_impl_t *db = dbuf_find(dn->dn_objset, dn->dn_object, level, blkid, NULL); if (db != NULL) { mutex_exit(&db->db_mtx); /* * This dbuf already exists. It is either CACHED, or * (we assume) about to be read or filled. */ goto no_issue; } /* * Find the closest ancestor (indirect block) of the target block * that is present in the cache. In this indirect block, we will * find the bp that is at curlevel, curblkid. */ curlevel = level; curblkid = blkid; while (curlevel < nlevels - 1) { int parent_level = curlevel + 1; uint64_t parent_blkid = curblkid >> epbs; dmu_buf_impl_t *db; if (dbuf_hold_impl(dn, parent_level, parent_blkid, FALSE, TRUE, FTAG, &db) == 0) { blkptr_t *bpp = db->db_buf->b_data; bp = bpp[P2PHASE(curblkid, 1 << epbs)]; dbuf_rele(db, FTAG); break; } curlevel = parent_level; curblkid = parent_blkid; } if (curlevel == nlevels - 1) { /* No cached indirect blocks found. */ ASSERT3U(curblkid, <, dn->dn_phys->dn_nblkptr); bp = dn->dn_phys->dn_blkptr[curblkid]; } ASSERT(!BP_IS_REDACTED(&bp) || dsl_dataset_feature_is_active(dn->dn_objset->os_dsl_dataset, SPA_FEATURE_REDACTED_DATASETS)); if (BP_IS_HOLE(&bp) || BP_IS_REDACTED(&bp)) goto no_issue; ASSERT3U(curlevel, ==, BP_GET_LEVEL(&bp)); zio_t *pio = zio_root(dmu_objset_spa(dn->dn_objset), NULL, NULL, ZIO_FLAG_CANFAIL); dbuf_prefetch_arg_t *dpa = kmem_zalloc(sizeof (*dpa), KM_SLEEP); dsl_dataset_t *ds = dn->dn_objset->os_dsl_dataset; SET_BOOKMARK(&dpa->dpa_zb, ds != NULL ? ds->ds_object : DMU_META_OBJSET, dn->dn_object, level, blkid); dpa->dpa_curlevel = curlevel; dpa->dpa_prio = prio; dpa->dpa_aflags = aflags; dpa->dpa_spa = dn->dn_objset->os_spa; dpa->dpa_dnode = dn; dpa->dpa_epbs = epbs; dpa->dpa_zio = pio; dpa->dpa_cb = cb; dpa->dpa_arg = arg; if (!DNODE_LEVEL_IS_CACHEABLE(dn, level)) dpa->dpa_aflags |= ARC_FLAG_UNCACHED; else if (dnode_level_is_l2cacheable(&bp, dn, level)) dpa->dpa_aflags |= ARC_FLAG_L2CACHE; /* * If we have the indirect just above us, no need to do the asynchronous * prefetch chain; we'll just run the last step ourselves. If we're at * a higher level, though, we want to issue the prefetches for all the * indirect blocks asynchronously, so we can go on with whatever we were * doing. */ if (curlevel == level) { ASSERT3U(curblkid, ==, blkid); dbuf_issue_final_prefetch(dpa, &bp); } else { arc_flags_t iter_aflags = ARC_FLAG_NOWAIT; zbookmark_phys_t zb; /* flag if L2ARC eligible, l2arc_noprefetch then decides */ if (dnode_level_is_l2cacheable(&bp, dn, curlevel)) iter_aflags |= ARC_FLAG_L2CACHE; SET_BOOKMARK(&zb, ds != NULL ? ds->ds_object : DMU_META_OBJSET, dn->dn_object, curlevel, curblkid); (void) arc_read(dpa->dpa_zio, dpa->dpa_spa, &bp, dbuf_prefetch_indirect_done, dpa, ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &iter_aflags, &zb); } /* * We use pio here instead of dpa_zio since it's possible that * dpa may have already been freed. */ zio_nowait(pio); return (1); no_issue: if (cb != NULL) cb(arg, level, blkid, B_FALSE); return (0); } int dbuf_prefetch(dnode_t *dn, int64_t level, uint64_t blkid, zio_priority_t prio, arc_flags_t aflags) { return (dbuf_prefetch_impl(dn, level, blkid, prio, aflags, NULL, NULL)); } /* * Helper function for dbuf_hold_impl() to copy a buffer. Handles * the case of encrypted, compressed and uncompressed buffers by * allocating the new buffer, respectively, with arc_alloc_raw_buf(), * arc_alloc_compressed_buf() or arc_alloc_buf().* * * NOTE: Declared noinline to avoid stack bloat in dbuf_hold_impl(). */ noinline static void dbuf_hold_copy(dnode_t *dn, dmu_buf_impl_t *db) { dbuf_dirty_record_t *dr = db->db_data_pending; arc_buf_t *data = dr->dt.dl.dr_data; arc_buf_t *db_data; enum zio_compress compress_type = arc_get_compression(data); uint8_t complevel = arc_get_complevel(data); if (arc_is_encrypted(data)) { boolean_t byteorder; uint8_t salt[ZIO_DATA_SALT_LEN]; uint8_t iv[ZIO_DATA_IV_LEN]; uint8_t mac[ZIO_DATA_MAC_LEN]; arc_get_raw_params(data, &byteorder, salt, iv, mac); db_data = arc_alloc_raw_buf(dn->dn_objset->os_spa, db, dmu_objset_id(dn->dn_objset), byteorder, salt, iv, mac, dn->dn_type, arc_buf_size(data), arc_buf_lsize(data), compress_type, complevel); } else if (compress_type != ZIO_COMPRESS_OFF) { db_data = arc_alloc_compressed_buf( dn->dn_objset->os_spa, db, arc_buf_size(data), arc_buf_lsize(data), compress_type, complevel); } else { db_data = arc_alloc_buf(dn->dn_objset->os_spa, db, DBUF_GET_BUFC_TYPE(db), db->db.db_size); } memcpy(db_data->b_data, data->b_data, arc_buf_size(data)); dbuf_set_data(db, db_data); } /* * Returns with db_holds incremented, and db_mtx not held. * Note: dn_struct_rwlock must be held. */ int dbuf_hold_impl(dnode_t *dn, uint8_t level, uint64_t blkid, boolean_t fail_sparse, boolean_t fail_uncached, const void *tag, dmu_buf_impl_t **dbp) { dmu_buf_impl_t *db, *parent = NULL; uint64_t hv; /* If the pool has been created, verify the tx_sync_lock is not held */ spa_t *spa = dn->dn_objset->os_spa; dsl_pool_t *dp = spa->spa_dsl_pool; if (dp != NULL) { ASSERT(!MUTEX_HELD(&dp->dp_tx.tx_sync_lock)); } ASSERT(blkid != DMU_BONUS_BLKID); ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock)); if (!fail_sparse) ASSERT3U(dn->dn_nlevels, >, level); *dbp = NULL; /* dbuf_find() returns with db_mtx held */ db = dbuf_find(dn->dn_objset, dn->dn_object, level, blkid, &hv); if (db == NULL) { blkptr_t *bp = NULL; int err; if (fail_uncached) return (SET_ERROR(ENOENT)); ASSERT3P(parent, ==, NULL); err = dbuf_findbp(dn, level, blkid, fail_sparse, &parent, &bp); if (fail_sparse) { if (err == 0 && bp && BP_IS_HOLE(bp)) err = SET_ERROR(ENOENT); if (err) { if (parent) dbuf_rele(parent, NULL); return (err); } } if (err && err != ENOENT) return (err); db = dbuf_create(dn, level, blkid, parent, bp, hv); } if (fail_uncached && db->db_state != DB_CACHED) { mutex_exit(&db->db_mtx); return (SET_ERROR(ENOENT)); } if (db->db_buf != NULL) { arc_buf_access(db->db_buf); ASSERT(MUTEX_HELD(&db->db_mtx)); ASSERT3P(db->db.db_data, ==, db->db_buf->b_data); } ASSERT(db->db_buf == NULL || arc_referenced(db->db_buf)); /* * If this buffer is currently syncing out, and we are * still referencing it from db_data, we need to make a copy * of it in case we decide we want to dirty it again in this txg. */ if (db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID && dn->dn_object != DMU_META_DNODE_OBJECT && db->db_state == DB_CACHED && db->db_data_pending) { dbuf_dirty_record_t *dr = db->db_data_pending; if (dr->dt.dl.dr_data == db->db_buf) { ASSERT3P(db->db_buf, !=, NULL); dbuf_hold_copy(dn, db); } } if (multilist_link_active(&db->db_cache_link)) { ASSERT(zfs_refcount_is_zero(&db->db_holds)); ASSERT(db->db_caching_status == DB_DBUF_CACHE || db->db_caching_status == DB_DBUF_METADATA_CACHE); multilist_remove(&dbuf_caches[db->db_caching_status].cache, db); uint64_t size = db->db.db_size; uint64_t usize = dmu_buf_user_size(&db->db); (void) zfs_refcount_remove_many( &dbuf_caches[db->db_caching_status].size, size, db); (void) zfs_refcount_remove_many( &dbuf_caches[db->db_caching_status].size, usize, db->db_user); if (db->db_caching_status == DB_DBUF_METADATA_CACHE) { DBUF_STAT_BUMPDOWN(metadata_cache_count); } else { DBUF_STAT_BUMPDOWN(cache_levels[db->db_level]); DBUF_STAT_BUMPDOWN(cache_count); DBUF_STAT_DECR(cache_levels_bytes[db->db_level], size + usize); } db->db_caching_status = DB_NO_CACHE; } (void) zfs_refcount_add(&db->db_holds, tag); DBUF_VERIFY(db); mutex_exit(&db->db_mtx); /* NOTE: we can't rele the parent until after we drop the db_mtx */ if (parent) dbuf_rele(parent, NULL); ASSERT3P(DB_DNODE(db), ==, dn); ASSERT3U(db->db_blkid, ==, blkid); ASSERT3U(db->db_level, ==, level); *dbp = db; return (0); } dmu_buf_impl_t * dbuf_hold(dnode_t *dn, uint64_t blkid, const void *tag) { return (dbuf_hold_level(dn, 0, blkid, tag)); } dmu_buf_impl_t * dbuf_hold_level(dnode_t *dn, int level, uint64_t blkid, const void *tag) { dmu_buf_impl_t *db; int err = dbuf_hold_impl(dn, level, blkid, FALSE, FALSE, tag, &db); return (err ? NULL : db); } void dbuf_create_bonus(dnode_t *dn) { ASSERT(RW_WRITE_HELD(&dn->dn_struct_rwlock)); ASSERT(dn->dn_bonus == NULL); dn->dn_bonus = dbuf_create(dn, 0, DMU_BONUS_BLKID, dn->dn_dbuf, NULL, dbuf_hash(dn->dn_objset, dn->dn_object, 0, DMU_BONUS_BLKID)); dn->dn_bonus->db_pending_evict = FALSE; } int dbuf_spill_set_blksz(dmu_buf_t *db_fake, uint64_t blksz, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; if (db->db_blkid != DMU_SPILL_BLKID) return (SET_ERROR(ENOTSUP)); if (blksz == 0) blksz = SPA_MINBLOCKSIZE; ASSERT3U(blksz, <=, spa_maxblocksize(dmu_objset_spa(db->db_objset))); blksz = P2ROUNDUP(blksz, SPA_MINBLOCKSIZE); dbuf_new_size(db, blksz, tx); return (0); } void dbuf_rm_spill(dnode_t *dn, dmu_tx_t *tx) { dbuf_free_range(dn, DMU_SPILL_BLKID, DMU_SPILL_BLKID, tx); } #pragma weak dmu_buf_add_ref = dbuf_add_ref void dbuf_add_ref(dmu_buf_impl_t *db, const void *tag) { int64_t holds = zfs_refcount_add(&db->db_holds, tag); VERIFY3S(holds, >, 1); } #pragma weak dmu_buf_try_add_ref = dbuf_try_add_ref boolean_t dbuf_try_add_ref(dmu_buf_t *db_fake, objset_t *os, uint64_t obj, uint64_t blkid, const void *tag) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dmu_buf_impl_t *found_db; boolean_t result = B_FALSE; if (blkid == DMU_BONUS_BLKID) found_db = dbuf_find_bonus(os, obj); else found_db = dbuf_find(os, obj, 0, blkid, NULL); if (found_db != NULL) { if (db == found_db && dbuf_refcount(db) > db->db_dirtycnt) { (void) zfs_refcount_add(&db->db_holds, tag); result = B_TRUE; } mutex_exit(&found_db->db_mtx); } return (result); } /* * If you call dbuf_rele() you had better not be referencing the dnode handle * unless you have some other direct or indirect hold on the dnode. (An indirect * hold is a hold on one of the dnode's dbufs, including the bonus buffer.) * Without that, the dbuf_rele() could lead to a dnode_rele() followed by the * dnode's parent dbuf evicting its dnode handles. */ void dbuf_rele(dmu_buf_impl_t *db, const void *tag) { mutex_enter(&db->db_mtx); dbuf_rele_and_unlock(db, tag, B_FALSE); } void dmu_buf_rele(dmu_buf_t *db, const void *tag) { dbuf_rele((dmu_buf_impl_t *)db, tag); } /* * dbuf_rele() for an already-locked dbuf. This is necessary to allow * db_dirtycnt and db_holds to be updated atomically. The 'evicting' * argument should be set if we are already in the dbuf-evicting code * path, in which case we don't want to recursively evict. This allows us to * avoid deeply nested stacks that would have a call flow similar to this: * * dbuf_rele()-->dbuf_rele_and_unlock()-->dbuf_evict_notify() * ^ | * | | * +-----dbuf_destroy()<--dbuf_evict_one()<--------+ * */ void dbuf_rele_and_unlock(dmu_buf_impl_t *db, const void *tag, boolean_t evicting) { int64_t holds; uint64_t size; ASSERT(MUTEX_HELD(&db->db_mtx)); DBUF_VERIFY(db); /* * Remove the reference to the dbuf before removing its hold on the * dnode so we can guarantee in dnode_move() that a referenced bonus * buffer has a corresponding dnode hold. */ holds = zfs_refcount_remove(&db->db_holds, tag); ASSERT(holds >= 0); /* * We can't freeze indirects if there is a possibility that they * may be modified in the current syncing context. */ if (db->db_buf != NULL && holds == (db->db_level == 0 ? db->db_dirtycnt : 0)) { arc_buf_freeze(db->db_buf); } if (holds == db->db_dirtycnt && db->db_level == 0 && db->db_user_immediate_evict) dbuf_evict_user(db); if (holds == 0) { if (db->db_blkid == DMU_BONUS_BLKID) { dnode_t *dn; boolean_t evict_dbuf = db->db_pending_evict; /* * If the dnode moves here, we cannot cross this * barrier until the move completes. */ DB_DNODE_ENTER(db); dn = DB_DNODE(db); atomic_dec_32(&dn->dn_dbufs_count); /* * Decrementing the dbuf count means that the bonus * buffer's dnode hold is no longer discounted in * dnode_move(). The dnode cannot move until after * the dnode_rele() below. */ DB_DNODE_EXIT(db); /* * Do not reference db after its lock is dropped. * Another thread may evict it. */ mutex_exit(&db->db_mtx); if (evict_dbuf) dnode_evict_bonus(dn); dnode_rele(dn, db); } else if (db->db_buf == NULL) { /* * This is a special case: we never associated this * dbuf with any data allocated from the ARC. */ ASSERT(db->db_state == DB_UNCACHED || db->db_state == DB_NOFILL); dbuf_destroy(db); } else if (arc_released(db->db_buf)) { /* * This dbuf has anonymous data associated with it. */ dbuf_destroy(db); } else if (!db->db_partial_read && !DBUF_IS_CACHEABLE(db)) { /* * We don't expect more accesses to the dbuf, and it * is either not cacheable or was marked for eviction. */ dbuf_destroy(db); } else if (!multilist_link_active(&db->db_cache_link)) { ASSERT3U(db->db_caching_status, ==, DB_NO_CACHE); dbuf_cached_state_t dcs = dbuf_include_in_metadata_cache(db) ? DB_DBUF_METADATA_CACHE : DB_DBUF_CACHE; db->db_caching_status = dcs; multilist_insert(&dbuf_caches[dcs].cache, db); uint64_t db_size = db->db.db_size; uint64_t dbu_size = dmu_buf_user_size(&db->db); (void) zfs_refcount_add_many( &dbuf_caches[dcs].size, db_size, db); size = zfs_refcount_add_many( &dbuf_caches[dcs].size, dbu_size, db->db_user); uint8_t db_level = db->db_level; mutex_exit(&db->db_mtx); if (dcs == DB_DBUF_METADATA_CACHE) { DBUF_STAT_BUMP(metadata_cache_count); DBUF_STAT_MAX(metadata_cache_size_bytes_max, size); } else { DBUF_STAT_BUMP(cache_count); DBUF_STAT_MAX(cache_size_bytes_max, size); DBUF_STAT_BUMP(cache_levels[db_level]); DBUF_STAT_INCR(cache_levels_bytes[db_level], db_size + dbu_size); } if (dcs == DB_DBUF_CACHE && !evicting) dbuf_evict_notify(size); } } else { mutex_exit(&db->db_mtx); } } #pragma weak dmu_buf_refcount = dbuf_refcount uint64_t dbuf_refcount(dmu_buf_impl_t *db) { return (zfs_refcount_count(&db->db_holds)); } uint64_t dmu_buf_user_refcount(dmu_buf_t *db_fake) { uint64_t holds; dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; mutex_enter(&db->db_mtx); ASSERT3U(zfs_refcount_count(&db->db_holds), >=, db->db_dirtycnt); holds = zfs_refcount_count(&db->db_holds) - db->db_dirtycnt; mutex_exit(&db->db_mtx); return (holds); } void * dmu_buf_replace_user(dmu_buf_t *db_fake, dmu_buf_user_t *old_user, dmu_buf_user_t *new_user) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; mutex_enter(&db->db_mtx); dbuf_verify_user(db, DBVU_NOT_EVICTING); if (db->db_user == old_user) db->db_user = new_user; else old_user = db->db_user; dbuf_verify_user(db, DBVU_NOT_EVICTING); mutex_exit(&db->db_mtx); return (old_user); } void * dmu_buf_set_user(dmu_buf_t *db_fake, dmu_buf_user_t *user) { return (dmu_buf_replace_user(db_fake, NULL, user)); } void * dmu_buf_set_user_ie(dmu_buf_t *db_fake, dmu_buf_user_t *user) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; db->db_user_immediate_evict = TRUE; return (dmu_buf_set_user(db_fake, user)); } void * dmu_buf_remove_user(dmu_buf_t *db_fake, dmu_buf_user_t *user) { return (dmu_buf_replace_user(db_fake, user, NULL)); } void * dmu_buf_get_user(dmu_buf_t *db_fake) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dbuf_verify_user(db, DBVU_NOT_EVICTING); return (db->db_user); } uint64_t dmu_buf_user_size(dmu_buf_t *db_fake) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; if (db->db_user == NULL) return (0); return (atomic_load_64(&db->db_user->dbu_size)); } void dmu_buf_add_user_size(dmu_buf_t *db_fake, uint64_t nadd) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; ASSERT3U(db->db_caching_status, ==, DB_NO_CACHE); ASSERT3P(db->db_user, !=, NULL); ASSERT3U(atomic_load_64(&db->db_user->dbu_size), <, UINT64_MAX - nadd); atomic_add_64(&db->db_user->dbu_size, nadd); } void dmu_buf_sub_user_size(dmu_buf_t *db_fake, uint64_t nsub) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; ASSERT3U(db->db_caching_status, ==, DB_NO_CACHE); ASSERT3P(db->db_user, !=, NULL); ASSERT3U(atomic_load_64(&db->db_user->dbu_size), >=, nsub); atomic_sub_64(&db->db_user->dbu_size, nsub); } void dmu_buf_user_evict_wait(void) { taskq_wait(dbu_evict_taskq); } blkptr_t * dmu_buf_get_blkptr(dmu_buf_t *db) { dmu_buf_impl_t *dbi = (dmu_buf_impl_t *)db; return (dbi->db_blkptr); } objset_t * dmu_buf_get_objset(dmu_buf_t *db) { dmu_buf_impl_t *dbi = (dmu_buf_impl_t *)db; return (dbi->db_objset); } static void dbuf_check_blkptr(dnode_t *dn, dmu_buf_impl_t *db) { /* ASSERT(dmu_tx_is_syncing(tx) */ ASSERT(MUTEX_HELD(&db->db_mtx)); if (db->db_blkptr != NULL) return; if (db->db_blkid == DMU_SPILL_BLKID) { db->db_blkptr = DN_SPILL_BLKPTR(dn->dn_phys); BP_ZERO(db->db_blkptr); return; } if (db->db_level == dn->dn_phys->dn_nlevels-1) { /* * This buffer was allocated at a time when there was * no available blkptrs from the dnode, or it was * inappropriate to hook it in (i.e., nlevels mismatch). */ ASSERT(db->db_blkid < dn->dn_phys->dn_nblkptr); ASSERT(db->db_parent == NULL); db->db_parent = dn->dn_dbuf; db->db_blkptr = &dn->dn_phys->dn_blkptr[db->db_blkid]; DBUF_VERIFY(db); } else { dmu_buf_impl_t *parent = db->db_parent; int epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT; ASSERT(dn->dn_phys->dn_nlevels > 1); if (parent == NULL) { mutex_exit(&db->db_mtx); rw_enter(&dn->dn_struct_rwlock, RW_READER); parent = dbuf_hold_level(dn, db->db_level + 1, db->db_blkid >> epbs, db); rw_exit(&dn->dn_struct_rwlock); mutex_enter(&db->db_mtx); db->db_parent = parent; } db->db_blkptr = (blkptr_t *)parent->db.db_data + (db->db_blkid & ((1ULL << epbs) - 1)); DBUF_VERIFY(db); } } static void dbuf_sync_bonus(dbuf_dirty_record_t *dr, dmu_tx_t *tx) { dmu_buf_impl_t *db = dr->dr_dbuf; void *data = dr->dt.dl.dr_data; ASSERT0(db->db_level); ASSERT(MUTEX_HELD(&db->db_mtx)); ASSERT(db->db_blkid == DMU_BONUS_BLKID); ASSERT(data != NULL); dnode_t *dn = dr->dr_dnode; ASSERT3U(DN_MAX_BONUS_LEN(dn->dn_phys), <=, DN_SLOTS_TO_BONUSLEN(dn->dn_phys->dn_extra_slots + 1)); memcpy(DN_BONUS(dn->dn_phys), data, DN_MAX_BONUS_LEN(dn->dn_phys)); dbuf_sync_leaf_verify_bonus_dnode(dr); dbuf_undirty_bonus(dr); dbuf_rele_and_unlock(db, (void *)(uintptr_t)tx->tx_txg, B_FALSE); } /* * When syncing out a blocks of dnodes, adjust the block to deal with * encryption. Normally, we make sure the block is decrypted before writing * it. If we have crypt params, then we are writing a raw (encrypted) block, * from a raw receive. In this case, set the ARC buf's crypt params so * that the BP will be filled with the correct byteorder, salt, iv, and mac. */ static void dbuf_prepare_encrypted_dnode_leaf(dbuf_dirty_record_t *dr) { int err; dmu_buf_impl_t *db = dr->dr_dbuf; ASSERT(MUTEX_HELD(&db->db_mtx)); ASSERT3U(db->db.db_object, ==, DMU_META_DNODE_OBJECT); ASSERT3U(db->db_level, ==, 0); if (!db->db_objset->os_raw_receive && arc_is_encrypted(db->db_buf)) { zbookmark_phys_t zb; /* * Unfortunately, there is currently no mechanism for * syncing context to handle decryption errors. An error * here is only possible if an attacker maliciously * changed a dnode block and updated the associated * checksums going up the block tree. */ SET_BOOKMARK(&zb, dmu_objset_id(db->db_objset), db->db.db_object, db->db_level, db->db_blkid); err = arc_untransform(db->db_buf, db->db_objset->os_spa, &zb, B_TRUE); if (err) panic("Invalid dnode block MAC"); } else if (dr->dt.dl.dr_has_raw_params) { (void) arc_release(dr->dt.dl.dr_data, db); arc_convert_to_raw(dr->dt.dl.dr_data, dmu_objset_id(db->db_objset), dr->dt.dl.dr_byteorder, DMU_OT_DNODE, dr->dt.dl.dr_salt, dr->dt.dl.dr_iv, dr->dt.dl.dr_mac); } } /* * dbuf_sync_indirect() is called recursively from dbuf_sync_list() so it * is critical the we not allow the compiler to inline this function in to * dbuf_sync_list() thereby drastically bloating the stack usage. */ noinline static void dbuf_sync_indirect(dbuf_dirty_record_t *dr, dmu_tx_t *tx) { dmu_buf_impl_t *db = dr->dr_dbuf; dnode_t *dn = dr->dr_dnode; ASSERT(dmu_tx_is_syncing(tx)); dprintf_dbuf_bp(db, db->db_blkptr, "blkptr=%p", db->db_blkptr); mutex_enter(&db->db_mtx); ASSERT(db->db_level > 0); DBUF_VERIFY(db); /* Read the block if it hasn't been read yet. */ if (db->db_buf == NULL) { mutex_exit(&db->db_mtx); (void) dbuf_read(db, NULL, DB_RF_MUST_SUCCEED); mutex_enter(&db->db_mtx); } ASSERT3U(db->db_state, ==, DB_CACHED); ASSERT(db->db_buf != NULL); /* Indirect block size must match what the dnode thinks it is. */ ASSERT3U(db->db.db_size, ==, 1<dn_phys->dn_indblkshift); dbuf_check_blkptr(dn, db); /* Provide the pending dirty record to child dbufs */ db->db_data_pending = dr; mutex_exit(&db->db_mtx); dbuf_write(dr, db->db_buf, tx); zio_t *zio = dr->dr_zio; mutex_enter(&dr->dt.di.dr_mtx); dbuf_sync_list(&dr->dt.di.dr_children, db->db_level - 1, tx); ASSERT(list_head(&dr->dt.di.dr_children) == NULL); mutex_exit(&dr->dt.di.dr_mtx); zio_nowait(zio); } /* * Verify that the size of the data in our bonus buffer does not exceed * its recorded size. * * The purpose of this verification is to catch any cases in development * where the size of a phys structure (i.e space_map_phys_t) grows and, * due to incorrect feature management, older pools expect to read more * data even though they didn't actually write it to begin with. * * For a example, this would catch an error in the feature logic where we * open an older pool and we expect to write the space map histogram of * a space map with size SPACE_MAP_SIZE_V0. */ static void dbuf_sync_leaf_verify_bonus_dnode(dbuf_dirty_record_t *dr) { #ifdef ZFS_DEBUG dnode_t *dn = dr->dr_dnode; /* * Encrypted bonus buffers can have data past their bonuslen. * Skip the verification of these blocks. */ if (DMU_OT_IS_ENCRYPTED(dn->dn_bonustype)) return; uint16_t bonuslen = dn->dn_phys->dn_bonuslen; uint16_t maxbonuslen = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots); ASSERT3U(bonuslen, <=, maxbonuslen); arc_buf_t *datap = dr->dt.dl.dr_data; char *datap_end = ((char *)datap) + bonuslen; char *datap_max = ((char *)datap) + maxbonuslen; /* ensure that everything is zero after our data */ for (; datap_end < datap_max; datap_end++) ASSERT(*datap_end == 0); #endif } static blkptr_t * dbuf_lightweight_bp(dbuf_dirty_record_t *dr) { /* This must be a lightweight dirty record. */ ASSERT3P(dr->dr_dbuf, ==, NULL); dnode_t *dn = dr->dr_dnode; if (dn->dn_phys->dn_nlevels == 1) { VERIFY3U(dr->dt.dll.dr_blkid, <, dn->dn_phys->dn_nblkptr); return (&dn->dn_phys->dn_blkptr[dr->dt.dll.dr_blkid]); } else { dmu_buf_impl_t *parent_db = dr->dr_parent->dr_dbuf; int epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT; VERIFY3U(parent_db->db_level, ==, 1); VERIFY3P(DB_DNODE(parent_db), ==, dn); VERIFY3U(dr->dt.dll.dr_blkid >> epbs, ==, parent_db->db_blkid); blkptr_t *bp = parent_db->db.db_data; return (&bp[dr->dt.dll.dr_blkid & ((1 << epbs) - 1)]); } } static void dbuf_lightweight_ready(zio_t *zio) { dbuf_dirty_record_t *dr = zio->io_private; blkptr_t *bp = zio->io_bp; if (zio->io_error != 0) return; dnode_t *dn = dr->dr_dnode; blkptr_t *bp_orig = dbuf_lightweight_bp(dr); spa_t *spa = dmu_objset_spa(dn->dn_objset); int64_t delta = bp_get_dsize_sync(spa, bp) - bp_get_dsize_sync(spa, bp_orig); dnode_diduse_space(dn, delta); uint64_t blkid = dr->dt.dll.dr_blkid; mutex_enter(&dn->dn_mtx); if (blkid > dn->dn_phys->dn_maxblkid) { ASSERT0(dn->dn_objset->os_raw_receive); dn->dn_phys->dn_maxblkid = blkid; } mutex_exit(&dn->dn_mtx); if (!BP_IS_EMBEDDED(bp)) { uint64_t fill = BP_IS_HOLE(bp) ? 0 : 1; BP_SET_FILL(bp, fill); } dmu_buf_impl_t *parent_db; EQUIV(dr->dr_parent == NULL, dn->dn_phys->dn_nlevels == 1); if (dr->dr_parent == NULL) { parent_db = dn->dn_dbuf; } else { parent_db = dr->dr_parent->dr_dbuf; } rw_enter(&parent_db->db_rwlock, RW_WRITER); *bp_orig = *bp; rw_exit(&parent_db->db_rwlock); } static void dbuf_lightweight_done(zio_t *zio) { dbuf_dirty_record_t *dr = zio->io_private; VERIFY0(zio->io_error); objset_t *os = dr->dr_dnode->dn_objset; dmu_tx_t *tx = os->os_synctx; if (zio->io_flags & (ZIO_FLAG_IO_REWRITE | ZIO_FLAG_NOPWRITE)) { ASSERT(BP_EQUAL(zio->io_bp, &zio->io_bp_orig)); } else { dsl_dataset_t *ds = os->os_dsl_dataset; (void) dsl_dataset_block_kill(ds, &zio->io_bp_orig, tx, B_TRUE); dsl_dataset_block_born(ds, zio->io_bp, tx); } dsl_pool_undirty_space(dmu_objset_pool(os), dr->dr_accounted, zio->io_txg); abd_free(dr->dt.dll.dr_abd); kmem_free(dr, sizeof (*dr)); } noinline static void dbuf_sync_lightweight(dbuf_dirty_record_t *dr, dmu_tx_t *tx) { dnode_t *dn = dr->dr_dnode; zio_t *pio; if (dn->dn_phys->dn_nlevels == 1) { pio = dn->dn_zio; } else { pio = dr->dr_parent->dr_zio; } zbookmark_phys_t zb = { .zb_objset = dmu_objset_id(dn->dn_objset), .zb_object = dn->dn_object, .zb_level = 0, .zb_blkid = dr->dt.dll.dr_blkid, }; /* * See comment in dbuf_write(). This is so that zio->io_bp_orig * will have the old BP in dbuf_lightweight_done(). */ dr->dr_bp_copy = *dbuf_lightweight_bp(dr); dr->dr_zio = zio_write(pio, dmu_objset_spa(dn->dn_objset), dmu_tx_get_txg(tx), &dr->dr_bp_copy, dr->dt.dll.dr_abd, dn->dn_datablksz, abd_get_size(dr->dt.dll.dr_abd), &dr->dt.dll.dr_props, dbuf_lightweight_ready, NULL, dbuf_lightweight_done, dr, ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED | dr->dt.dll.dr_flags, &zb); zio_nowait(dr->dr_zio); } /* * dbuf_sync_leaf() is called recursively from dbuf_sync_list() so it is * critical the we not allow the compiler to inline this function in to * dbuf_sync_list() thereby drastically bloating the stack usage. */ noinline static void dbuf_sync_leaf(dbuf_dirty_record_t *dr, dmu_tx_t *tx) { arc_buf_t **datap = &dr->dt.dl.dr_data; dmu_buf_impl_t *db = dr->dr_dbuf; dnode_t *dn = dr->dr_dnode; objset_t *os; uint64_t txg = tx->tx_txg; ASSERT(dmu_tx_is_syncing(tx)); dprintf_dbuf_bp(db, db->db_blkptr, "blkptr=%p", db->db_blkptr); mutex_enter(&db->db_mtx); /* * To be synced, we must be dirtied. But we might have been freed * after the dirty. */ if (db->db_state == DB_UNCACHED) { /* This buffer has been freed since it was dirtied */ ASSERT3P(db->db.db_data, ==, NULL); } else if (db->db_state == DB_FILL) { /* This buffer was freed and is now being re-filled */ ASSERT(db->db.db_data != dr->dt.dl.dr_data); } else if (db->db_state == DB_READ) { /* * This buffer was either cloned or had a Direct I/O write * occur and has an in-flgiht read on the BP. It is safe to * issue the write here, because the read has already been * issued and the contents won't change. * * We can verify the case of both the clone and Direct I/O * write by making sure the first dirty record for the dbuf * has no ARC buffer associated with it. */ dbuf_dirty_record_t *dr_head = list_head(&db->db_dirty_records); ASSERT3P(db->db_buf, ==, NULL); ASSERT3P(db->db.db_data, ==, NULL); ASSERT3P(dr_head->dt.dl.dr_data, ==, NULL); ASSERT3U(dr_head->dt.dl.dr_override_state, ==, DR_OVERRIDDEN); } else { ASSERT(db->db_state == DB_CACHED || db->db_state == DB_NOFILL); } DBUF_VERIFY(db); if (db->db_blkid == DMU_SPILL_BLKID) { mutex_enter(&dn->dn_mtx); if (!(dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR)) { /* * In the previous transaction group, the bonus buffer * was entirely used to store the attributes for the * dnode which overrode the dn_spill field. However, * when adding more attributes to the file a spill * block was required to hold the extra attributes. * * Make sure to clear the garbage left in the dn_spill * field from the previous attributes in the bonus * buffer. Otherwise, after writing out the spill * block to the new allocated dva, it will free * the old block pointed to by the invalid dn_spill. */ db->db_blkptr = NULL; } dn->dn_phys->dn_flags |= DNODE_FLAG_SPILL_BLKPTR; mutex_exit(&dn->dn_mtx); } /* * If this is a bonus buffer, simply copy the bonus data into the * dnode. It will be written out when the dnode is synced (and it * will be synced, since it must have been dirty for dbuf_sync to * be called). */ if (db->db_blkid == DMU_BONUS_BLKID) { ASSERT(dr->dr_dbuf == db); dbuf_sync_bonus(dr, tx); return; } os = dn->dn_objset; /* * This function may have dropped the db_mtx lock allowing a dmu_sync * operation to sneak in. As a result, we need to ensure that we * don't check the dr_override_state until we have returned from * dbuf_check_blkptr. */ dbuf_check_blkptr(dn, db); /* * If this buffer is in the middle of an immediate write, wait for the * synchronous IO to complete. * * This is also valid even with Direct I/O writes setting a dirty * records override state into DR_IN_DMU_SYNC, because all * Direct I/O writes happen in open-context. */ while (dr->dt.dl.dr_override_state == DR_IN_DMU_SYNC) { ASSERT(dn->dn_object != DMU_META_DNODE_OBJECT); cv_wait(&db->db_changed, &db->db_mtx); } /* * If this is a dnode block, ensure it is appropriately encrypted * or decrypted, depending on what we are writing to it this txg. */ if (os->os_encrypted && dn->dn_object == DMU_META_DNODE_OBJECT) dbuf_prepare_encrypted_dnode_leaf(dr); if (*datap != NULL && *datap == db->db_buf && dn->dn_object != DMU_META_DNODE_OBJECT && zfs_refcount_count(&db->db_holds) > 1) { /* * If this buffer is currently "in use" (i.e., there * are active holds and db_data still references it), * then make a copy before we start the write so that * any modifications from the open txg will not leak * into this write. * * NOTE: this copy does not need to be made for * objects only modified in the syncing context (e.g. * DNONE_DNODE blocks). */ int psize = arc_buf_size(*datap); int lsize = arc_buf_lsize(*datap); arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db); enum zio_compress compress_type = arc_get_compression(*datap); uint8_t complevel = arc_get_complevel(*datap); if (arc_is_encrypted(*datap)) { boolean_t byteorder; uint8_t salt[ZIO_DATA_SALT_LEN]; uint8_t iv[ZIO_DATA_IV_LEN]; uint8_t mac[ZIO_DATA_MAC_LEN]; arc_get_raw_params(*datap, &byteorder, salt, iv, mac); *datap = arc_alloc_raw_buf(os->os_spa, db, dmu_objset_id(os), byteorder, salt, iv, mac, dn->dn_type, psize, lsize, compress_type, complevel); } else if (compress_type != ZIO_COMPRESS_OFF) { ASSERT3U(type, ==, ARC_BUFC_DATA); *datap = arc_alloc_compressed_buf(os->os_spa, db, psize, lsize, compress_type, complevel); } else { *datap = arc_alloc_buf(os->os_spa, db, type, psize); } memcpy((*datap)->b_data, db->db.db_data, psize); } db->db_data_pending = dr; mutex_exit(&db->db_mtx); dbuf_write(dr, *datap, tx); ASSERT(!list_link_active(&dr->dr_dirty_node)); if (dn->dn_object == DMU_META_DNODE_OBJECT) { list_insert_tail(&dn->dn_dirty_records[txg & TXG_MASK], dr); } else { zio_nowait(dr->dr_zio); } } /* * Syncs out a range of dirty records for indirect or leaf dbufs. May be * called recursively from dbuf_sync_indirect(). */ void dbuf_sync_list(list_t *list, int level, dmu_tx_t *tx) { dbuf_dirty_record_t *dr; while ((dr = list_head(list))) { if (dr->dr_zio != NULL) { /* * If we find an already initialized zio then we * are processing the meta-dnode, and we have finished. * The dbufs for all dnodes are put back on the list * during processing, so that we can zio_wait() * these IOs after initiating all child IOs. */ ASSERT3U(dr->dr_dbuf->db.db_object, ==, DMU_META_DNODE_OBJECT); break; } list_remove(list, dr); if (dr->dr_dbuf == NULL) { dbuf_sync_lightweight(dr, tx); } else { if (dr->dr_dbuf->db_blkid != DMU_BONUS_BLKID && dr->dr_dbuf->db_blkid != DMU_SPILL_BLKID) { VERIFY3U(dr->dr_dbuf->db_level, ==, level); } if (dr->dr_dbuf->db_level > 0) dbuf_sync_indirect(dr, tx); else dbuf_sync_leaf(dr, tx); } } } static void dbuf_write_ready(zio_t *zio, arc_buf_t *buf, void *vdb) { (void) buf; dmu_buf_impl_t *db = vdb; dnode_t *dn; blkptr_t *bp = zio->io_bp; blkptr_t *bp_orig = &zio->io_bp_orig; spa_t *spa = zio->io_spa; int64_t delta; uint64_t fill = 0; int i; ASSERT3P(db->db_blkptr, !=, NULL); ASSERT3P(&db->db_data_pending->dr_bp_copy, ==, bp); DB_DNODE_ENTER(db); dn = DB_DNODE(db); delta = bp_get_dsize_sync(spa, bp) - bp_get_dsize_sync(spa, bp_orig); dnode_diduse_space(dn, delta - zio->io_prev_space_delta); zio->io_prev_space_delta = delta; if (BP_GET_BIRTH(bp) != 0) { ASSERT((db->db_blkid != DMU_SPILL_BLKID && BP_GET_TYPE(bp) == dn->dn_type) || (db->db_blkid == DMU_SPILL_BLKID && BP_GET_TYPE(bp) == dn->dn_bonustype) || BP_IS_EMBEDDED(bp)); ASSERT(BP_GET_LEVEL(bp) == db->db_level); } mutex_enter(&db->db_mtx); #ifdef ZFS_DEBUG if (db->db_blkid == DMU_SPILL_BLKID) { ASSERT(dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR); ASSERT(!(BP_IS_HOLE(bp)) && db->db_blkptr == DN_SPILL_BLKPTR(dn->dn_phys)); } #endif if (db->db_level == 0) { mutex_enter(&dn->dn_mtx); if (db->db_blkid > dn->dn_phys->dn_maxblkid && db->db_blkid != DMU_SPILL_BLKID) { ASSERT0(db->db_objset->os_raw_receive); dn->dn_phys->dn_maxblkid = db->db_blkid; } mutex_exit(&dn->dn_mtx); if (dn->dn_type == DMU_OT_DNODE) { i = 0; while (i < db->db.db_size) { dnode_phys_t *dnp = (void *)(((char *)db->db.db_data) + i); i += DNODE_MIN_SIZE; if (dnp->dn_type != DMU_OT_NONE) { fill++; for (int j = 0; j < dnp->dn_nblkptr; j++) { (void) zfs_blkptr_verify(spa, &dnp->dn_blkptr[j], BLK_CONFIG_SKIP, BLK_VERIFY_HALT); } if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) { (void) zfs_blkptr_verify(spa, DN_SPILL_BLKPTR(dnp), BLK_CONFIG_SKIP, BLK_VERIFY_HALT); } i += dnp->dn_extra_slots * DNODE_MIN_SIZE; } } } else { if (BP_IS_HOLE(bp)) { fill = 0; } else { fill = 1; } } } else { blkptr_t *ibp = db->db.db_data; ASSERT3U(db->db.db_size, ==, 1<dn_phys->dn_indblkshift); for (i = db->db.db_size >> SPA_BLKPTRSHIFT; i > 0; i--, ibp++) { if (BP_IS_HOLE(ibp)) continue; (void) zfs_blkptr_verify(spa, ibp, BLK_CONFIG_SKIP, BLK_VERIFY_HALT); fill += BP_GET_FILL(ibp); } } DB_DNODE_EXIT(db); if (!BP_IS_EMBEDDED(bp)) BP_SET_FILL(bp, fill); mutex_exit(&db->db_mtx); db_lock_type_t dblt = dmu_buf_lock_parent(db, RW_WRITER, FTAG); *db->db_blkptr = *bp; dmu_buf_unlock_parent(db, dblt, FTAG); } /* * This function gets called just prior to running through the compression * stage of the zio pipeline. If we're an indirect block comprised of only * holes, then we want this indirect to be compressed away to a hole. In * order to do that we must zero out any information about the holes that * this indirect points to prior to before we try to compress it. */ static void dbuf_write_children_ready(zio_t *zio, arc_buf_t *buf, void *vdb) { (void) zio, (void) buf; dmu_buf_impl_t *db = vdb; blkptr_t *bp; unsigned int epbs, i; ASSERT3U(db->db_level, >, 0); DB_DNODE_ENTER(db); epbs = DB_DNODE(db)->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT; DB_DNODE_EXIT(db); ASSERT3U(epbs, <, 31); /* Determine if all our children are holes */ for (i = 0, bp = db->db.db_data; i < 1ULL << epbs; i++, bp++) { if (!BP_IS_HOLE(bp)) break; } /* * If all the children are holes, then zero them all out so that * we may get compressed away. */ if (i == 1ULL << epbs) { /* * We only found holes. Grab the rwlock to prevent * anybody from reading the blocks we're about to * zero out. */ rw_enter(&db->db_rwlock, RW_WRITER); memset(db->db.db_data, 0, db->db.db_size); rw_exit(&db->db_rwlock); } } static void dbuf_write_done(zio_t *zio, arc_buf_t *buf, void *vdb) { (void) buf; dmu_buf_impl_t *db = vdb; blkptr_t *bp_orig = &zio->io_bp_orig; blkptr_t *bp = db->db_blkptr; objset_t *os = db->db_objset; dmu_tx_t *tx = os->os_synctx; ASSERT0(zio->io_error); ASSERT(db->db_blkptr == bp); /* * For nopwrites and rewrites we ensure that the bp matches our * original and bypass all the accounting. */ if (zio->io_flags & (ZIO_FLAG_IO_REWRITE | ZIO_FLAG_NOPWRITE)) { ASSERT(BP_EQUAL(bp, bp_orig)); } else { dsl_dataset_t *ds = os->os_dsl_dataset; (void) dsl_dataset_block_kill(ds, bp_orig, tx, B_TRUE); dsl_dataset_block_born(ds, bp, tx); } mutex_enter(&db->db_mtx); DBUF_VERIFY(db); dbuf_dirty_record_t *dr = db->db_data_pending; dnode_t *dn = dr->dr_dnode; ASSERT(!list_link_active(&dr->dr_dirty_node)); ASSERT(dr->dr_dbuf == db); ASSERT(list_next(&db->db_dirty_records, dr) == NULL); list_remove(&db->db_dirty_records, dr); #ifdef ZFS_DEBUG if (db->db_blkid == DMU_SPILL_BLKID) { ASSERT(dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR); ASSERT(!(BP_IS_HOLE(db->db_blkptr)) && db->db_blkptr == DN_SPILL_BLKPTR(dn->dn_phys)); } #endif if (db->db_level == 0) { ASSERT(db->db_blkid != DMU_BONUS_BLKID); ASSERT(dr->dt.dl.dr_override_state == DR_NOT_OVERRIDDEN); /* no dr_data if this is a NO_FILL or Direct I/O */ if (dr->dt.dl.dr_data != NULL && dr->dt.dl.dr_data != db->db_buf) { ASSERT3B(dr->dt.dl.dr_brtwrite, ==, B_FALSE); ASSERT3B(dr->dt.dl.dr_diowrite, ==, B_FALSE); arc_buf_destroy(dr->dt.dl.dr_data, db); } } else { ASSERT(list_head(&dr->dt.di.dr_children) == NULL); ASSERT3U(db->db.db_size, ==, 1 << dn->dn_phys->dn_indblkshift); if (!BP_IS_HOLE(db->db_blkptr)) { int epbs __maybe_unused = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT; ASSERT3U(db->db_blkid, <=, dn->dn_phys->dn_maxblkid >> (db->db_level * epbs)); ASSERT3U(BP_GET_LSIZE(db->db_blkptr), ==, db->db.db_size); } mutex_destroy(&dr->dt.di.dr_mtx); list_destroy(&dr->dt.di.dr_children); } cv_broadcast(&db->db_changed); ASSERT(db->db_dirtycnt > 0); db->db_dirtycnt -= 1; db->db_data_pending = NULL; dbuf_rele_and_unlock(db, (void *)(uintptr_t)tx->tx_txg, B_FALSE); dsl_pool_undirty_space(dmu_objset_pool(os), dr->dr_accounted, zio->io_txg); kmem_cache_free(dbuf_dirty_kmem_cache, dr); } static void dbuf_write_nofill_ready(zio_t *zio) { dbuf_write_ready(zio, NULL, zio->io_private); } static void dbuf_write_nofill_done(zio_t *zio) { dbuf_write_done(zio, NULL, zio->io_private); } static void dbuf_write_override_ready(zio_t *zio) { dbuf_dirty_record_t *dr = zio->io_private; dmu_buf_impl_t *db = dr->dr_dbuf; dbuf_write_ready(zio, NULL, db); } static void dbuf_write_override_done(zio_t *zio) { dbuf_dirty_record_t *dr = zio->io_private; dmu_buf_impl_t *db = dr->dr_dbuf; blkptr_t *obp = &dr->dt.dl.dr_overridden_by; mutex_enter(&db->db_mtx); if (!BP_EQUAL(zio->io_bp, obp)) { if (!BP_IS_HOLE(obp)) dsl_free(spa_get_dsl(zio->io_spa), zio->io_txg, obp); arc_release(dr->dt.dl.dr_data, db); } mutex_exit(&db->db_mtx); dbuf_write_done(zio, NULL, db); if (zio->io_abd != NULL) abd_free(zio->io_abd); } typedef struct dbuf_remap_impl_callback_arg { objset_t *drica_os; uint64_t drica_blk_birth; dmu_tx_t *drica_tx; } dbuf_remap_impl_callback_arg_t; static void dbuf_remap_impl_callback(uint64_t vdev, uint64_t offset, uint64_t size, void *arg) { dbuf_remap_impl_callback_arg_t *drica = arg; objset_t *os = drica->drica_os; spa_t *spa = dmu_objset_spa(os); dmu_tx_t *tx = drica->drica_tx; ASSERT(dsl_pool_sync_context(spa_get_dsl(spa))); if (os == spa_meta_objset(spa)) { spa_vdev_indirect_mark_obsolete(spa, vdev, offset, size, tx); } else { dsl_dataset_block_remapped(dmu_objset_ds(os), vdev, offset, size, drica->drica_blk_birth, tx); } } static void dbuf_remap_impl(dnode_t *dn, blkptr_t *bp, krwlock_t *rw, dmu_tx_t *tx) { blkptr_t bp_copy = *bp; spa_t *spa = dmu_objset_spa(dn->dn_objset); dbuf_remap_impl_callback_arg_t drica; ASSERT(dsl_pool_sync_context(spa_get_dsl(spa))); drica.drica_os = dn->dn_objset; drica.drica_blk_birth = BP_GET_BIRTH(bp); drica.drica_tx = tx; if (spa_remap_blkptr(spa, &bp_copy, dbuf_remap_impl_callback, &drica)) { /* * If the blkptr being remapped is tracked by a livelist, * then we need to make sure the livelist reflects the update. * First, cancel out the old blkptr by appending a 'FREE' * entry. Next, add an 'ALLOC' to track the new version. This * way we avoid trying to free an inaccurate blkptr at delete. * Note that embedded blkptrs are not tracked in livelists. */ if (dn->dn_objset != spa_meta_objset(spa)) { dsl_dataset_t *ds = dmu_objset_ds(dn->dn_objset); if (dsl_deadlist_is_open(&ds->ds_dir->dd_livelist) && BP_GET_BIRTH(bp) > ds->ds_dir->dd_origin_txg) { ASSERT(!BP_IS_EMBEDDED(bp)); ASSERT(dsl_dir_is_clone(ds->ds_dir)); ASSERT(spa_feature_is_enabled(spa, SPA_FEATURE_LIVELIST)); bplist_append(&ds->ds_dir->dd_pending_frees, bp); bplist_append(&ds->ds_dir->dd_pending_allocs, &bp_copy); } } /* * The db_rwlock prevents dbuf_read_impl() from * dereferencing the BP while we are changing it. To * avoid lock contention, only grab it when we are actually * changing the BP. */ if (rw != NULL) rw_enter(rw, RW_WRITER); *bp = bp_copy; if (rw != NULL) rw_exit(rw); } } /* * Remap any existing BP's to concrete vdevs, if possible. */ static void dbuf_remap(dnode_t *dn, dmu_buf_impl_t *db, dmu_tx_t *tx) { spa_t *spa = dmu_objset_spa(db->db_objset); ASSERT(dsl_pool_sync_context(spa_get_dsl(spa))); if (!spa_feature_is_active(spa, SPA_FEATURE_DEVICE_REMOVAL)) return; if (db->db_level > 0) { blkptr_t *bp = db->db.db_data; for (int i = 0; i < db->db.db_size >> SPA_BLKPTRSHIFT; i++) { dbuf_remap_impl(dn, &bp[i], &db->db_rwlock, tx); } } else if (db->db.db_object == DMU_META_DNODE_OBJECT) { dnode_phys_t *dnp = db->db.db_data; ASSERT3U(dn->dn_type, ==, DMU_OT_DNODE); for (int i = 0; i < db->db.db_size >> DNODE_SHIFT; i += dnp[i].dn_extra_slots + 1) { for (int j = 0; j < dnp[i].dn_nblkptr; j++) { krwlock_t *lock = (dn->dn_dbuf == NULL ? NULL : &dn->dn_dbuf->db_rwlock); dbuf_remap_impl(dn, &dnp[i].dn_blkptr[j], lock, tx); } } } } /* * Populate dr->dr_zio with a zio to commit a dirty buffer to disk. * Caller is responsible for issuing the zio_[no]wait(dr->dr_zio). */ static void dbuf_write(dbuf_dirty_record_t *dr, arc_buf_t *data, dmu_tx_t *tx) { dmu_buf_impl_t *db = dr->dr_dbuf; dnode_t *dn = dr->dr_dnode; objset_t *os; dmu_buf_impl_t *parent = db->db_parent; uint64_t txg = tx->tx_txg; zbookmark_phys_t zb; zio_prop_t zp; zio_t *pio; /* parent I/O */ int wp_flag = 0; ASSERT(dmu_tx_is_syncing(tx)); os = dn->dn_objset; if (db->db_level > 0 || dn->dn_type == DMU_OT_DNODE) { /* * Private object buffers are released here rather than in * dbuf_dirty() since they are only modified in the syncing * context and we don't want the overhead of making multiple * copies of the data. */ if (BP_IS_HOLE(db->db_blkptr)) arc_buf_thaw(data); else dbuf_release_bp(db); dbuf_remap(dn, db, tx); } if (parent != dn->dn_dbuf) { /* Our parent is an indirect block. */ /* We have a dirty parent that has been scheduled for write. */ ASSERT(parent && parent->db_data_pending); /* Our parent's buffer is one level closer to the dnode. */ ASSERT(db->db_level == parent->db_level-1); /* * We're about to modify our parent's db_data by modifying * our block pointer, so the parent must be released. */ ASSERT(arc_released(parent->db_buf)); pio = parent->db_data_pending->dr_zio; } else { /* Our parent is the dnode itself. */ ASSERT((db->db_level == dn->dn_phys->dn_nlevels-1 && db->db_blkid != DMU_SPILL_BLKID) || (db->db_blkid == DMU_SPILL_BLKID && db->db_level == 0)); if (db->db_blkid != DMU_SPILL_BLKID) ASSERT3P(db->db_blkptr, ==, &dn->dn_phys->dn_blkptr[db->db_blkid]); pio = dn->dn_zio; } ASSERT(db->db_level == 0 || data == db->db_buf); ASSERT3U(BP_GET_BIRTH(db->db_blkptr), <=, txg); ASSERT(pio); SET_BOOKMARK(&zb, os->os_dsl_dataset ? os->os_dsl_dataset->ds_object : DMU_META_OBJSET, db->db.db_object, db->db_level, db->db_blkid); if (db->db_blkid == DMU_SPILL_BLKID) wp_flag = WP_SPILL; wp_flag |= (data == NULL) ? WP_NOFILL : 0; dmu_write_policy(os, dn, db->db_level, wp_flag, &zp); + /* + * Set rewrite properties for zfs_rewrite() operations. + */ + if (db->db_level == 0 && dr->dt.dl.dr_rewrite) { + zp.zp_rewrite = B_TRUE; + + /* + * Mark physical rewrite feature for activation. + * This will be activated automatically during dataset sync. + */ + dsl_dataset_t *ds = os->os_dsl_dataset; + if (!dsl_dataset_feature_is_active(ds, + SPA_FEATURE_PHYSICAL_REWRITE)) { + ds->ds_feature_activation[ + SPA_FEATURE_PHYSICAL_REWRITE] = (void *)B_TRUE; + } + } + /* * We copy the blkptr now (rather than when we instantiate the dirty * record), because its value can change between open context and * syncing context. We do not need to hold dn_struct_rwlock to read * db_blkptr because we are in syncing context. */ dr->dr_bp_copy = *db->db_blkptr; if (db->db_level == 0 && dr->dt.dl.dr_override_state == DR_OVERRIDDEN) { /* * The BP for this block has been provided by open context * (by dmu_sync(), dmu_write_direct(), * or dmu_buf_write_embedded()). */ abd_t *contents = (data != NULL) ? abd_get_from_buf(data->b_data, arc_buf_size(data)) : NULL; dr->dr_zio = zio_write(pio, os->os_spa, txg, &dr->dr_bp_copy, contents, db->db.db_size, db->db.db_size, &zp, dbuf_write_override_ready, NULL, dbuf_write_override_done, dr, ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED, &zb); mutex_enter(&db->db_mtx); dr->dt.dl.dr_override_state = DR_NOT_OVERRIDDEN; zio_write_override(dr->dr_zio, &dr->dt.dl.dr_overridden_by, dr->dt.dl.dr_copies, dr->dt.dl.dr_gang_copies, dr->dt.dl.dr_nopwrite, dr->dt.dl.dr_brtwrite); mutex_exit(&db->db_mtx); } else if (data == NULL) { ASSERT(zp.zp_checksum == ZIO_CHECKSUM_OFF || zp.zp_checksum == ZIO_CHECKSUM_NOPARITY); dr->dr_zio = zio_write(pio, os->os_spa, txg, &dr->dr_bp_copy, NULL, db->db.db_size, db->db.db_size, &zp, dbuf_write_nofill_ready, NULL, dbuf_write_nofill_done, db, ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED | ZIO_FLAG_NODATA, &zb); } else { ASSERT(arc_released(data)); /* * For indirect blocks, we want to setup the children * ready callback so that we can properly handle an indirect * block that only contains holes. */ arc_write_done_func_t *children_ready_cb = NULL; if (db->db_level != 0) children_ready_cb = dbuf_write_children_ready; dr->dr_zio = arc_write(pio, os->os_spa, txg, &dr->dr_bp_copy, data, !DBUF_IS_CACHEABLE(db), dbuf_is_l2cacheable(db, NULL), &zp, dbuf_write_ready, children_ready_cb, dbuf_write_done, db, ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED, &zb); } } EXPORT_SYMBOL(dbuf_find); EXPORT_SYMBOL(dbuf_is_metadata); EXPORT_SYMBOL(dbuf_destroy); EXPORT_SYMBOL(dbuf_whichblock); EXPORT_SYMBOL(dbuf_read); EXPORT_SYMBOL(dbuf_unoverride); EXPORT_SYMBOL(dbuf_free_range); EXPORT_SYMBOL(dbuf_new_size); EXPORT_SYMBOL(dbuf_release_bp); EXPORT_SYMBOL(dbuf_dirty); EXPORT_SYMBOL(dmu_buf_set_crypt_params); EXPORT_SYMBOL(dmu_buf_will_dirty); +EXPORT_SYMBOL(dmu_buf_will_rewrite); EXPORT_SYMBOL(dmu_buf_is_dirty); EXPORT_SYMBOL(dmu_buf_will_clone_or_dio); EXPORT_SYMBOL(dmu_buf_will_not_fill); EXPORT_SYMBOL(dmu_buf_will_fill); EXPORT_SYMBOL(dmu_buf_fill_done); EXPORT_SYMBOL(dmu_buf_rele); EXPORT_SYMBOL(dbuf_assign_arcbuf); EXPORT_SYMBOL(dbuf_prefetch); EXPORT_SYMBOL(dbuf_hold_impl); EXPORT_SYMBOL(dbuf_hold); EXPORT_SYMBOL(dbuf_hold_level); EXPORT_SYMBOL(dbuf_create_bonus); EXPORT_SYMBOL(dbuf_spill_set_blksz); EXPORT_SYMBOL(dbuf_rm_spill); EXPORT_SYMBOL(dbuf_add_ref); EXPORT_SYMBOL(dbuf_rele); EXPORT_SYMBOL(dbuf_rele_and_unlock); EXPORT_SYMBOL(dbuf_refcount); EXPORT_SYMBOL(dbuf_sync_list); EXPORT_SYMBOL(dmu_buf_set_user); EXPORT_SYMBOL(dmu_buf_set_user_ie); EXPORT_SYMBOL(dmu_buf_get_user); EXPORT_SYMBOL(dmu_buf_get_blkptr); ZFS_MODULE_PARAM(zfs_dbuf_cache, dbuf_cache_, max_bytes, U64, ZMOD_RW, "Maximum size in bytes of the dbuf cache."); ZFS_MODULE_PARAM(zfs_dbuf_cache, dbuf_cache_, hiwater_pct, UINT, ZMOD_RW, "Percentage over dbuf_cache_max_bytes for direct dbuf eviction."); ZFS_MODULE_PARAM(zfs_dbuf_cache, dbuf_cache_, lowater_pct, UINT, ZMOD_RW, "Percentage below dbuf_cache_max_bytes when dbuf eviction stops."); ZFS_MODULE_PARAM(zfs_dbuf, dbuf_, metadata_cache_max_bytes, U64, ZMOD_RW, "Maximum size in bytes of dbuf metadata cache."); ZFS_MODULE_PARAM(zfs_dbuf, dbuf_, cache_shift, UINT, ZMOD_RW, "Set size of dbuf cache to log2 fraction of arc size."); ZFS_MODULE_PARAM(zfs_dbuf, dbuf_, metadata_cache_shift, UINT, ZMOD_RW, "Set size of dbuf metadata cache to log2 fraction of arc size."); ZFS_MODULE_PARAM(zfs_dbuf, dbuf_, mutex_cache_shift, UINT, ZMOD_RD, "Set size of dbuf cache mutex array as log2 shift."); diff --git a/module/zfs/dmu.c b/module/zfs/dmu.c index 690227a30938..296e58ef9cd8 100644 --- a/module/zfs/dmu.c +++ b/module/zfs/dmu.c @@ -1,2989 +1,2990 @@ // SPDX-License-Identifier: CDDL-1.0 /* * 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 https://opensource.org/licenses/CDDL-1.0. * 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) 2013 by Saso Kiselkov. All rights reserved. * Copyright (c) 2013, Joyent, Inc. All rights reserved. * Copyright (c) 2016, Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2015 by Chunwei Chen. All rights reserved. * Copyright (c) 2019 Datto Inc. * Copyright (c) 2019, 2023, Klara Inc. * Copyright (c) 2019, Allan Jude * Copyright (c) 2022 Hewlett Packard Enterprise Development LP. * Copyright (c) 2021, 2022 by Pawel Jakub Dawidek */ #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 #include #endif /* * Enable/disable nopwrite feature. */ static int zfs_nopwrite_enabled = 1; /* * Tunable to control percentage of dirtied L1 blocks from frees allowed into * one TXG. After this threshold is crossed, additional dirty blocks from frees * will wait until the next TXG. * A value of zero will disable this throttle. */ static uint_t zfs_per_txg_dirty_frees_percent = 30; /* * Enable/disable forcing txg sync when dirty checking for holes with lseek(). * By default this is enabled to ensure accurate hole reporting, it can result * in a significant performance penalty for lseek(SEEK_HOLE) heavy workloads. * Disabling this option will result in holes never being reported in dirty * files which is always safe. */ static int zfs_dmu_offset_next_sync = 1; /* * Limit the amount we can prefetch with one call to this amount. This * helps to limit the amount of memory that can be used by prefetching. * Larger objects should be prefetched a bit at a time. */ #ifdef _ILP32 uint_t dmu_prefetch_max = 8 * 1024 * 1024; #else uint_t dmu_prefetch_max = 8 * SPA_MAXBLOCKSIZE; #endif /* * Override copies= for dedup state objects. 0 means the traditional behaviour * (ie the default for the containing objset ie 3 for the MOS). */ uint_t dmu_ddt_copies = 0; const dmu_object_type_info_t dmu_ot[DMU_OT_NUMTYPES] = { {DMU_BSWAP_UINT8, TRUE, FALSE, FALSE, "unallocated" }, {DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "object directory" }, {DMU_BSWAP_UINT64, TRUE, TRUE, FALSE, "object array" }, {DMU_BSWAP_UINT8, TRUE, FALSE, FALSE, "packed nvlist" }, {DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "packed nvlist size" }, {DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "bpobj" }, {DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "bpobj header" }, {DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "SPA space map header" }, {DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "SPA space map" }, {DMU_BSWAP_UINT64, TRUE, FALSE, TRUE, "ZIL intent log" }, {DMU_BSWAP_DNODE, TRUE, FALSE, TRUE, "DMU dnode" }, {DMU_BSWAP_OBJSET, TRUE, TRUE, FALSE, "DMU objset" }, {DMU_BSWAP_UINT64, TRUE, TRUE, FALSE, "DSL directory" }, {DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "DSL directory child map"}, {DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "DSL dataset snap map" }, {DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "DSL props" }, {DMU_BSWAP_UINT64, TRUE, TRUE, FALSE, "DSL dataset" }, {DMU_BSWAP_ZNODE, TRUE, FALSE, FALSE, "ZFS znode" }, {DMU_BSWAP_OLDACL, TRUE, FALSE, TRUE, "ZFS V0 ACL" }, {DMU_BSWAP_UINT8, FALSE, FALSE, TRUE, "ZFS plain file" }, {DMU_BSWAP_ZAP, TRUE, FALSE, TRUE, "ZFS directory" }, {DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "ZFS master node" }, {DMU_BSWAP_ZAP, TRUE, FALSE, TRUE, "ZFS delete queue" }, {DMU_BSWAP_UINT8, FALSE, FALSE, TRUE, "zvol object" }, {DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "zvol prop" }, {DMU_BSWAP_UINT8, FALSE, FALSE, TRUE, "other uint8[]" }, {DMU_BSWAP_UINT64, FALSE, FALSE, TRUE, "other uint64[]" }, {DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "other ZAP" }, {DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "persistent error log" }, {DMU_BSWAP_UINT8, TRUE, FALSE, FALSE, "SPA history" }, {DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "SPA history offsets" }, {DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "Pool properties" }, {DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "DSL permissions" }, {DMU_BSWAP_ACL, TRUE, FALSE, TRUE, "ZFS ACL" }, {DMU_BSWAP_UINT8, TRUE, FALSE, TRUE, "ZFS SYSACL" }, {DMU_BSWAP_UINT8, TRUE, FALSE, TRUE, "FUID table" }, {DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "FUID table size" }, {DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "DSL dataset next clones"}, {DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "scan work queue" }, {DMU_BSWAP_ZAP, TRUE, FALSE, TRUE, "ZFS user/group/project used" }, {DMU_BSWAP_ZAP, TRUE, FALSE, TRUE, "ZFS user/group/project quota"}, {DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "snapshot refcount tags"}, {DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "DDT ZAP algorithm" }, {DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "DDT statistics" }, {DMU_BSWAP_UINT8, TRUE, FALSE, TRUE, "System attributes" }, {DMU_BSWAP_ZAP, TRUE, FALSE, TRUE, "SA master node" }, {DMU_BSWAP_ZAP, TRUE, FALSE, TRUE, "SA attr registration" }, {DMU_BSWAP_ZAP, TRUE, FALSE, TRUE, "SA attr layouts" }, {DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "scan translations" }, {DMU_BSWAP_UINT8, FALSE, FALSE, TRUE, "deduplicated block" }, {DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "DSL deadlist map" }, {DMU_BSWAP_UINT64, TRUE, TRUE, FALSE, "DSL deadlist map hdr" }, {DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "DSL dir clones" }, {DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "bpobj subobj" } }; dmu_object_byteswap_info_t dmu_ot_byteswap[DMU_BSWAP_NUMFUNCS] = { { byteswap_uint8_array, "uint8" }, { byteswap_uint16_array, "uint16" }, { byteswap_uint32_array, "uint32" }, { byteswap_uint64_array, "uint64" }, { zap_byteswap, "zap" }, { dnode_buf_byteswap, "dnode" }, { dmu_objset_byteswap, "objset" }, { zfs_znode_byteswap, "znode" }, { zfs_oldacl_byteswap, "oldacl" }, { zfs_acl_byteswap, "acl" } }; int dmu_buf_hold_noread_by_dnode(dnode_t *dn, uint64_t offset, const void *tag, dmu_buf_t **dbp) { uint64_t blkid; dmu_buf_impl_t *db; rw_enter(&dn->dn_struct_rwlock, RW_READER); blkid = dbuf_whichblock(dn, 0, offset); db = dbuf_hold(dn, blkid, tag); rw_exit(&dn->dn_struct_rwlock); if (db == NULL) { *dbp = NULL; return (SET_ERROR(EIO)); } *dbp = &db->db; return (0); } int dmu_buf_hold_noread(objset_t *os, uint64_t object, uint64_t offset, const void *tag, dmu_buf_t **dbp) { dnode_t *dn; uint64_t blkid; dmu_buf_impl_t *db; int err; err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); rw_enter(&dn->dn_struct_rwlock, RW_READER); blkid = dbuf_whichblock(dn, 0, offset); db = dbuf_hold(dn, blkid, tag); rw_exit(&dn->dn_struct_rwlock); dnode_rele(dn, FTAG); if (db == NULL) { *dbp = NULL; return (SET_ERROR(EIO)); } *dbp = &db->db; return (err); } int dmu_buf_hold_by_dnode(dnode_t *dn, uint64_t offset, const void *tag, dmu_buf_t **dbp, dmu_flags_t flags) { int err; err = dmu_buf_hold_noread_by_dnode(dn, offset, tag, dbp); if (err == 0) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)(*dbp); err = dbuf_read(db, NULL, flags | DB_RF_CANFAIL); if (err != 0) { dbuf_rele(db, tag); *dbp = NULL; } } return (err); } int dmu_buf_hold(objset_t *os, uint64_t object, uint64_t offset, const void *tag, dmu_buf_t **dbp, dmu_flags_t flags) { int err; err = dmu_buf_hold_noread(os, object, offset, tag, dbp); if (err == 0) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)(*dbp); err = dbuf_read(db, NULL, flags | DB_RF_CANFAIL); if (err != 0) { dbuf_rele(db, tag); *dbp = NULL; } } return (err); } int dmu_bonus_max(void) { return (DN_OLD_MAX_BONUSLEN); } int dmu_set_bonus(dmu_buf_t *db_fake, int newsize, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dnode_t *dn; int error; if (newsize < 0 || newsize > db_fake->db_size) return (SET_ERROR(EINVAL)); DB_DNODE_ENTER(db); dn = DB_DNODE(db); if (dn->dn_bonus != db) { error = SET_ERROR(EINVAL); } else { dnode_setbonuslen(dn, newsize, tx); error = 0; } DB_DNODE_EXIT(db); return (error); } int dmu_set_bonustype(dmu_buf_t *db_fake, dmu_object_type_t type, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dnode_t *dn; int error; if (!DMU_OT_IS_VALID(type)) return (SET_ERROR(EINVAL)); DB_DNODE_ENTER(db); dn = DB_DNODE(db); if (dn->dn_bonus != db) { error = SET_ERROR(EINVAL); } else { dnode_setbonus_type(dn, type, tx); error = 0; } DB_DNODE_EXIT(db); return (error); } dmu_object_type_t dmu_get_bonustype(dmu_buf_t *db_fake) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dmu_object_type_t type; DB_DNODE_ENTER(db); type = DB_DNODE(db)->dn_bonustype; DB_DNODE_EXIT(db); return (type); } int dmu_rm_spill(objset_t *os, uint64_t object, dmu_tx_t *tx) { dnode_t *dn; int error; error = dnode_hold(os, object, FTAG, &dn); dbuf_rm_spill(dn, tx); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); dnode_rm_spill(dn, tx); rw_exit(&dn->dn_struct_rwlock); dnode_rele(dn, FTAG); return (error); } /* * Lookup and hold the bonus buffer for the provided dnode. If the dnode * has not yet been allocated a new bonus dbuf a will be allocated. * Returns ENOENT, EIO, or 0. */ int dmu_bonus_hold_by_dnode(dnode_t *dn, const void *tag, dmu_buf_t **dbp, dmu_flags_t flags) { dmu_buf_impl_t *db; int error; rw_enter(&dn->dn_struct_rwlock, RW_READER); if (dn->dn_bonus == NULL) { if (!rw_tryupgrade(&dn->dn_struct_rwlock)) { rw_exit(&dn->dn_struct_rwlock); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); } if (dn->dn_bonus == NULL) dbuf_create_bonus(dn); } db = dn->dn_bonus; /* as long as the bonus buf is held, the dnode will be held */ if (zfs_refcount_add(&db->db_holds, tag) == 1) { VERIFY(dnode_add_ref(dn, db)); atomic_inc_32(&dn->dn_dbufs_count); } /* * Wait to drop dn_struct_rwlock until after adding the bonus dbuf's * hold and incrementing the dbuf count to ensure that dnode_move() sees * a dnode hold for every dbuf. */ rw_exit(&dn->dn_struct_rwlock); error = dbuf_read(db, NULL, flags | DB_RF_CANFAIL); if (error) { dnode_evict_bonus(dn); dbuf_rele(db, tag); *dbp = NULL; return (error); } *dbp = &db->db; return (0); } int dmu_bonus_hold(objset_t *os, uint64_t object, const void *tag, dmu_buf_t **dbp) { dnode_t *dn; int error; error = dnode_hold(os, object, FTAG, &dn); if (error) return (error); error = dmu_bonus_hold_by_dnode(dn, tag, dbp, DMU_READ_NO_PREFETCH); dnode_rele(dn, FTAG); return (error); } /* * returns ENOENT, EIO, or 0. * * This interface will allocate a blank spill dbuf when a spill blk * doesn't already exist on the dnode. * * if you only want to find an already existing spill db, then * dmu_spill_hold_existing() should be used. */ int dmu_spill_hold_by_dnode(dnode_t *dn, dmu_flags_t flags, const void *tag, dmu_buf_t **dbp) { dmu_buf_impl_t *db = NULL; int err; if ((flags & DB_RF_HAVESTRUCT) == 0) rw_enter(&dn->dn_struct_rwlock, RW_READER); db = dbuf_hold(dn, DMU_SPILL_BLKID, tag); if ((flags & DB_RF_HAVESTRUCT) == 0) rw_exit(&dn->dn_struct_rwlock); if (db == NULL) { *dbp = NULL; return (SET_ERROR(EIO)); } err = dbuf_read(db, NULL, flags); if (err == 0) *dbp = &db->db; else { dbuf_rele(db, tag); *dbp = NULL; } return (err); } int dmu_spill_hold_existing(dmu_buf_t *bonus, const void *tag, dmu_buf_t **dbp) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)bonus; dnode_t *dn; int err; DB_DNODE_ENTER(db); dn = DB_DNODE(db); if (spa_version(dn->dn_objset->os_spa) < SPA_VERSION_SA) { err = SET_ERROR(EINVAL); } else { rw_enter(&dn->dn_struct_rwlock, RW_READER); if (!dn->dn_have_spill) { err = SET_ERROR(ENOENT); } else { err = dmu_spill_hold_by_dnode(dn, DB_RF_HAVESTRUCT | DB_RF_CANFAIL, tag, dbp); } rw_exit(&dn->dn_struct_rwlock); } DB_DNODE_EXIT(db); return (err); } int dmu_spill_hold_by_bonus(dmu_buf_t *bonus, dmu_flags_t flags, const void *tag, dmu_buf_t **dbp) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)bonus; int err; DB_DNODE_ENTER(db); err = dmu_spill_hold_by_dnode(DB_DNODE(db), flags, tag, dbp); DB_DNODE_EXIT(db); return (err); } /* * Note: longer-term, we should modify all of the dmu_buf_*() interfaces * to take a held dnode rather than -- the lookup is wasteful, * and can induce severe lock contention when writing to several files * whose dnodes are in the same block. */ int dmu_buf_hold_array_by_dnode(dnode_t *dn, uint64_t offset, uint64_t length, boolean_t read, const void *tag, int *numbufsp, dmu_buf_t ***dbpp, dmu_flags_t flags) { dmu_buf_t **dbp; zstream_t *zs = NULL; uint64_t blkid, nblks, i; dmu_flags_t dbuf_flags; int err; zio_t *zio = NULL; boolean_t missed = B_FALSE; ASSERT(!read || length <= DMU_MAX_ACCESS); /* * Note: We directly notify the prefetch code of this read, so that * we can tell it about the multi-block read. dbuf_read() only knows * about the one block it is accessing. */ dbuf_flags = (flags & ~DMU_READ_PREFETCH) | DMU_READ_NO_PREFETCH | DB_RF_CANFAIL | DB_RF_NEVERWAIT | DB_RF_HAVESTRUCT; rw_enter(&dn->dn_struct_rwlock, RW_READER); if (dn->dn_datablkshift) { int blkshift = dn->dn_datablkshift; nblks = (P2ROUNDUP(offset + length, 1ULL << blkshift) - P2ALIGN_TYPED(offset, 1ULL << blkshift, uint64_t)) >> blkshift; } else { if (offset + length > dn->dn_datablksz) { zfs_panic_recover("zfs: accessing past end of object " "%llx/%llx (size=%u access=%llu+%llu)", (longlong_t)dn->dn_objset-> os_dsl_dataset->ds_object, (longlong_t)dn->dn_object, dn->dn_datablksz, (longlong_t)offset, (longlong_t)length); rw_exit(&dn->dn_struct_rwlock); return (SET_ERROR(EIO)); } nblks = 1; } dbp = kmem_zalloc(sizeof (dmu_buf_t *) * nblks, KM_SLEEP); if (read) zio = zio_root(dn->dn_objset->os_spa, NULL, NULL, ZIO_FLAG_CANFAIL); blkid = dbuf_whichblock(dn, 0, offset); if ((flags & DMU_READ_NO_PREFETCH) == 0) { /* * Prepare the zfetch before initiating the demand reads, so * that if multiple threads block on same indirect block, we * base predictions on the original less racy request order. */ zs = dmu_zfetch_prepare(&dn->dn_zfetch, blkid, nblks, read && !(flags & DMU_DIRECTIO), B_TRUE); } for (i = 0; i < nblks; i++) { dmu_buf_impl_t *db = dbuf_hold(dn, blkid + i, tag); if (db == NULL) { if (zs) { dmu_zfetch_run(&dn->dn_zfetch, zs, missed, B_TRUE, (flags & DMU_UNCACHEDIO)); } rw_exit(&dn->dn_struct_rwlock); dmu_buf_rele_array(dbp, nblks, tag); if (read) zio_nowait(zio); return (SET_ERROR(EIO)); } /* * Initiate async demand data read. * We check the db_state after calling dbuf_read() because * (1) dbuf_read() may change the state to CACHED due to a * hit in the ARC, and (2) on a cache miss, a child will * have been added to "zio" but not yet completed, so the * state will not yet be CACHED. */ if (read) { if (i == nblks - 1 && blkid + i < dn->dn_maxblkid && offset + length < db->db.db_offset + db->db.db_size) { if (offset <= db->db.db_offset) dbuf_flags |= DMU_PARTIAL_FIRST; else dbuf_flags |= DMU_PARTIAL_MORE; } (void) dbuf_read(db, zio, dbuf_flags); if (db->db_state != DB_CACHED) missed = B_TRUE; } dbp[i] = &db->db; } /* * If we are doing O_DIRECT we still hold the dbufs, even for reads, * but we do not issue any reads here. We do not want to account for * writes in this case. * * O_DIRECT write/read accounting takes place in * dmu_{write/read}_abd(). */ if (!read && ((flags & DMU_DIRECTIO) == 0)) zfs_racct_write(dn->dn_objset->os_spa, length, nblks, flags); if (zs) { dmu_zfetch_run(&dn->dn_zfetch, zs, missed, B_TRUE, (flags & DMU_UNCACHEDIO)); } rw_exit(&dn->dn_struct_rwlock); if (read) { /* wait for async read i/o */ err = zio_wait(zio); if (err) { dmu_buf_rele_array(dbp, nblks, tag); return (err); } /* wait for other io to complete */ for (i = 0; i < nblks; i++) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)dbp[i]; mutex_enter(&db->db_mtx); while (db->db_state == DB_READ || db->db_state == DB_FILL) cv_wait(&db->db_changed, &db->db_mtx); if (db->db_state == DB_UNCACHED) err = SET_ERROR(EIO); mutex_exit(&db->db_mtx); if (err) { dmu_buf_rele_array(dbp, nblks, tag); return (err); } } } *numbufsp = nblks; *dbpp = dbp; return (0); } int dmu_buf_hold_array(objset_t *os, uint64_t object, uint64_t offset, uint64_t length, int read, const void *tag, int *numbufsp, dmu_buf_t ***dbpp) { dnode_t *dn; int err; err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); err = dmu_buf_hold_array_by_dnode(dn, offset, length, read, tag, numbufsp, dbpp, DMU_READ_PREFETCH); dnode_rele(dn, FTAG); return (err); } int dmu_buf_hold_array_by_bonus(dmu_buf_t *db_fake, uint64_t offset, uint64_t length, boolean_t read, const void *tag, int *numbufsp, dmu_buf_t ***dbpp) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; int err; DB_DNODE_ENTER(db); err = dmu_buf_hold_array_by_dnode(DB_DNODE(db), offset, length, read, tag, numbufsp, dbpp, DMU_READ_PREFETCH); DB_DNODE_EXIT(db); return (err); } void dmu_buf_rele_array(dmu_buf_t **dbp_fake, int numbufs, const void *tag) { int i; dmu_buf_impl_t **dbp = (dmu_buf_impl_t **)dbp_fake; if (numbufs == 0) return; for (i = 0; i < numbufs; i++) { if (dbp[i]) dbuf_rele(dbp[i], tag); } kmem_free(dbp, sizeof (dmu_buf_t *) * numbufs); } /* * Issue prefetch I/Os for the given blocks. If level is greater than 0, the * indirect blocks prefetched will be those that point to the blocks containing * the data starting at offset, and continuing to offset + len. If the range * is too long, prefetch the first dmu_prefetch_max bytes as requested, while * for the rest only a higher level, also fitting within dmu_prefetch_max. It * should primarily help random reads, since for long sequential reads there is * a speculative prefetcher. * * Note that if the indirect blocks above the blocks being prefetched are not * in cache, they will be asynchronously read in. Dnode read by dnode_hold() * is currently synchronous. */ void dmu_prefetch(objset_t *os, uint64_t object, int64_t level, uint64_t offset, uint64_t len, zio_priority_t pri) { dnode_t *dn; if (dmu_prefetch_max == 0 || len == 0) { dmu_prefetch_dnode(os, object, pri); return; } if (dnode_hold(os, object, FTAG, &dn) != 0) return; dmu_prefetch_by_dnode(dn, level, offset, len, pri); dnode_rele(dn, FTAG); } void dmu_prefetch_by_dnode(dnode_t *dn, int64_t level, uint64_t offset, uint64_t len, zio_priority_t pri) { int64_t level2 = level; uint64_t start, end, start2, end2; /* * Depending on len we may do two prefetches: blocks [start, end) at * level, and following blocks [start2, end2) at higher level2. */ rw_enter(&dn->dn_struct_rwlock, RW_READER); if (dn->dn_datablkshift != 0) { /* * Limit prefetch to present blocks. */ uint64_t size = (dn->dn_maxblkid + 1) << dn->dn_datablkshift; if (offset >= size) { rw_exit(&dn->dn_struct_rwlock); return; } if (offset + len < offset || offset + len > size) len = size - offset; /* * The object has multiple blocks. Calculate the full range * of blocks [start, end2) and then split it into two parts, * so that the first [start, end) fits into dmu_prefetch_max. */ start = dbuf_whichblock(dn, level, offset); end2 = dbuf_whichblock(dn, level, offset + len - 1) + 1; uint8_t ibs = dn->dn_indblkshift; uint8_t bs = (level == 0) ? dn->dn_datablkshift : ibs; uint_t limit = P2ROUNDUP(dmu_prefetch_max, 1 << bs) >> bs; start2 = end = MIN(end2, start + limit); /* * Find level2 where [start2, end2) fits into dmu_prefetch_max. */ uint8_t ibps = ibs - SPA_BLKPTRSHIFT; limit = P2ROUNDUP(dmu_prefetch_max, 1 << ibs) >> ibs; do { level2++; start2 = P2ROUNDUP(start2, 1 << ibps) >> ibps; end2 = P2ROUNDUP(end2, 1 << ibps) >> ibps; } while (end2 - start2 > limit); } else { /* There is only one block. Prefetch it or nothing. */ start = start2 = end2 = 0; end = start + (level == 0 && offset < dn->dn_datablksz); } for (uint64_t i = start; i < end; i++) dbuf_prefetch(dn, level, i, pri, 0); for (uint64_t i = start2; i < end2; i++) dbuf_prefetch(dn, level2, i, pri, 0); rw_exit(&dn->dn_struct_rwlock); } typedef struct { kmutex_t dpa_lock; kcondvar_t dpa_cv; uint64_t dpa_pending_io; } dmu_prefetch_arg_t; static void dmu_prefetch_done(void *arg, uint64_t level, uint64_t blkid, boolean_t issued) { (void) level; (void) blkid; (void)issued; dmu_prefetch_arg_t *dpa = arg; ASSERT0(level); mutex_enter(&dpa->dpa_lock); ASSERT3U(dpa->dpa_pending_io, >, 0); if (--dpa->dpa_pending_io == 0) cv_broadcast(&dpa->dpa_cv); mutex_exit(&dpa->dpa_lock); } static void dmu_prefetch_wait_by_dnode(dnode_t *dn, uint64_t offset, uint64_t len) { dmu_prefetch_arg_t dpa; mutex_init(&dpa.dpa_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&dpa.dpa_cv, NULL, CV_DEFAULT, NULL); rw_enter(&dn->dn_struct_rwlock, RW_READER); uint64_t start = dbuf_whichblock(dn, 0, offset); uint64_t end = dbuf_whichblock(dn, 0, offset + len - 1) + 1; dpa.dpa_pending_io = end - start; for (uint64_t blk = start; blk < end; blk++) { (void) dbuf_prefetch_impl(dn, 0, blk, ZIO_PRIORITY_ASYNC_READ, 0, dmu_prefetch_done, &dpa); } rw_exit(&dn->dn_struct_rwlock); /* wait for prefetch L0 reads to finish */ mutex_enter(&dpa.dpa_lock); while (dpa.dpa_pending_io > 0) { cv_wait(&dpa.dpa_cv, &dpa.dpa_lock); } mutex_exit(&dpa.dpa_lock); mutex_destroy(&dpa.dpa_lock); cv_destroy(&dpa.dpa_cv); } /* * Issue prefetch I/Os for the given L0 block range and wait for the I/O * to complete. This does not enforce dmu_prefetch_max and will prefetch * the entire range. The blocks are read from disk into the ARC but no * decompression occurs (i.e., the dbuf cache is not required). */ int dmu_prefetch_wait(objset_t *os, uint64_t object, uint64_t offset, uint64_t size) { dnode_t *dn; int err = 0; err = dnode_hold(os, object, FTAG, &dn); if (err != 0) return (err); /* * Chunk the requests (16 indirects worth) so that we can be interrupted */ uint64_t chunksize; if (dn->dn_indblkshift) { uint64_t nbps = bp_span_in_blocks(dn->dn_indblkshift, 1); chunksize = (nbps * 16) << dn->dn_datablkshift; } else { chunksize = dn->dn_datablksz; } while (size > 0) { uint64_t mylen = MIN(size, chunksize); dmu_prefetch_wait_by_dnode(dn, offset, mylen); offset += mylen; size -= mylen; if (issig()) { err = SET_ERROR(EINTR); break; } } dnode_rele(dn, FTAG); return (err); } /* * Issue prefetch I/Os for the given object's dnode. */ void dmu_prefetch_dnode(objset_t *os, uint64_t object, zio_priority_t pri) { if (object == 0 || object >= DN_MAX_OBJECT) return; dnode_t *dn = DMU_META_DNODE(os); rw_enter(&dn->dn_struct_rwlock, RW_READER); uint64_t blkid = dbuf_whichblock(dn, 0, object * sizeof (dnode_phys_t)); dbuf_prefetch(dn, 0, blkid, pri, 0); rw_exit(&dn->dn_struct_rwlock); } /* * Get the next "chunk" of file data to free. We traverse the file from * the end so that the file gets shorter over time (if we crash in the * middle, this will leave us in a better state). We find allocated file * data by simply searching the allocated level 1 indirects. * * On input, *start should be the first offset that does not need to be * freed (e.g. "offset + length"). On return, *start will be the first * offset that should be freed and l1blks is set to the number of level 1 * indirect blocks found within the chunk. */ static int get_next_chunk(dnode_t *dn, uint64_t *start, uint64_t minimum, uint64_t *l1blks) { uint64_t blks; uint64_t maxblks = DMU_MAX_ACCESS >> (dn->dn_indblkshift + 1); /* bytes of data covered by a level-1 indirect block */ uint64_t iblkrange = (uint64_t)dn->dn_datablksz * EPB(dn->dn_indblkshift, SPA_BLKPTRSHIFT); ASSERT3U(minimum, <=, *start); /* dn_nlevels == 1 means we don't have any L1 blocks */ if (dn->dn_nlevels <= 1) { *l1blks = 0; *start = minimum; return (0); } /* * Check if we can free the entire range assuming that all of the * L1 blocks in this range have data. If we can, we use this * worst case value as an estimate so we can avoid having to look * at the object's actual data. */ uint64_t total_l1blks = (roundup(*start, iblkrange) - (minimum / iblkrange * iblkrange)) / iblkrange; if (total_l1blks <= maxblks) { *l1blks = total_l1blks; *start = minimum; return (0); } ASSERT(ISP2(iblkrange)); for (blks = 0; *start > minimum && blks < maxblks; blks++) { int err; /* * dnode_next_offset(BACKWARDS) will find an allocated L1 * indirect block at or before the input offset. We must * decrement *start so that it is at the end of the region * to search. */ (*start)--; err = dnode_next_offset(dn, DNODE_FIND_BACKWARDS, start, 2, 1, 0); /* if there are no indirect blocks before start, we are done */ if (err == ESRCH) { *start = minimum; break; } else if (err != 0) { *l1blks = blks; return (err); } /* set start to the beginning of this L1 indirect */ *start = P2ALIGN_TYPED(*start, iblkrange, uint64_t); } if (*start < minimum) *start = minimum; *l1blks = blks; return (0); } /* * If this objset is of type OST_ZFS return true if vfs's unmounted flag is set, * otherwise return false. * Used below in dmu_free_long_range_impl() to enable abort when unmounting */ static boolean_t dmu_objset_zfs_unmounting(objset_t *os) { #ifdef _KERNEL if (dmu_objset_type(os) == DMU_OST_ZFS) return (zfs_get_vfs_flag_unmounted(os)); #else (void) os; #endif return (B_FALSE); } static int dmu_free_long_range_impl(objset_t *os, dnode_t *dn, uint64_t offset, uint64_t length) { uint64_t object_size; int err; uint64_t dirty_frees_threshold; dsl_pool_t *dp = dmu_objset_pool(os); if (dn == NULL) return (SET_ERROR(EINVAL)); object_size = (dn->dn_maxblkid + 1) * dn->dn_datablksz; if (offset >= object_size) return (0); if (zfs_per_txg_dirty_frees_percent <= 100) dirty_frees_threshold = zfs_per_txg_dirty_frees_percent * zfs_dirty_data_max / 100; else dirty_frees_threshold = zfs_dirty_data_max / 20; if (length == DMU_OBJECT_END || offset + length > object_size) length = object_size - offset; while (length != 0) { uint64_t chunk_end, chunk_begin, chunk_len; uint64_t l1blks; dmu_tx_t *tx; if (dmu_objset_zfs_unmounting(dn->dn_objset)) return (SET_ERROR(EINTR)); chunk_end = chunk_begin = offset + length; /* move chunk_begin backwards to the beginning of this chunk */ err = get_next_chunk(dn, &chunk_begin, offset, &l1blks); if (err) return (err); ASSERT3U(chunk_begin, >=, offset); ASSERT3U(chunk_begin, <=, chunk_end); chunk_len = chunk_end - chunk_begin; tx = dmu_tx_create(os); dmu_tx_hold_free(tx, dn->dn_object, chunk_begin, chunk_len); /* * Mark this transaction as typically resulting in a net * reduction in space used. */ dmu_tx_mark_netfree(tx); err = dmu_tx_assign(tx, DMU_TX_WAIT); if (err) { dmu_tx_abort(tx); return (err); } uint64_t txg = dmu_tx_get_txg(tx); mutex_enter(&dp->dp_lock); uint64_t long_free_dirty = dp->dp_long_free_dirty_pertxg[txg & TXG_MASK]; mutex_exit(&dp->dp_lock); /* * To avoid filling up a TXG with just frees, wait for * the next TXG to open before freeing more chunks if * we have reached the threshold of frees. */ if (dirty_frees_threshold != 0 && long_free_dirty >= dirty_frees_threshold) { DMU_TX_STAT_BUMP(dmu_tx_dirty_frees_delay); dmu_tx_commit(tx); txg_wait_open(dp, 0, B_TRUE); continue; } /* * In order to prevent unnecessary write throttling, for each * TXG, we track the cumulative size of L1 blocks being dirtied * in dnode_free_range() below. We compare this number to a * tunable threshold, past which we prevent new L1 dirty freeing * blocks from being added into the open TXG. See * dmu_free_long_range_impl() for details. The threshold * prevents write throttle activation due to dirty freeing L1 * blocks taking up a large percentage of zfs_dirty_data_max. */ mutex_enter(&dp->dp_lock); dp->dp_long_free_dirty_pertxg[txg & TXG_MASK] += l1blks << dn->dn_indblkshift; mutex_exit(&dp->dp_lock); DTRACE_PROBE3(free__long__range, uint64_t, long_free_dirty, uint64_t, chunk_len, uint64_t, txg); dnode_free_range(dn, chunk_begin, chunk_len, tx); dmu_tx_commit(tx); length -= chunk_len; } return (0); } int dmu_free_long_range(objset_t *os, uint64_t object, uint64_t offset, uint64_t length) { dnode_t *dn; int err; err = dnode_hold(os, object, FTAG, &dn); if (err != 0) return (err); err = dmu_free_long_range_impl(os, dn, offset, length); /* * It is important to zero out the maxblkid when freeing the entire * file, so that (a) subsequent calls to dmu_free_long_range_impl() * will take the fast path, and (b) dnode_reallocate() can verify * that the entire file has been freed. */ if (err == 0 && offset == 0 && length == DMU_OBJECT_END) dn->dn_maxblkid = 0; dnode_rele(dn, FTAG); return (err); } int dmu_free_long_object(objset_t *os, uint64_t object) { dmu_tx_t *tx; int err; err = dmu_free_long_range(os, object, 0, DMU_OBJECT_END); if (err != 0) return (err); tx = dmu_tx_create(os); dmu_tx_hold_bonus(tx, object); dmu_tx_hold_free(tx, object, 0, DMU_OBJECT_END); dmu_tx_mark_netfree(tx); err = dmu_tx_assign(tx, DMU_TX_WAIT); if (err == 0) { err = dmu_object_free(os, object, tx); dmu_tx_commit(tx); } else { dmu_tx_abort(tx); } return (err); } int dmu_free_range(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, dmu_tx_t *tx) { dnode_t *dn; int err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); ASSERT(offset < UINT64_MAX); ASSERT(size == DMU_OBJECT_END || size <= UINT64_MAX - offset); dnode_free_range(dn, offset, size, tx); dnode_rele(dn, FTAG); return (0); } static int dmu_read_impl(dnode_t *dn, uint64_t offset, uint64_t size, void *buf, dmu_flags_t flags) { dmu_buf_t **dbp; int numbufs, err = 0; /* * Deal with odd block sizes, where there can't be data past the first * block. If we ever do the tail block optimization, we will need to * handle that here as well. */ if (dn->dn_maxblkid == 0) { uint64_t newsz = offset > dn->dn_datablksz ? 0 : MIN(size, dn->dn_datablksz - offset); memset((char *)buf + newsz, 0, size - newsz); size = newsz; } if (size == 0) return (0); /* Allow Direct I/O when requested and properly aligned */ if ((flags & DMU_DIRECTIO) && zfs_dio_page_aligned(buf) && zfs_dio_aligned(offset, size, PAGESIZE)) { abd_t *data = abd_get_from_buf(buf, size); err = dmu_read_abd(dn, offset, size, data, flags); abd_free(data); return (err); } flags &= ~DMU_DIRECTIO; while (size > 0) { uint64_t mylen = MIN(size, DMU_MAX_ACCESS / 2); int i; /* * NB: we could do this block-at-a-time, but it's nice * to be reading in parallel. */ err = dmu_buf_hold_array_by_dnode(dn, offset, mylen, TRUE, FTAG, &numbufs, &dbp, flags); if (err) break; for (i = 0; i < numbufs; i++) { uint64_t tocpy; int64_t bufoff; dmu_buf_t *db = dbp[i]; ASSERT(size > 0); bufoff = offset - db->db_offset; tocpy = MIN(db->db_size - bufoff, size); ASSERT(db->db_data != NULL); (void) memcpy(buf, (char *)db->db_data + bufoff, tocpy); offset += tocpy; size -= tocpy; buf = (char *)buf + tocpy; } dmu_buf_rele_array(dbp, numbufs, FTAG); } return (err); } int dmu_read(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, void *buf, dmu_flags_t flags) { dnode_t *dn; int err; err = dnode_hold(os, object, FTAG, &dn); if (err != 0) return (err); err = dmu_read_impl(dn, offset, size, buf, flags); dnode_rele(dn, FTAG); return (err); } int dmu_read_by_dnode(dnode_t *dn, uint64_t offset, uint64_t size, void *buf, dmu_flags_t flags) { return (dmu_read_impl(dn, offset, size, buf, flags)); } static void dmu_write_impl(dmu_buf_t **dbp, int numbufs, uint64_t offset, uint64_t size, const void *buf, dmu_tx_t *tx, dmu_flags_t flags) { int i; for (i = 0; i < numbufs; i++) { uint64_t tocpy; int64_t bufoff; dmu_buf_t *db = dbp[i]; ASSERT(size > 0); bufoff = offset - db->db_offset; tocpy = MIN(db->db_size - bufoff, size); ASSERT(i == 0 || i == numbufs-1 || tocpy == db->db_size); if (tocpy == db->db_size) { dmu_buf_will_fill_flags(db, tx, B_FALSE, flags); } else { if (i == numbufs - 1 && bufoff + tocpy < db->db_size) { if (bufoff == 0) flags |= DMU_PARTIAL_FIRST; else flags |= DMU_PARTIAL_MORE; } dmu_buf_will_dirty_flags(db, tx, flags); } ASSERT(db->db_data != NULL); (void) memcpy((char *)db->db_data + bufoff, buf, tocpy); if (tocpy == db->db_size) dmu_buf_fill_done(db, tx, B_FALSE); offset += tocpy; size -= tocpy; buf = (char *)buf + tocpy; } } void dmu_write(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, const void *buf, dmu_tx_t *tx) { dmu_buf_t **dbp; int numbufs; if (size == 0) return; VERIFY0(dmu_buf_hold_array(os, object, offset, size, FALSE, FTAG, &numbufs, &dbp)); dmu_write_impl(dbp, numbufs, offset, size, buf, tx, DMU_READ_PREFETCH); dmu_buf_rele_array(dbp, numbufs, FTAG); } int dmu_write_by_dnode(dnode_t *dn, uint64_t offset, uint64_t size, const void *buf, dmu_tx_t *tx, dmu_flags_t flags) { dmu_buf_t **dbp; int numbufs; int error; if (size == 0) return (0); /* Allow Direct I/O when requested and properly aligned */ if ((flags & DMU_DIRECTIO) && zfs_dio_page_aligned((void *)buf) && zfs_dio_aligned(offset, size, dn->dn_datablksz)) { abd_t *data = abd_get_from_buf((void *)buf, size); error = dmu_write_abd(dn, offset, size, data, flags, tx); abd_free(data); return (error); } flags &= ~DMU_DIRECTIO; VERIFY0(dmu_buf_hold_array_by_dnode(dn, offset, size, FALSE, FTAG, &numbufs, &dbp, flags)); dmu_write_impl(dbp, numbufs, offset, size, buf, tx, flags); dmu_buf_rele_array(dbp, numbufs, FTAG); return (0); } void dmu_prealloc(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, dmu_tx_t *tx) { dmu_buf_t **dbp; int numbufs, i; if (size == 0) return; VERIFY(0 == dmu_buf_hold_array(os, object, offset, size, FALSE, FTAG, &numbufs, &dbp)); for (i = 0; i < numbufs; i++) { dmu_buf_t *db = dbp[i]; dmu_buf_will_not_fill(db, tx); } dmu_buf_rele_array(dbp, numbufs, FTAG); } void dmu_write_embedded(objset_t *os, uint64_t object, uint64_t offset, void *data, uint8_t etype, uint8_t comp, int uncompressed_size, int compressed_size, int byteorder, dmu_tx_t *tx) { dmu_buf_t *db; ASSERT3U(etype, <, NUM_BP_EMBEDDED_TYPES); ASSERT3U(comp, <, ZIO_COMPRESS_FUNCTIONS); VERIFY0(dmu_buf_hold_noread(os, object, offset, FTAG, &db)); dmu_buf_write_embedded(db, data, (bp_embedded_type_t)etype, (enum zio_compress)comp, uncompressed_size, compressed_size, byteorder, tx); dmu_buf_rele(db, FTAG); } void dmu_redact(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, dmu_tx_t *tx) { int numbufs, i; dmu_buf_t **dbp; VERIFY0(dmu_buf_hold_array(os, object, offset, size, FALSE, FTAG, &numbufs, &dbp)); for (i = 0; i < numbufs; i++) dmu_buf_redact(dbp[i], tx); dmu_buf_rele_array(dbp, numbufs, FTAG); } #ifdef _KERNEL int dmu_read_uio_dnode(dnode_t *dn, zfs_uio_t *uio, uint64_t size, dmu_flags_t flags) { dmu_buf_t **dbp; int numbufs, i, err; if ((flags & DMU_DIRECTIO) && (uio->uio_extflg & UIO_DIRECT)) return (dmu_read_uio_direct(dn, uio, size, flags)); flags &= ~DMU_DIRECTIO; /* * NB: we could do this block-at-a-time, but it's nice * to be reading in parallel. */ err = dmu_buf_hold_array_by_dnode(dn, zfs_uio_offset(uio), size, TRUE, FTAG, &numbufs, &dbp, flags); if (err) return (err); for (i = 0; i < numbufs; i++) { uint64_t tocpy; int64_t bufoff; dmu_buf_t *db = dbp[i]; ASSERT(size > 0); bufoff = zfs_uio_offset(uio) - db->db_offset; tocpy = MIN(db->db_size - bufoff, size); ASSERT(db->db_data != NULL); err = zfs_uio_fault_move((char *)db->db_data + bufoff, tocpy, UIO_READ, uio); if (err) break; size -= tocpy; } dmu_buf_rele_array(dbp, numbufs, FTAG); return (err); } /* * Read 'size' bytes into the uio buffer. * From object zdb->db_object. * Starting at zfs_uio_offset(uio). * * If the caller already has a dbuf in the target object * (e.g. its bonus buffer), this routine is faster than dmu_read_uio(), * because we don't have to find the dnode_t for the object. */ int dmu_read_uio_dbuf(dmu_buf_t *zdb, zfs_uio_t *uio, uint64_t size, dmu_flags_t flags) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)zdb; int err; if (size == 0) return (0); DB_DNODE_ENTER(db); err = dmu_read_uio_dnode(DB_DNODE(db), uio, size, flags); DB_DNODE_EXIT(db); return (err); } /* * Read 'size' bytes into the uio buffer. * From the specified object * Starting at offset zfs_uio_offset(uio). */ int dmu_read_uio(objset_t *os, uint64_t object, zfs_uio_t *uio, uint64_t size, dmu_flags_t flags) { dnode_t *dn; int err; if (size == 0) return (0); err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); err = dmu_read_uio_dnode(dn, uio, size, flags); dnode_rele(dn, FTAG); return (err); } int dmu_write_uio_dnode(dnode_t *dn, zfs_uio_t *uio, uint64_t size, dmu_tx_t *tx, dmu_flags_t flags) { dmu_buf_t **dbp; int numbufs; int err = 0; uint64_t write_size; dmu_flags_t oflags = flags; top: write_size = size; /* * We only allow Direct I/O writes to happen if we are block * sized aligned. Otherwise, we pass the write off to the ARC. */ if ((flags & DMU_DIRECTIO) && (uio->uio_extflg & UIO_DIRECT) && (write_size >= dn->dn_datablksz)) { if (zfs_dio_aligned(zfs_uio_offset(uio), write_size, dn->dn_datablksz)) { return (dmu_write_uio_direct(dn, uio, size, flags, tx)); } else if (write_size > dn->dn_datablksz && zfs_dio_offset_aligned(zfs_uio_offset(uio), dn->dn_datablksz)) { write_size = dn->dn_datablksz * (write_size / dn->dn_datablksz); err = dmu_write_uio_direct(dn, uio, write_size, flags, tx); if (err == 0) { size -= write_size; goto top; } else { return (err); } } else { write_size = P2PHASE(zfs_uio_offset(uio), dn->dn_datablksz); } } flags &= ~DMU_DIRECTIO; err = dmu_buf_hold_array_by_dnode(dn, zfs_uio_offset(uio), write_size, FALSE, FTAG, &numbufs, &dbp, flags); if (err) return (err); for (int i = 0; i < numbufs; i++) { uint64_t tocpy; int64_t bufoff; dmu_buf_t *db = dbp[i]; ASSERT(write_size > 0); offset_t off = zfs_uio_offset(uio); bufoff = off - db->db_offset; tocpy = MIN(db->db_size - bufoff, write_size); ASSERT(i == 0 || i == numbufs-1 || tocpy == db->db_size); if (tocpy == db->db_size) { dmu_buf_will_fill_flags(db, tx, B_TRUE, flags); } else { if (i == numbufs - 1 && bufoff + tocpy < db->db_size) { if (bufoff == 0) flags |= DMU_PARTIAL_FIRST; else flags |= DMU_PARTIAL_MORE; } dmu_buf_will_dirty_flags(db, tx, flags); } ASSERT(db->db_data != NULL); err = zfs_uio_fault_move((char *)db->db_data + bufoff, tocpy, UIO_WRITE, uio); if (tocpy == db->db_size && dmu_buf_fill_done(db, tx, err)) { /* The fill was reverted. Undo any uio progress. */ zfs_uio_advance(uio, off - zfs_uio_offset(uio)); } if (err) break; write_size -= tocpy; size -= tocpy; } IMPLY(err == 0, write_size == 0); dmu_buf_rele_array(dbp, numbufs, FTAG); if ((oflags & DMU_DIRECTIO) && (uio->uio_extflg & UIO_DIRECT) && err == 0 && size > 0) { flags = oflags; goto top; } IMPLY(err == 0, size == 0); return (err); } /* * Write 'size' bytes from the uio buffer. * To object zdb->db_object. * Starting at offset zfs_uio_offset(uio). * * If the caller already has a dbuf in the target object * (e.g. its bonus buffer), this routine is faster than dmu_write_uio(), * because we don't have to find the dnode_t for the object. */ int dmu_write_uio_dbuf(dmu_buf_t *zdb, zfs_uio_t *uio, uint64_t size, dmu_tx_t *tx, dmu_flags_t flags) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)zdb; int err; if (size == 0) return (0); DB_DNODE_ENTER(db); err = dmu_write_uio_dnode(DB_DNODE(db), uio, size, tx, flags); DB_DNODE_EXIT(db); return (err); } /* * Write 'size' bytes from the uio buffer. * To the specified object. * Starting at offset zfs_uio_offset(uio). */ int dmu_write_uio(objset_t *os, uint64_t object, zfs_uio_t *uio, uint64_t size, dmu_tx_t *tx, dmu_flags_t flags) { dnode_t *dn; int err; if (size == 0) return (0); err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); err = dmu_write_uio_dnode(dn, uio, size, tx, flags); dnode_rele(dn, FTAG); return (err); } #endif /* _KERNEL */ static void dmu_cached_bps(spa_t *spa, blkptr_t *bps, uint_t nbps, uint64_t *l1sz, uint64_t *l2sz) { int cached_flags; if (bps == NULL) return; for (size_t blk_off = 0; blk_off < nbps; blk_off++) { blkptr_t *bp = &bps[blk_off]; if (BP_IS_HOLE(bp)) continue; cached_flags = arc_cached(spa, bp); if (cached_flags == 0) continue; if ((cached_flags & (ARC_CACHED_IN_L1 | ARC_CACHED_IN_L2)) == ARC_CACHED_IN_L2) *l2sz += BP_GET_LSIZE(bp); else *l1sz += BP_GET_LSIZE(bp); } } /* * Estimate DMU object cached size. */ int dmu_object_cached_size(objset_t *os, uint64_t object, uint64_t *l1sz, uint64_t *l2sz) { dnode_t *dn; dmu_object_info_t doi; int err = 0; *l1sz = *l2sz = 0; if (dnode_hold(os, object, FTAG, &dn) != 0) return (0); if (dn->dn_nlevels < 2) { dnode_rele(dn, FTAG); return (0); } dmu_object_info_from_dnode(dn, &doi); for (uint64_t off = 0; off < doi.doi_max_offset; off += dmu_prefetch_max) { /* dbuf_read doesn't prefetch L1 blocks. */ dmu_prefetch_by_dnode(dn, 1, off, dmu_prefetch_max, ZIO_PRIORITY_SYNC_READ); } /* * Hold all valid L1 blocks, asking ARC the status of each BP * contained in each such L1 block. */ uint_t nbps = bp_span_in_blocks(dn->dn_indblkshift, 1); uint64_t l1blks = 1 + (dn->dn_maxblkid / nbps); rw_enter(&dn->dn_struct_rwlock, RW_READER); for (uint64_t blk = 0; blk < l1blks; blk++) { dmu_buf_impl_t *db = NULL; if (issig()) { /* * On interrupt, get out, and bubble up EINTR */ err = EINTR; break; } /* * If we get an i/o error here, the L1 can't be read, * and nothing under it could be cached, so we just * continue. Ignoring the error from dbuf_hold_impl * or from dbuf_read is then a reasonable choice. */ err = dbuf_hold_impl(dn, 1, blk, B_TRUE, B_FALSE, FTAG, &db); if (err != 0) { /* * ignore error and continue */ err = 0; continue; } err = dbuf_read(db, NULL, DB_RF_CANFAIL); if (err == 0) { dmu_cached_bps(dmu_objset_spa(os), db->db.db_data, nbps, l1sz, l2sz); } /* * error may be ignored, and we continue */ err = 0; dbuf_rele(db, FTAG); } rw_exit(&dn->dn_struct_rwlock); dnode_rele(dn, FTAG); return (err); } /* * Allocate a loaned anonymous arc buffer. */ arc_buf_t * dmu_request_arcbuf(dmu_buf_t *handle, int size) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)handle; return (arc_loan_buf(db->db_objset->os_spa, B_FALSE, size)); } /* * Free a loaned arc buffer. */ void dmu_return_arcbuf(arc_buf_t *buf) { arc_return_buf(buf, FTAG); arc_buf_destroy(buf, FTAG); } /* * A "lightweight" write is faster than a regular write (e.g. * dmu_write_by_dnode() or dmu_assign_arcbuf_by_dnode()), because it avoids the * CPU cost of creating a dmu_buf_impl_t and arc_buf_[hdr_]_t. However, the * data can not be read or overwritten until the transaction's txg has been * synced. This makes it appropriate for workloads that are known to be * (temporarily) write-only, like "zfs receive". * * A single block is written, starting at the specified offset in bytes. If * the call is successful, it returns 0 and the provided abd has been * consumed (the caller should not free it). */ int dmu_lightweight_write_by_dnode(dnode_t *dn, uint64_t offset, abd_t *abd, const zio_prop_t *zp, zio_flag_t flags, dmu_tx_t *tx) { dbuf_dirty_record_t *dr = dbuf_dirty_lightweight(dn, dbuf_whichblock(dn, 0, offset), tx); if (dr == NULL) return (SET_ERROR(EIO)); dr->dt.dll.dr_abd = abd; dr->dt.dll.dr_props = *zp; dr->dt.dll.dr_flags = flags; return (0); } /* * When possible directly assign passed loaned arc buffer to a dbuf. * If this is not possible copy the contents of passed arc buf via * dmu_write(). */ int dmu_assign_arcbuf_by_dnode(dnode_t *dn, uint64_t offset, arc_buf_t *buf, dmu_tx_t *tx, dmu_flags_t flags) { dmu_buf_impl_t *db; objset_t *os = dn->dn_objset; uint32_t blksz = (uint32_t)arc_buf_lsize(buf); uint64_t blkid; rw_enter(&dn->dn_struct_rwlock, RW_READER); blkid = dbuf_whichblock(dn, 0, offset); db = dbuf_hold(dn, blkid, FTAG); rw_exit(&dn->dn_struct_rwlock); if (db == NULL) return (SET_ERROR(EIO)); /* * We can only assign if the offset is aligned and the arc buf is the * same size as the dbuf. */ if (offset == db->db.db_offset && blksz == db->db.db_size) { zfs_racct_write(os->os_spa, blksz, 1, flags); dbuf_assign_arcbuf(db, buf, tx, flags); dbuf_rele(db, FTAG); } else { /* compressed bufs must always be assignable to their dbuf */ ASSERT3U(arc_get_compression(buf), ==, ZIO_COMPRESS_OFF); ASSERT(!(buf->b_flags & ARC_BUF_FLAG_COMPRESSED)); dbuf_rele(db, FTAG); dmu_write_by_dnode(dn, offset, blksz, buf->b_data, tx, flags); dmu_return_arcbuf(buf); } return (0); } int dmu_assign_arcbuf_by_dbuf(dmu_buf_t *handle, uint64_t offset, arc_buf_t *buf, dmu_tx_t *tx, dmu_flags_t flags) { int err; dmu_buf_impl_t *db = (dmu_buf_impl_t *)handle; DB_DNODE_ENTER(db); err = dmu_assign_arcbuf_by_dnode(DB_DNODE(db), offset, buf, tx, flags); DB_DNODE_EXIT(db); return (err); } void dmu_sync_ready(zio_t *zio, arc_buf_t *buf, void *varg) { (void) buf; dmu_sync_arg_t *dsa = varg; if (zio->io_error == 0) { dbuf_dirty_record_t *dr = dsa->dsa_dr; blkptr_t *bp = zio->io_bp; if (BP_IS_HOLE(bp)) { dmu_buf_t *db = NULL; if (dr) db = &(dr->dr_dbuf->db); else db = dsa->dsa_zgd->zgd_db; /* * A block of zeros may compress to a hole, but the * block size still needs to be known for replay. */ BP_SET_LSIZE(bp, db->db_size); } else if (!BP_IS_EMBEDDED(bp)) { ASSERT(BP_GET_LEVEL(bp) == 0); BP_SET_FILL(bp, 1); } } } static void dmu_sync_late_arrival_ready(zio_t *zio) { dmu_sync_ready(zio, NULL, zio->io_private); } void dmu_sync_done(zio_t *zio, arc_buf_t *buf, void *varg) { (void) buf; dmu_sync_arg_t *dsa = varg; dbuf_dirty_record_t *dr = dsa->dsa_dr; dmu_buf_impl_t *db = dr->dr_dbuf; zgd_t *zgd = dsa->dsa_zgd; /* * Record the vdev(s) backing this blkptr so they can be flushed after * the writes for the lwb have completed. */ if (zgd && zio->io_error == 0) { zil_lwb_add_block(zgd->zgd_lwb, zgd->zgd_bp); } mutex_enter(&db->db_mtx); ASSERT(dr->dt.dl.dr_override_state == DR_IN_DMU_SYNC); if (zio->io_error == 0) { ASSERT0(dr->dt.dl.dr_has_raw_params); dr->dt.dl.dr_nopwrite = !!(zio->io_flags & ZIO_FLAG_NOPWRITE); if (dr->dt.dl.dr_nopwrite) { blkptr_t *bp = zio->io_bp; blkptr_t *bp_orig = &zio->io_bp_orig; uint8_t chksum = BP_GET_CHECKSUM(bp_orig); ASSERT(BP_EQUAL(bp, bp_orig)); VERIFY(BP_EQUAL(bp, db->db_blkptr)); ASSERT(zio->io_prop.zp_compress != ZIO_COMPRESS_OFF); VERIFY(zio_checksum_table[chksum].ci_flags & ZCHECKSUM_FLAG_NOPWRITE); } dr->dt.dl.dr_overridden_by = *zio->io_bp; dr->dt.dl.dr_override_state = DR_OVERRIDDEN; dr->dt.dl.dr_copies = zio->io_prop.zp_copies; dr->dt.dl.dr_gang_copies = zio->io_prop.zp_gang_copies; /* * Old style holes are filled with all zeros, whereas * new-style holes maintain their lsize, type, level, * and birth time (see zio_write_compress). While we * need to reset the BP_SET_LSIZE() call that happened * in dmu_sync_ready for old style holes, we do *not* * want to wipe out the information contained in new * style holes. Thus, only zero out the block pointer if * it's an old style hole. */ if (BP_IS_HOLE(&dr->dt.dl.dr_overridden_by) && BP_GET_LOGICAL_BIRTH(&dr->dt.dl.dr_overridden_by) == 0) BP_ZERO(&dr->dt.dl.dr_overridden_by); } else { dr->dt.dl.dr_override_state = DR_NOT_OVERRIDDEN; } cv_broadcast(&db->db_changed); mutex_exit(&db->db_mtx); if (dsa->dsa_done) dsa->dsa_done(dsa->dsa_zgd, zio->io_error); kmem_free(dsa, sizeof (*dsa)); } static void dmu_sync_late_arrival_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; dmu_sync_arg_t *dsa = zio->io_private; zgd_t *zgd = dsa->dsa_zgd; if (zio->io_error == 0) { /* * Record the vdev(s) backing this blkptr so they can be * flushed after the writes for the lwb have completed. */ zil_lwb_add_block(zgd->zgd_lwb, zgd->zgd_bp); if (!BP_IS_HOLE(bp)) { blkptr_t *bp_orig __maybe_unused = &zio->io_bp_orig; ASSERT(!(zio->io_flags & ZIO_FLAG_NOPWRITE)); ASSERT(BP_IS_HOLE(bp_orig) || !BP_EQUAL(bp, bp_orig)); ASSERT(BP_GET_BIRTH(zio->io_bp) == zio->io_txg); ASSERT(zio->io_txg > spa_syncing_txg(zio->io_spa)); zio_free(zio->io_spa, zio->io_txg, zio->io_bp); } } dmu_tx_commit(dsa->dsa_tx); dsa->dsa_done(dsa->dsa_zgd, zio->io_error); abd_free(zio->io_abd); kmem_free(dsa, sizeof (*dsa)); } static int dmu_sync_late_arrival(zio_t *pio, objset_t *os, dmu_sync_cb_t *done, zgd_t *zgd, zio_prop_t *zp, zbookmark_phys_t *zb) { dmu_sync_arg_t *dsa; dmu_tx_t *tx; int error; error = dbuf_read((dmu_buf_impl_t *)zgd->zgd_db, NULL, DB_RF_CANFAIL | DMU_READ_NO_PREFETCH | DMU_KEEP_CACHING); if (error != 0) return (error); tx = dmu_tx_create(os); dmu_tx_hold_space(tx, zgd->zgd_db->db_size); /* * This transaction does not produce any dirty data or log blocks, so * it should not be throttled. All other cases wait for TXG sync, by * which time the log block we are writing will be obsolete, so we can * skip waiting and just return error here instead. */ if (dmu_tx_assign(tx, DMU_TX_NOWAIT | DMU_TX_NOTHROTTLE) != 0) { dmu_tx_abort(tx); /* Make zl_get_data do txg_waited_synced() */ return (SET_ERROR(EIO)); } /* * In order to prevent the zgd's lwb from being free'd prior to * dmu_sync_late_arrival_done() being called, we have to ensure * the lwb's "max txg" takes this tx's txg into account. */ zil_lwb_add_txg(zgd->zgd_lwb, dmu_tx_get_txg(tx)); dsa = kmem_alloc(sizeof (dmu_sync_arg_t), KM_SLEEP); dsa->dsa_dr = NULL; dsa->dsa_done = done; dsa->dsa_zgd = zgd; dsa->dsa_tx = tx; /* * Since we are currently syncing this txg, it's nontrivial to * determine what BP to nopwrite against, so we disable nopwrite. * * When syncing, the db_blkptr is initially the BP of the previous * txg. We can not nopwrite against it because it will be changed * (this is similar to the non-late-arrival case where the dbuf is * dirty in a future txg). * * Then dbuf_write_ready() sets bp_blkptr to the location we will write. * We can not nopwrite against it because although the BP will not * (typically) be changed, the data has not yet been persisted to this * location. * * Finally, when dbuf_write_done() is called, it is theoretically * possible to always nopwrite, because the data that was written in * this txg is the same data that we are trying to write. However we * would need to check that this dbuf is not dirty in any future * txg's (as we do in the normal dmu_sync() path). For simplicity, we * don't nopwrite in this case. */ zp->zp_nopwrite = B_FALSE; zio_nowait(zio_write(pio, os->os_spa, dmu_tx_get_txg(tx), zgd->zgd_bp, abd_get_from_buf(zgd->zgd_db->db_data, zgd->zgd_db->db_size), zgd->zgd_db->db_size, zgd->zgd_db->db_size, zp, dmu_sync_late_arrival_ready, NULL, dmu_sync_late_arrival_done, dsa, ZIO_PRIORITY_SYNC_WRITE, ZIO_FLAG_CANFAIL, zb)); return (0); } /* * Intent log support: sync the block associated with db to disk. * N.B. and XXX: the caller is responsible for making sure that the * data isn't changing while dmu_sync() is writing it. * * Return values: * * EEXIST: this txg has already been synced, so there's nothing to do. * The caller should not log the write. * * ENOENT: the block was dbuf_free_range()'d, so there's nothing to do. * The caller should not log the write. * * EALREADY: this block is already in the process of being synced. * The caller should track its progress (somehow). * * EIO: could not do the I/O. * The caller should do a txg_wait_synced(). * * 0: the I/O has been initiated. * The caller should log this blkptr in the done callback. * It is possible that the I/O will fail, in which case * the error will be reported to the done callback and * propagated to pio from zio_done(). */ int dmu_sync(zio_t *pio, uint64_t txg, dmu_sync_cb_t *done, zgd_t *zgd) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)zgd->zgd_db; objset_t *os = db->db_objset; dsl_dataset_t *ds = os->os_dsl_dataset; dbuf_dirty_record_t *dr, *dr_next; dmu_sync_arg_t *dsa; zbookmark_phys_t zb; zio_prop_t zp; ASSERT(pio != NULL); ASSERT(txg != 0); SET_BOOKMARK(&zb, ds->ds_object, db->db.db_object, db->db_level, db->db_blkid); DB_DNODE_ENTER(db); dmu_write_policy(os, DB_DNODE(db), db->db_level, WP_DMU_SYNC, &zp); DB_DNODE_EXIT(db); /* * If we're frozen (running ziltest), we always need to generate a bp. */ if (txg > spa_freeze_txg(os->os_spa)) return (dmu_sync_late_arrival(pio, os, done, zgd, &zp, &zb)); /* * Grabbing db_mtx now provides a barrier between dbuf_sync_leaf() * and us. If we determine that this txg is not yet syncing, * but it begins to sync a moment later, that's OK because the * sync thread will block in dbuf_sync_leaf() until we drop db_mtx. */ mutex_enter(&db->db_mtx); if (txg <= spa_last_synced_txg(os->os_spa)) { /* * This txg has already synced. There's nothing to do. */ mutex_exit(&db->db_mtx); return (SET_ERROR(EEXIST)); } if (txg <= spa_syncing_txg(os->os_spa)) { /* * This txg is currently syncing, so we can't mess with * the dirty record anymore; just write a new log block. */ mutex_exit(&db->db_mtx); return (dmu_sync_late_arrival(pio, os, done, zgd, &zp, &zb)); } dr = dbuf_find_dirty_eq(db, txg); if (dr == NULL) { /* * There's no dr for this dbuf, so it must have been freed. * There's no need to log writes to freed blocks, so we're done. */ mutex_exit(&db->db_mtx); return (SET_ERROR(ENOENT)); } dr_next = list_next(&db->db_dirty_records, dr); ASSERT(dr_next == NULL || dr_next->dr_txg < txg); if (db->db_blkptr != NULL) { /* * We need to fill in zgd_bp with the current blkptr so that * the nopwrite code can check if we're writing the same * data that's already on disk. We can only nopwrite if we * are sure that after making the copy, db_blkptr will not * change until our i/o completes. We ensure this by * holding the db_mtx, and only allowing nopwrite if the * block is not already dirty (see below). This is verified * by dmu_sync_done(), which VERIFYs that the db_blkptr has * not changed. */ *zgd->zgd_bp = *db->db_blkptr; } /* * Assume the on-disk data is X, the current syncing data (in * txg - 1) is Y, and the current in-memory data is Z (currently * in dmu_sync). * * We usually want to perform a nopwrite if X and Z are the * same. However, if Y is different (i.e. the BP is going to * change before this write takes effect), then a nopwrite will * be incorrect - we would override with X, which could have * been freed when Y was written. * * (Note that this is not a concern when we are nop-writing from * syncing context, because X and Y must be identical, because * all previous txgs have been synced.) * * Therefore, we disable nopwrite if the current BP could change * before this TXG. There are two ways it could change: by * being dirty (dr_next is non-NULL), or by being freed * (dnode_block_freed()). This behavior is verified by * zio_done(), which VERIFYs that the override BP is identical * to the on-disk BP. */ if (dr_next != NULL) { zp.zp_nopwrite = B_FALSE; } else { DB_DNODE_ENTER(db); if (dnode_block_freed(DB_DNODE(db), db->db_blkid)) zp.zp_nopwrite = B_FALSE; DB_DNODE_EXIT(db); } ASSERT(dr->dr_txg == txg); if (dr->dt.dl.dr_override_state == DR_IN_DMU_SYNC || dr->dt.dl.dr_override_state == DR_OVERRIDDEN) { /* * We have already issued a sync write for this buffer, * or this buffer has already been synced. It could not * have been dirtied since, or we would have cleared the state. */ mutex_exit(&db->db_mtx); return (SET_ERROR(EALREADY)); } ASSERT0(dr->dt.dl.dr_has_raw_params); ASSERT(dr->dt.dl.dr_override_state == DR_NOT_OVERRIDDEN); dr->dt.dl.dr_override_state = DR_IN_DMU_SYNC; mutex_exit(&db->db_mtx); dsa = kmem_alloc(sizeof (dmu_sync_arg_t), KM_SLEEP); dsa->dsa_dr = dr; dsa->dsa_done = done; dsa->dsa_zgd = zgd; dsa->dsa_tx = NULL; zio_nowait(arc_write(pio, os->os_spa, txg, zgd->zgd_bp, dr->dt.dl.dr_data, !DBUF_IS_CACHEABLE(db), dbuf_is_l2cacheable(db, NULL), &zp, dmu_sync_ready, NULL, dmu_sync_done, dsa, ZIO_PRIORITY_SYNC_WRITE, ZIO_FLAG_CANFAIL, &zb)); return (0); } int dmu_object_set_nlevels(objset_t *os, uint64_t object, int nlevels, dmu_tx_t *tx) { dnode_t *dn; int err; err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); err = dnode_set_nlevels(dn, nlevels, tx); dnode_rele(dn, FTAG); return (err); } int dmu_object_set_blocksize(objset_t *os, uint64_t object, uint64_t size, int ibs, dmu_tx_t *tx) { dnode_t *dn; int err; err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); err = dnode_set_blksz(dn, size, ibs, tx); dnode_rele(dn, FTAG); return (err); } int dmu_object_set_maxblkid(objset_t *os, uint64_t object, uint64_t maxblkid, dmu_tx_t *tx) { dnode_t *dn; int err; err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); dnode_new_blkid(dn, maxblkid, tx, B_FALSE, B_TRUE); rw_exit(&dn->dn_struct_rwlock); dnode_rele(dn, FTAG); return (0); } void dmu_object_set_checksum(objset_t *os, uint64_t object, uint8_t checksum, dmu_tx_t *tx) { dnode_t *dn; /* * Send streams include each object's checksum function. This * check ensures that the receiving system can understand the * checksum function transmitted. */ ASSERT3U(checksum, <, ZIO_CHECKSUM_LEGACY_FUNCTIONS); VERIFY0(dnode_hold(os, object, FTAG, &dn)); ASSERT3U(checksum, <, ZIO_CHECKSUM_FUNCTIONS); dn->dn_checksum = checksum; dnode_setdirty(dn, tx); dnode_rele(dn, FTAG); } void dmu_object_set_compress(objset_t *os, uint64_t object, uint8_t compress, dmu_tx_t *tx) { dnode_t *dn; /* * Send streams include each object's compression function. This * check ensures that the receiving system can understand the * compression function transmitted. */ ASSERT3U(compress, <, ZIO_COMPRESS_LEGACY_FUNCTIONS); VERIFY0(dnode_hold(os, object, FTAG, &dn)); dn->dn_compress = compress; dnode_setdirty(dn, tx); dnode_rele(dn, FTAG); } /* * When the "redundant_metadata" property is set to "most", only indirect * blocks of this level and higher will have an additional ditto block. */ static const int zfs_redundant_metadata_most_ditto_level = 2; void dmu_write_policy(objset_t *os, dnode_t *dn, int level, int wp, zio_prop_t *zp) { dmu_object_type_t type = dn ? dn->dn_type : DMU_OT_OBJSET; boolean_t ismd = (level > 0 || DMU_OT_IS_METADATA(type) || (wp & WP_SPILL)); enum zio_checksum checksum = os->os_checksum; enum zio_compress compress = os->os_compress; uint8_t complevel = os->os_complevel; enum zio_checksum dedup_checksum = os->os_dedup_checksum; boolean_t dedup = B_FALSE; boolean_t nopwrite = B_FALSE; boolean_t dedup_verify = os->os_dedup_verify; boolean_t encrypt = B_FALSE; int copies = os->os_copies; int gang_copies = os->os_copies; /* * We maintain different write policies for each of the following * types of data: * 1. metadata * 2. preallocated blocks (i.e. level-0 blocks of a dump device) * 3. all other level 0 blocks */ if (ismd) { /* * XXX -- we should design a compression algorithm * that specializes in arrays of bps. */ compress = zio_compress_select(os->os_spa, ZIO_COMPRESS_ON, ZIO_COMPRESS_ON); /* * Metadata always gets checksummed. If the data * checksum is multi-bit correctable, and it's not a * ZBT-style checksum, then it's suitable for metadata * as well. Otherwise, the metadata checksum defaults * to fletcher4. */ if (!(zio_checksum_table[checksum].ci_flags & ZCHECKSUM_FLAG_METADATA) || (zio_checksum_table[checksum].ci_flags & ZCHECKSUM_FLAG_EMBEDDED)) checksum = ZIO_CHECKSUM_FLETCHER_4; switch (os->os_redundant_metadata) { case ZFS_REDUNDANT_METADATA_ALL: copies++; gang_copies++; break; case ZFS_REDUNDANT_METADATA_MOST: if (level >= zfs_redundant_metadata_most_ditto_level || DMU_OT_IS_METADATA(type) || (wp & WP_SPILL)) copies++; if (level + 1 >= zfs_redundant_metadata_most_ditto_level || DMU_OT_IS_METADATA(type) || (wp & WP_SPILL)) gang_copies++; break; case ZFS_REDUNDANT_METADATA_SOME: if (DMU_OT_IS_CRITICAL(type, level)) { copies++; gang_copies++; } else if (DMU_OT_IS_METADATA(type)) { gang_copies++; } break; case ZFS_REDUNDANT_METADATA_NONE: break; } if (dmu_ddt_copies > 0) { /* * If this tunable is set, and this is a write for a * dedup entry store (zap or log), then we treat it * something like ZFS_REDUNDANT_METADATA_MOST on a * regular dataset: this many copies, and one more for * "higher" indirect blocks. This specific exception is * necessary because dedup objects are stored in the * MOS, which always has the highest possible copies. */ dmu_object_type_t stype = dn ? dn->dn_storage_type : DMU_OT_NONE; if (stype == DMU_OT_NONE) stype = type; if (stype == DMU_OT_DDT_ZAP) { copies = dmu_ddt_copies; if (level >= zfs_redundant_metadata_most_ditto_level) copies++; } } } else if (wp & WP_NOFILL) { ASSERT(level == 0); /* * If we're writing preallocated blocks, we aren't actually * writing them so don't set any policy properties. These * blocks are currently only used by an external subsystem * outside of zfs (i.e. dump) and not written by the zio * pipeline. */ compress = ZIO_COMPRESS_OFF; checksum = ZIO_CHECKSUM_OFF; } else { compress = zio_compress_select(os->os_spa, dn->dn_compress, compress); complevel = zio_complevel_select(os->os_spa, compress, complevel, complevel); /* * Storing many references to an all zeros block in the dedup * table would be expensive. Instead, if dedup is enabled, * store them as holes even if compression is not enabled. */ if (compress == ZIO_COMPRESS_OFF && dedup_checksum != ZIO_CHECKSUM_OFF) compress = ZIO_COMPRESS_EMPTY; checksum = (dedup_checksum == ZIO_CHECKSUM_OFF) ? zio_checksum_select(dn->dn_checksum, checksum) : dedup_checksum; /* * Determine dedup setting. If we are in dmu_sync(), * we won't actually dedup now because that's all * done in syncing context; but we do want to use the * dedup checksum. If the checksum is not strong * enough to ensure unique signatures, force * dedup_verify. */ if (dedup_checksum != ZIO_CHECKSUM_OFF) { dedup = (wp & WP_DMU_SYNC) ? B_FALSE : B_TRUE; if (!(zio_checksum_table[checksum].ci_flags & ZCHECKSUM_FLAG_DEDUP)) dedup_verify = B_TRUE; } /* * Enable nopwrite if we have secure enough checksum * algorithm (see comment in zio_nop_write) and * compression is enabled. We don't enable nopwrite if * dedup is enabled as the two features are mutually * exclusive. */ nopwrite = (!dedup && (zio_checksum_table[checksum].ci_flags & ZCHECKSUM_FLAG_NOPWRITE) && compress != ZIO_COMPRESS_OFF && zfs_nopwrite_enabled); if (os->os_redundant_metadata == ZFS_REDUNDANT_METADATA_ALL || (os->os_redundant_metadata == ZFS_REDUNDANT_METADATA_MOST && zfs_redundant_metadata_most_ditto_level <= 1)) gang_copies++; } /* * All objects in an encrypted objset are protected from modification * via a MAC. Encrypted objects store their IV and salt in the last DVA * in the bp, so we cannot use all copies. Encrypted objects are also * not subject to nopwrite since writing the same data will still * result in a new ciphertext. Only encrypted blocks can be dedup'd * to avoid ambiguity in the dedup code since the DDT does not store * object types. */ if (os->os_encrypted && (wp & WP_NOFILL) == 0) { encrypt = B_TRUE; if (DMU_OT_IS_ENCRYPTED(type)) { copies = MIN(copies, SPA_DVAS_PER_BP - 1); gang_copies = MIN(gang_copies, SPA_DVAS_PER_BP - 1); nopwrite = B_FALSE; } else { dedup = B_FALSE; } if (level <= 0 && (type == DMU_OT_DNODE || type == DMU_OT_OBJSET)) { compress = ZIO_COMPRESS_EMPTY; } } zp->zp_compress = compress; zp->zp_complevel = complevel; zp->zp_checksum = checksum; zp->zp_type = (wp & WP_SPILL) ? dn->dn_bonustype : type; zp->zp_level = level; zp->zp_copies = MIN(copies, spa_max_replication(os->os_spa)); zp->zp_gang_copies = MIN(MAX(gang_copies, copies), spa_max_replication(os->os_spa)); zp->zp_dedup = dedup; zp->zp_dedup_verify = dedup && dedup_verify; zp->zp_nopwrite = nopwrite; zp->zp_encrypt = encrypt; zp->zp_byteorder = ZFS_HOST_BYTEORDER; zp->zp_direct_write = (wp & WP_DIRECT_WR) ? B_TRUE : B_FALSE; + zp->zp_rewrite = B_FALSE; memset(zp->zp_salt, 0, ZIO_DATA_SALT_LEN); memset(zp->zp_iv, 0, ZIO_DATA_IV_LEN); memset(zp->zp_mac, 0, ZIO_DATA_MAC_LEN); zp->zp_zpl_smallblk = (DMU_OT_IS_FILE(zp->zp_type) || zp->zp_type == DMU_OT_ZVOL) ? os->os_zpl_special_smallblock : 0; zp->zp_storage_type = dn ? dn->dn_storage_type : DMU_OT_NONE; ASSERT3U(zp->zp_compress, !=, ZIO_COMPRESS_INHERIT); } /* * Reports the location of data and holes in an object. In order to * accurately report holes all dirty data must be synced to disk. This * causes extremely poor performance when seeking for holes in a dirty file. * As a compromise, only provide hole data when the dnode is clean. When * a dnode is dirty report the dnode as having no holes by returning EBUSY * which is always safe to do. */ int dmu_offset_next(objset_t *os, uint64_t object, boolean_t hole, uint64_t *off) { dnode_t *dn; uint64_t txg, maxtxg = 0; int err; restart: err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); rw_enter(&dn->dn_struct_rwlock, RW_READER); if (dnode_is_dirty(dn)) { /* * If the zfs_dmu_offset_next_sync module option is enabled * then hole reporting has been requested. Dirty dnodes * must be synced to disk to accurately report holes. * * Provided a RL_READER rangelock spanning 0-UINT64_MAX is * held by the caller only limited restarts will be required. * We tolerate callers which do not hold the rangelock by * returning EBUSY and not reporting holes after at most * TXG_CONCURRENT_STATES (3) restarts. */ if (zfs_dmu_offset_next_sync) { rw_exit(&dn->dn_struct_rwlock); dnode_rele(dn, FTAG); if (maxtxg == 0) { txg = spa_last_synced_txg(dmu_objset_spa(os)); maxtxg = txg + TXG_CONCURRENT_STATES; } else if (txg >= maxtxg) return (SET_ERROR(EBUSY)); txg_wait_synced(dmu_objset_pool(os), ++txg); goto restart; } err = SET_ERROR(EBUSY); } else { err = dnode_next_offset(dn, DNODE_FIND_HAVELOCK | (hole ? DNODE_FIND_HOLE : 0), off, 1, 1, 0); } rw_exit(&dn->dn_struct_rwlock); dnode_rele(dn, FTAG); return (err); } int dmu_read_l0_bps(objset_t *os, uint64_t object, uint64_t offset, uint64_t length, blkptr_t *bps, size_t *nbpsp) { dmu_buf_t **dbp, *dbuf; dmu_buf_impl_t *db; blkptr_t *bp; int error, numbufs; error = dmu_buf_hold_array(os, object, offset, length, FALSE, FTAG, &numbufs, &dbp); if (error != 0) { if (error == ESRCH) { error = SET_ERROR(ENXIO); } return (error); } ASSERT3U(numbufs, <=, *nbpsp); for (int i = 0; i < numbufs; i++) { dbuf = dbp[i]; db = (dmu_buf_impl_t *)dbuf; mutex_enter(&db->db_mtx); if (!list_is_empty(&db->db_dirty_records)) { dbuf_dirty_record_t *dr; dr = list_head(&db->db_dirty_records); if (dr->dt.dl.dr_brtwrite) { /* * This is very special case where we clone a * block and in the same transaction group we * read its BP (most likely to clone the clone). */ bp = &dr->dt.dl.dr_overridden_by; } else { /* * The block was modified in the same * transaction group. */ mutex_exit(&db->db_mtx); error = SET_ERROR(EAGAIN); goto out; } } else { bp = db->db_blkptr; } mutex_exit(&db->db_mtx); if (bp == NULL) { /* * The file size was increased, but the block was never * written, otherwise we would either have the block * pointer or the dirty record and would not get here. * It is effectively a hole, so report it as such. */ BP_ZERO(&bps[i]); continue; } /* * Make sure we clone only data blocks. */ if (BP_IS_METADATA(bp) && !BP_IS_HOLE(bp)) { error = SET_ERROR(EINVAL); goto out; } /* * If the block was allocated in transaction group that is not * yet synced, we could clone it, but we couldn't write this * operation into ZIL, or it may be impossible to replay, since * the block may appear not yet allocated at that point. */ if (BP_GET_PHYSICAL_BIRTH(bp) > spa_freeze_txg(os->os_spa)) { error = SET_ERROR(EINVAL); goto out; } if (BP_GET_PHYSICAL_BIRTH(bp) > spa_last_synced_txg(os->os_spa)) { error = SET_ERROR(EAGAIN); goto out; } bps[i] = *bp; } *nbpsp = numbufs; out: dmu_buf_rele_array(dbp, numbufs, FTAG); return (error); } int dmu_brt_clone(objset_t *os, uint64_t object, uint64_t offset, uint64_t length, dmu_tx_t *tx, const blkptr_t *bps, size_t nbps) { spa_t *spa; dmu_buf_t **dbp, *dbuf; dmu_buf_impl_t *db; struct dirty_leaf *dl; dbuf_dirty_record_t *dr; const blkptr_t *bp; int error = 0, i, numbufs; spa = os->os_spa; VERIFY0(dmu_buf_hold_array(os, object, offset, length, FALSE, FTAG, &numbufs, &dbp)); ASSERT3U(nbps, ==, numbufs); /* * Before we start cloning make sure that the dbufs sizes match new BPs * sizes. If they don't, that's a no-go, as we are not able to shrink * dbufs. */ for (i = 0; i < numbufs; i++) { dbuf = dbp[i]; db = (dmu_buf_impl_t *)dbuf; bp = &bps[i]; ASSERT3U(db->db.db_object, !=, DMU_META_DNODE_OBJECT); ASSERT0(db->db_level); ASSERT(db->db_blkid != DMU_BONUS_BLKID); ASSERT(db->db_blkid != DMU_SPILL_BLKID); if (!BP_IS_HOLE(bp) && BP_GET_LSIZE(bp) != dbuf->db_size) { error = SET_ERROR(EXDEV); goto out; } } for (i = 0; i < numbufs; i++) { dbuf = dbp[i]; db = (dmu_buf_impl_t *)dbuf; bp = &bps[i]; dmu_buf_will_clone_or_dio(dbuf, tx); mutex_enter(&db->db_mtx); dr = list_head(&db->db_dirty_records); VERIFY(dr != NULL); ASSERT3U(dr->dr_txg, ==, tx->tx_txg); dl = &dr->dt.dl; ASSERT0(dl->dr_has_raw_params); dl->dr_overridden_by = *bp; if (!BP_IS_HOLE(bp) || BP_GET_LOGICAL_BIRTH(bp) != 0) { if (!BP_IS_EMBEDDED(bp)) { BP_SET_BIRTH(&dl->dr_overridden_by, dr->dr_txg, BP_GET_PHYSICAL_BIRTH(bp)); BP_SET_REWRITE(&dl->dr_overridden_by, 0); } else { BP_SET_LOGICAL_BIRTH(&dl->dr_overridden_by, dr->dr_txg); } } dl->dr_brtwrite = B_TRUE; dl->dr_override_state = DR_OVERRIDDEN; mutex_exit(&db->db_mtx); /* * When data in embedded into BP there is no need to create * BRT entry as there is no data block. Just copy the BP as * it contains the data. */ if (!BP_IS_HOLE(bp) && !BP_IS_EMBEDDED(bp)) { brt_pending_add(spa, bp, tx); } } out: dmu_buf_rele_array(dbp, numbufs, FTAG); return (error); } void __dmu_object_info_from_dnode(dnode_t *dn, dmu_object_info_t *doi) { dnode_phys_t *dnp = dn->dn_phys; doi->doi_data_block_size = dn->dn_datablksz; doi->doi_metadata_block_size = dn->dn_indblkshift ? 1ULL << dn->dn_indblkshift : 0; doi->doi_type = dn->dn_type; doi->doi_bonus_type = dn->dn_bonustype; doi->doi_bonus_size = dn->dn_bonuslen; doi->doi_dnodesize = dn->dn_num_slots << DNODE_SHIFT; doi->doi_indirection = dn->dn_nlevels; doi->doi_checksum = dn->dn_checksum; doi->doi_compress = dn->dn_compress; doi->doi_nblkptr = dn->dn_nblkptr; doi->doi_physical_blocks_512 = (DN_USED_BYTES(dnp) + 256) >> 9; doi->doi_max_offset = (dn->dn_maxblkid + 1) * dn->dn_datablksz; doi->doi_fill_count = 0; for (int i = 0; i < dnp->dn_nblkptr; i++) doi->doi_fill_count += BP_GET_FILL(&dnp->dn_blkptr[i]); } void dmu_object_info_from_dnode(dnode_t *dn, dmu_object_info_t *doi) { rw_enter(&dn->dn_struct_rwlock, RW_READER); mutex_enter(&dn->dn_mtx); __dmu_object_info_from_dnode(dn, doi); mutex_exit(&dn->dn_mtx); rw_exit(&dn->dn_struct_rwlock); } /* * Get information on a DMU object. * If doi is NULL, just indicates whether the object exists. */ int dmu_object_info(objset_t *os, uint64_t object, dmu_object_info_t *doi) { dnode_t *dn; int err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); if (doi != NULL) dmu_object_info_from_dnode(dn, doi); dnode_rele(dn, FTAG); return (0); } /* * As above, but faster; can be used when you have a held dbuf in hand. */ void dmu_object_info_from_db(dmu_buf_t *db_fake, dmu_object_info_t *doi) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; DB_DNODE_ENTER(db); dmu_object_info_from_dnode(DB_DNODE(db), doi); DB_DNODE_EXIT(db); } /* * Faster still when you only care about the size. */ void dmu_object_size_from_db(dmu_buf_t *db_fake, uint32_t *blksize, u_longlong_t *nblk512) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dnode_t *dn; DB_DNODE_ENTER(db); dn = DB_DNODE(db); *blksize = dn->dn_datablksz; /* add in number of slots used for the dnode itself */ *nblk512 = ((DN_USED_BYTES(dn->dn_phys) + SPA_MINBLOCKSIZE/2) >> SPA_MINBLOCKSHIFT) + dn->dn_num_slots; DB_DNODE_EXIT(db); } void dmu_object_dnsize_from_db(dmu_buf_t *db_fake, int *dnsize) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; DB_DNODE_ENTER(db); *dnsize = DB_DNODE(db)->dn_num_slots << DNODE_SHIFT; DB_DNODE_EXIT(db); } void byteswap_uint64_array(void *vbuf, size_t size) { uint64_t *buf = vbuf; size_t count = size >> 3; int i; ASSERT((size & 7) == 0); for (i = 0; i < count; i++) buf[i] = BSWAP_64(buf[i]); } void byteswap_uint32_array(void *vbuf, size_t size) { uint32_t *buf = vbuf; size_t count = size >> 2; int i; ASSERT((size & 3) == 0); for (i = 0; i < count; i++) buf[i] = BSWAP_32(buf[i]); } void byteswap_uint16_array(void *vbuf, size_t size) { uint16_t *buf = vbuf; size_t count = size >> 1; int i; ASSERT((size & 1) == 0); for (i = 0; i < count; i++) buf[i] = BSWAP_16(buf[i]); } void byteswap_uint8_array(void *vbuf, size_t size) { (void) vbuf, (void) size; } void dmu_init(void) { abd_init(); zfs_dbgmsg_init(); sa_cache_init(); dmu_objset_init(); dnode_init(); zfetch_init(); dmu_tx_init(); l2arc_init(); arc_init(); dbuf_init(); } void dmu_fini(void) { arc_fini(); /* arc depends on l2arc, so arc must go first */ l2arc_fini(); dmu_tx_fini(); zfetch_fini(); dbuf_fini(); dnode_fini(); dmu_objset_fini(); sa_cache_fini(); zfs_dbgmsg_fini(); abd_fini(); } EXPORT_SYMBOL(dmu_bonus_hold); EXPORT_SYMBOL(dmu_bonus_hold_by_dnode); EXPORT_SYMBOL(dmu_buf_hold_array_by_bonus); EXPORT_SYMBOL(dmu_buf_rele_array); EXPORT_SYMBOL(dmu_prefetch); EXPORT_SYMBOL(dmu_prefetch_by_dnode); EXPORT_SYMBOL(dmu_prefetch_dnode); EXPORT_SYMBOL(dmu_free_range); EXPORT_SYMBOL(dmu_free_long_range); EXPORT_SYMBOL(dmu_free_long_object); EXPORT_SYMBOL(dmu_read); EXPORT_SYMBOL(dmu_read_by_dnode); EXPORT_SYMBOL(dmu_read_uio); EXPORT_SYMBOL(dmu_read_uio_dbuf); EXPORT_SYMBOL(dmu_read_uio_dnode); EXPORT_SYMBOL(dmu_write); EXPORT_SYMBOL(dmu_write_by_dnode); EXPORT_SYMBOL(dmu_write_uio); EXPORT_SYMBOL(dmu_write_uio_dbuf); EXPORT_SYMBOL(dmu_write_uio_dnode); EXPORT_SYMBOL(dmu_prealloc); EXPORT_SYMBOL(dmu_object_info); EXPORT_SYMBOL(dmu_object_info_from_dnode); EXPORT_SYMBOL(dmu_object_info_from_db); EXPORT_SYMBOL(dmu_object_size_from_db); EXPORT_SYMBOL(dmu_object_dnsize_from_db); EXPORT_SYMBOL(dmu_object_set_nlevels); EXPORT_SYMBOL(dmu_object_set_blocksize); EXPORT_SYMBOL(dmu_object_set_maxblkid); EXPORT_SYMBOL(dmu_object_set_checksum); EXPORT_SYMBOL(dmu_object_set_compress); EXPORT_SYMBOL(dmu_offset_next); EXPORT_SYMBOL(dmu_write_policy); EXPORT_SYMBOL(dmu_sync); EXPORT_SYMBOL(dmu_request_arcbuf); EXPORT_SYMBOL(dmu_return_arcbuf); EXPORT_SYMBOL(dmu_assign_arcbuf_by_dnode); EXPORT_SYMBOL(dmu_assign_arcbuf_by_dbuf); EXPORT_SYMBOL(dmu_buf_hold); EXPORT_SYMBOL(dmu_ot); ZFS_MODULE_PARAM(zfs, zfs_, nopwrite_enabled, INT, ZMOD_RW, "Enable NOP writes"); ZFS_MODULE_PARAM(zfs, zfs_, per_txg_dirty_frees_percent, UINT, ZMOD_RW, "Percentage of dirtied blocks from frees in one TXG"); ZFS_MODULE_PARAM(zfs, zfs_, dmu_offset_next_sync, INT, ZMOD_RW, "Enable forcing txg sync to find holes"); ZFS_MODULE_PARAM(zfs, , dmu_prefetch_max, UINT, ZMOD_RW, "Limit one prefetch call to this size"); ZFS_MODULE_PARAM(zfs, , dmu_ddt_copies, UINT, ZMOD_RW, "Override copies= for dedup objects"); diff --git a/module/zfs/zfs_vnops.c b/module/zfs/zfs_vnops.c index 8ad992f5b62b..74aa91a4f2eb 100644 --- a/module/zfs/zfs_vnops.c +++ b/module/zfs/zfs_vnops.c @@ -1,2066 +1,2078 @@ // SPDX-License-Identifier: CDDL-1.0 /* * 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 https://opensource.org/licenses/CDDL-1.0. * 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. * Copyright (c) 2015 by Chunwei Chen. All rights reserved. * Copyright 2017 Nexenta Systems, Inc. * Copyright (c) 2021, 2022 by Pawel Jakub Dawidek * Copyright (c) 2025, Rob Norris */ /* Portions Copyright 2007 Jeremy Teo */ /* 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 /* * Enables access to the block cloning feature. If this setting is 0, then even * if feature@block_cloning is enabled, using functions and system calls that * attempt to clone blocks will act as though the feature is disabled. */ int zfs_bclone_enabled = 1; /* * When set to 1 the FICLONE and FICLONERANGE ioctls will wait for any dirty * data to be written to disk before proceeding. This ensures that the clone * operation reliably succeeds, even if a file is modified and then immediately * cloned. Note that for small files this may be slower than simply copying * the file. When set to 0 the clone operation will immediately fail if it * encounters any dirty blocks. By default waiting is enabled. */ int zfs_bclone_wait_dirty = 1; /* * Enable Direct I/O. If this setting is 0, then all I/O requests will be * directed through the ARC acting as though the dataset property direct was * set to disabled. * * Disabled by default on FreeBSD until a potential range locking issue in * zfs_getpages() can be resolved. */ #ifdef __FreeBSD__ static int zfs_dio_enabled = 0; #else static int zfs_dio_enabled = 1; #endif /* * Strictly enforce alignment for Direct I/O requests, returning EINVAL * if not page-aligned instead of silently falling back to uncached I/O. */ static int zfs_dio_strict = 0; /* * Maximum bytes to read per chunk in zfs_read(). */ #ifdef _ILP32 static uint64_t zfs_vnops_read_chunk_size = 1024 * 1024; #else static uint64_t zfs_vnops_read_chunk_size = DMU_MAX_ACCESS / 2; #endif int zfs_fsync(znode_t *zp, int syncflag, cred_t *cr) { int error = 0; zfsvfs_t *zfsvfs = ZTOZSB(zp); if (zfsvfs->z_os->os_sync != ZFS_SYNC_DISABLED) { if ((error = zfs_enter_verify_zp(zfsvfs, zp, FTAG)) != 0) return (error); zil_commit(zfsvfs->z_log, zp->z_id); zfs_exit(zfsvfs, FTAG); } return (error); } #if defined(SEEK_HOLE) && defined(SEEK_DATA) /* * Lseek support for finding holes (cmd == SEEK_HOLE) and * data (cmd == SEEK_DATA). "off" is an in/out parameter. */ static int zfs_holey_common(znode_t *zp, ulong_t cmd, loff_t *off) { zfs_locked_range_t *lr; uint64_t noff = (uint64_t)*off; /* new offset */ uint64_t file_sz; int error; boolean_t hole; file_sz = zp->z_size; if (noff >= file_sz) { return (SET_ERROR(ENXIO)); } if (cmd == F_SEEK_HOLE) hole = B_TRUE; else hole = B_FALSE; /* Flush any mmap()'d data to disk */ if (zn_has_cached_data(zp, 0, file_sz - 1)) zn_flush_cached_data(zp, B_TRUE); lr = zfs_rangelock_enter(&zp->z_rangelock, 0, UINT64_MAX, RL_READER); error = dmu_offset_next(ZTOZSB(zp)->z_os, zp->z_id, hole, &noff); zfs_rangelock_exit(lr); if (error == ESRCH) return (SET_ERROR(ENXIO)); /* File was dirty, so fall back to using generic logic */ if (error == EBUSY) { if (hole) *off = file_sz; return (0); } /* * We could find a hole that begins after the logical end-of-file, * because dmu_offset_next() only works on whole blocks. If the * EOF falls mid-block, then indicate that the "virtual hole" * at the end of the file begins at the logical EOF, rather than * at the end of the last block. */ if (noff > file_sz) { ASSERT(hole); noff = file_sz; } if (noff < *off) return (error); *off = noff; return (error); } int zfs_holey(znode_t *zp, ulong_t cmd, loff_t *off) { zfsvfs_t *zfsvfs = ZTOZSB(zp); int error; if ((error = zfs_enter_verify_zp(zfsvfs, zp, FTAG)) != 0) return (error); error = zfs_holey_common(zp, cmd, off); zfs_exit(zfsvfs, FTAG); return (error); } #endif /* SEEK_HOLE && SEEK_DATA */ int zfs_access(znode_t *zp, int mode, int flag, cred_t *cr) { zfsvfs_t *zfsvfs = ZTOZSB(zp); int error; if ((error = zfs_enter_verify_zp(zfsvfs, zp, FTAG)) != 0) return (error); if (flag & V_ACE_MASK) #if defined(__linux__) error = zfs_zaccess(zp, mode, flag, B_FALSE, cr, zfs_init_idmap); #else error = zfs_zaccess(zp, mode, flag, B_FALSE, cr, NULL); #endif else #if defined(__linux__) error = zfs_zaccess_rwx(zp, mode, flag, cr, zfs_init_idmap); #else error = zfs_zaccess_rwx(zp, mode, flag, cr, NULL); #endif zfs_exit(zfsvfs, FTAG); return (error); } /* * Determine if Direct I/O has been requested (either via the O_DIRECT flag or * the "direct" dataset property). When inherited by the property only apply * the O_DIRECT flag to correctly aligned IO requests. The rational for this * is it allows the property to be safely set on a dataset without forcing * all of the applications to be aware of the alignment restrictions. When * O_DIRECT is explicitly requested by an application return EINVAL if the * request is unaligned. In all cases, if the range for this request has * been mmap'ed then we will perform buffered I/O to keep the mapped region * synhronized with the ARC. * * It is possible that a file's pages could be mmap'ed after it is checked * here. If so, that is handled coorarding in zfs_write(). See comments in the * following area for how this is handled: * zfs_write() -> update_pages() */ static int zfs_setup_direct(struct znode *zp, zfs_uio_t *uio, zfs_uio_rw_t rw, int *ioflagp) { zfsvfs_t *zfsvfs = ZTOZSB(zp); objset_t *os = zfsvfs->z_os; int ioflag = *ioflagp; int error = 0; if (os->os_direct == ZFS_DIRECT_ALWAYS) { /* Force either direct or uncached I/O. */ ioflag |= O_DIRECT; } if ((ioflag & O_DIRECT) == 0) goto out; if (!zfs_dio_enabled || os->os_direct == ZFS_DIRECT_DISABLED) { /* * Direct I/O is disabled. The I/O request will be directed * through the ARC as uncached I/O. */ goto out; } if (!zfs_uio_page_aligned(uio) || !zfs_uio_aligned(uio, PAGE_SIZE)) { /* * Misaligned requests can be executed through the ARC as * uncached I/O. But if O_DIRECT was set by user and we * were set to be strict, then it is a failure. */ if ((*ioflagp & O_DIRECT) && zfs_dio_strict) error = SET_ERROR(EINVAL); goto out; } if (zn_has_cached_data(zp, zfs_uio_offset(uio), zfs_uio_offset(uio) + zfs_uio_resid(uio) - 1)) { /* * The region is mmap'ed. The I/O request will be directed * through the ARC as uncached I/O. */ goto out; } /* * For short writes the page mapping of Direct I/O makes no sense. * Direct them through the ARC as uncached I/O. */ if (rw == UIO_WRITE && zfs_uio_resid(uio) < zp->z_blksz) goto out; error = zfs_uio_get_dio_pages_alloc(uio, rw); if (error) goto out; ASSERT(uio->uio_extflg & UIO_DIRECT); out: *ioflagp = ioflag; return (error); } /* * Read bytes from specified file into supplied buffer. * * IN: zp - inode of file to be read from. * uio - structure supplying read location, range info, * and return buffer. * ioflag - O_SYNC flags; used to provide FRSYNC semantics. * O_DIRECT flag; used to bypass page cache. * cr - credentials of caller. * * OUT: uio - updated offset and range, buffer filled. * * RETURN: 0 on success, error code on failure. * * Side Effects: * inode - atime updated if byte count > 0 */ int zfs_read(struct znode *zp, zfs_uio_t *uio, int ioflag, cred_t *cr) { (void) cr; int error = 0; boolean_t frsync = B_FALSE; boolean_t dio_checksum_failure = B_FALSE; zfsvfs_t *zfsvfs = ZTOZSB(zp); if ((error = zfs_enter_verify_zp(zfsvfs, zp, FTAG)) != 0) return (error); if (zp->z_pflags & ZFS_AV_QUARANTINED) { zfs_exit(zfsvfs, FTAG); return (SET_ERROR(EACCES)); } /* We don't copy out anything useful for directories. */ if (Z_ISDIR(ZTOTYPE(zp))) { zfs_exit(zfsvfs, FTAG); return (SET_ERROR(EISDIR)); } /* * Validate file offset */ if (zfs_uio_offset(uio) < (offset_t)0) { zfs_exit(zfsvfs, FTAG); return (SET_ERROR(EINVAL)); } /* * Fasttrack empty reads */ if (zfs_uio_resid(uio) == 0) { zfs_exit(zfsvfs, FTAG); return (0); } #ifdef FRSYNC /* * If we're in FRSYNC mode, sync out this znode before reading it. * Only do this for non-snapshots. * * Some platforms do not support FRSYNC and instead map it * to O_SYNC, which results in unnecessary calls to zil_commit. We * only honor FRSYNC requests on platforms which support it. */ frsync = !!(ioflag & FRSYNC); #endif if (zfsvfs->z_log && (frsync || zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)) zil_commit(zfsvfs->z_log, zp->z_id); /* * Lock the range against changes. */ zfs_locked_range_t *lr = zfs_rangelock_enter(&zp->z_rangelock, zfs_uio_offset(uio), zfs_uio_resid(uio), RL_READER); /* * If we are reading past end-of-file we can skip * to the end; but we might still need to set atime. */ if (zfs_uio_offset(uio) >= zp->z_size) { error = 0; goto out; } ASSERT(zfs_uio_offset(uio) < zp->z_size); /* * Setting up Direct I/O if requested. */ error = zfs_setup_direct(zp, uio, UIO_READ, &ioflag); if (error) { goto out; } #if defined(__linux__) ssize_t start_offset = zfs_uio_offset(uio); #endif uint_t blksz = zp->z_blksz; ssize_t chunk_size; ssize_t n = MIN(zfs_uio_resid(uio), zp->z_size - zfs_uio_offset(uio)); ssize_t start_resid = n; ssize_t dio_remaining_resid = 0; dmu_flags_t dflags = DMU_READ_PREFETCH; if (ioflag & O_DIRECT) dflags |= DMU_UNCACHEDIO; if (uio->uio_extflg & UIO_DIRECT) { /* * All pages for an O_DIRECT request ahve already been mapped * so there's no compelling reason to handle this uio in * smaller chunks. */ chunk_size = DMU_MAX_ACCESS; /* * In the event that the O_DIRECT request is reading the entire * file, it is possible file's length is not page sized * aligned. However, lower layers expect that the Direct I/O * request is page-aligned. In this case, as much of the file * that can be read using Direct I/O happens and the remaining * amount will be read through the ARC. * * This is still consistent with the semantics of Direct I/O in * ZFS as at a minimum the I/O request must be page-aligned. */ dio_remaining_resid = n - P2ALIGN_TYPED(n, PAGE_SIZE, ssize_t); if (dio_remaining_resid != 0) n -= dio_remaining_resid; dflags |= DMU_DIRECTIO; } else { chunk_size = MIN(MAX(zfs_vnops_read_chunk_size, blksz), DMU_MAX_ACCESS / 2); } while (n > 0) { ssize_t nbytes = MIN(n, chunk_size - P2PHASE(zfs_uio_offset(uio), blksz)); #ifdef UIO_NOCOPY if (zfs_uio_segflg(uio) == UIO_NOCOPY) error = mappedread_sf(zp, nbytes, uio); else #endif if (zn_has_cached_data(zp, zfs_uio_offset(uio), zfs_uio_offset(uio) + nbytes - 1)) { error = mappedread(zp, nbytes, uio); } else { error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl), uio, nbytes, dflags); } if (error) { /* convert checksum errors into IO errors */ if (error == ECKSUM) { /* * If a Direct I/O read returned a checksum * verify error, then it must be treated as * suspicious. The contents of the buffer could * have beeen manipulated while the I/O was in * flight. In this case, the remainder of I/O * request will just be reissued through the * ARC. */ if (uio->uio_extflg & UIO_DIRECT) { dio_checksum_failure = B_TRUE; uio->uio_extflg &= ~UIO_DIRECT; n += dio_remaining_resid; dio_remaining_resid = 0; continue; } else { error = SET_ERROR(EIO); } } #if defined(__linux__) /* * if we actually read some bytes, bubbling EFAULT * up to become EAGAIN isn't what we want here... * * ...on Linux, at least. On FBSD, doing this breaks. */ if (error == EFAULT && (zfs_uio_offset(uio) - start_offset) != 0) error = 0; #endif break; } n -= nbytes; } if (error == 0 && (uio->uio_extflg & UIO_DIRECT) && dio_remaining_resid != 0) { /* * Temporarily remove the UIO_DIRECT flag from the UIO so the * remainder of the file can be read using the ARC. */ uio->uio_extflg &= ~UIO_DIRECT; dflags &= ~DMU_DIRECTIO; if (zn_has_cached_data(zp, zfs_uio_offset(uio), zfs_uio_offset(uio) + dio_remaining_resid - 1)) { error = mappedread(zp, dio_remaining_resid, uio); } else { error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl), uio, dio_remaining_resid, dflags); } uio->uio_extflg |= UIO_DIRECT; dflags |= DMU_DIRECTIO; if (error != 0) n += dio_remaining_resid; } else if (error && (uio->uio_extflg & UIO_DIRECT)) { n += dio_remaining_resid; } int64_t nread = start_resid - n; dataset_kstats_update_read_kstats(&zfsvfs->z_kstat, nread); out: zfs_rangelock_exit(lr); if (dio_checksum_failure == B_TRUE) uio->uio_extflg |= UIO_DIRECT; /* * Cleanup for Direct I/O if requested. */ if (uio->uio_extflg & UIO_DIRECT) zfs_uio_free_dio_pages(uio, UIO_READ); ZFS_ACCESSTIME_STAMP(zfsvfs, zp); zfs_exit(zfsvfs, FTAG); return (error); } static void zfs_clear_setid_bits_if_necessary(zfsvfs_t *zfsvfs, znode_t *zp, cred_t *cr, uint64_t *clear_setid_bits_txgp, dmu_tx_t *tx) { zilog_t *zilog = zfsvfs->z_log; const uint64_t uid = KUID_TO_SUID(ZTOUID(zp)); ASSERT(clear_setid_bits_txgp != NULL); ASSERT(tx != NULL); /* * Clear Set-UID/Set-GID bits on successful write if not * privileged and at least one of the execute bits is set. * * It would be nice to do this after all writes have * been done, but that would still expose the ISUID/ISGID * to another app after the partial write is committed. * * Note: we don't call zfs_fuid_map_id() here because * user 0 is not an ephemeral uid. */ mutex_enter(&zp->z_acl_lock); if ((zp->z_mode & (S_IXUSR | (S_IXUSR >> 3) | (S_IXUSR >> 6))) != 0 && (zp->z_mode & (S_ISUID | S_ISGID)) != 0 && secpolicy_vnode_setid_retain(zp, cr, ((zp->z_mode & S_ISUID) != 0 && uid == 0)) != 0) { uint64_t newmode; zp->z_mode &= ~(S_ISUID | S_ISGID); newmode = zp->z_mode; (void) sa_update(zp->z_sa_hdl, SA_ZPL_MODE(zfsvfs), (void *)&newmode, sizeof (uint64_t), tx); mutex_exit(&zp->z_acl_lock); /* * Make sure SUID/SGID bits will be removed when we replay the * log. If the setid bits are keep coming back, don't log more * than one TX_SETATTR per transaction group. */ if (*clear_setid_bits_txgp != dmu_tx_get_txg(tx)) { vattr_t va = {0}; va.va_mask = ATTR_MODE; va.va_nodeid = zp->z_id; va.va_mode = newmode; zfs_log_setattr(zilog, tx, TX_SETATTR, zp, &va, ATTR_MODE, NULL); *clear_setid_bits_txgp = dmu_tx_get_txg(tx); } } else { mutex_exit(&zp->z_acl_lock); } } /* * Write the bytes to a file. * * IN: zp - znode of file to be written to. * uio - structure supplying write location, range info, * and data buffer. * ioflag - O_APPEND flag set if in append mode. * O_DIRECT flag; used to bypass page cache. * cr - credentials of caller. * * OUT: uio - updated offset and range. * * RETURN: 0 if success * error code if failure * * Timestamps: * ip - ctime|mtime updated if byte count > 0 */ int zfs_write(znode_t *zp, zfs_uio_t *uio, int ioflag, cred_t *cr) { int error = 0, error1; ssize_t start_resid = zfs_uio_resid(uio); uint64_t clear_setid_bits_txg = 0; boolean_t o_direct_defer = B_FALSE; /* * Fasttrack empty write */ ssize_t n = start_resid; if (n == 0) return (0); zfsvfs_t *zfsvfs = ZTOZSB(zp); if ((error = zfs_enter_verify_zp(zfsvfs, zp, FTAG)) != 0) return (error); sa_bulk_attr_t bulk[4]; int count = 0; uint64_t mtime[2], ctime[2]; SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zfsvfs), NULL, &zp->z_size, 8); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL, &zp->z_pflags, 8); /* * Callers might not be able to detect properly that we are read-only, * so check it explicitly here. */ if (zfs_is_readonly(zfsvfs)) { zfs_exit(zfsvfs, FTAG); return (SET_ERROR(EROFS)); } /* * If immutable or not appending then return EPERM. * Intentionally allow ZFS_READONLY through here. * See zfs_zaccess_common() */ if ((zp->z_pflags & ZFS_IMMUTABLE) || ((zp->z_pflags & ZFS_APPENDONLY) && !(ioflag & O_APPEND) && (zfs_uio_offset(uio) < zp->z_size))) { zfs_exit(zfsvfs, FTAG); return (SET_ERROR(EPERM)); } /* * Validate file offset */ offset_t woff = ioflag & O_APPEND ? zp->z_size : zfs_uio_offset(uio); if (woff < 0) { zfs_exit(zfsvfs, FTAG); return (SET_ERROR(EINVAL)); } /* * Setting up Direct I/O if requested. */ error = zfs_setup_direct(zp, uio, UIO_WRITE, &ioflag); if (error) { zfs_exit(zfsvfs, FTAG); return (SET_ERROR(error)); } /* * Pre-fault the pages to ensure slow (eg NFS) pages * don't hold up txg. */ ssize_t pfbytes = MIN(n, DMU_MAX_ACCESS >> 1); if (zfs_uio_prefaultpages(pfbytes, uio)) { zfs_exit(zfsvfs, FTAG); return (SET_ERROR(EFAULT)); } /* * If in append mode, set the io offset pointer to eof. */ zfs_locked_range_t *lr; if (ioflag & O_APPEND) { /* * Obtain an appending range lock to guarantee file append * semantics. We reset the write offset once we have the lock. */ lr = zfs_rangelock_enter(&zp->z_rangelock, 0, n, RL_APPEND); woff = lr->lr_offset; if (lr->lr_length == UINT64_MAX) { /* * We overlocked the file because this write will cause * the file block size to increase. * Note that zp_size cannot change with this lock held. */ woff = zp->z_size; } zfs_uio_setoffset(uio, woff); /* * We need to update the starting offset as well because it is * set previously in the ZPL (Linux) and VNOPS (FreeBSD) * layers. */ zfs_uio_setsoffset(uio, woff); } else { /* * Note that if the file block size will change as a result of * this write, then this range lock will lock the entire file * so that we can re-write the block safely. */ lr = zfs_rangelock_enter(&zp->z_rangelock, woff, n, RL_WRITER); } if (zn_rlimit_fsize_uio(zp, uio)) { zfs_rangelock_exit(lr); zfs_exit(zfsvfs, FTAG); return (SET_ERROR(EFBIG)); } const rlim64_t limit = MAXOFFSET_T; if (woff >= limit) { zfs_rangelock_exit(lr); zfs_exit(zfsvfs, FTAG); return (SET_ERROR(EFBIG)); } if (n > limit - woff) n = limit - woff; uint64_t end_size = MAX(zp->z_size, woff + n); zilog_t *zilog = zfsvfs->z_log; boolean_t commit = (ioflag & (O_SYNC | O_DSYNC)) || (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS); const uint64_t uid = KUID_TO_SUID(ZTOUID(zp)); const uint64_t gid = KGID_TO_SGID(ZTOGID(zp)); const uint64_t projid = zp->z_projid; /* * In the event we are increasing the file block size * (lr_length == UINT64_MAX), we will direct the write to the ARC. * Because zfs_grow_blocksize() will read from the ARC in order to * grow the dbuf, we avoid doing Direct I/O here as that would cause * data written to disk to be overwritten by data in the ARC during * the sync phase. Besides writing data twice to disk, we also * want to avoid consistency concerns between data in the the ARC and * on disk while growing the file's blocksize. * * We will only temporarily remove Direct I/O and put it back after * we have grown the blocksize. We do this in the event a request * is larger than max_blksz, so further requests to * dmu_write_uio_dbuf() will still issue the requests using Direct * IO. * * As an example: * The first block to file is being written as a 4k request with * a recorsize of 1K. The first 1K issued in the loop below will go * through the ARC; however, the following 3 1K requests will * use Direct I/O. */ if (uio->uio_extflg & UIO_DIRECT && lr->lr_length == UINT64_MAX) { uio->uio_extflg &= ~UIO_DIRECT; o_direct_defer = B_TRUE; } /* * Write the file in reasonable size chunks. Each chunk is written * in a separate transaction; this keeps the intent log records small * and allows us to do more fine-grained space accounting. */ while (n > 0) { woff = zfs_uio_offset(uio); if (zfs_id_overblockquota(zfsvfs, DMU_USERUSED_OBJECT, uid) || zfs_id_overblockquota(zfsvfs, DMU_GROUPUSED_OBJECT, gid) || (projid != ZFS_DEFAULT_PROJID && zfs_id_overblockquota(zfsvfs, DMU_PROJECTUSED_OBJECT, projid))) { error = SET_ERROR(EDQUOT); break; } uint64_t blksz; if (lr->lr_length == UINT64_MAX && zp->z_size <= zp->z_blksz) { if (zp->z_blksz > zfsvfs->z_max_blksz && !ISP2(zp->z_blksz)) { /* * File's blocksize is already larger than the * "recordsize" property. Only let it grow to * the next power of 2. */ blksz = 1 << highbit64(zp->z_blksz); } else { blksz = zfsvfs->z_max_blksz; } blksz = MIN(blksz, P2ROUNDUP(end_size, SPA_MINBLOCKSIZE)); blksz = MAX(blksz, zp->z_blksz); } else { blksz = zp->z_blksz; } arc_buf_t *abuf = NULL; ssize_t nbytes = n; if (n >= blksz && woff >= zp->z_size && P2PHASE(woff, blksz) == 0 && !(uio->uio_extflg & UIO_DIRECT) && (blksz >= SPA_OLD_MAXBLOCKSIZE || n < 4 * blksz)) { /* * This write covers a full block. "Borrow" a buffer * from the dmu so that we can fill it before we enter * a transaction. This avoids the possibility of * holding up the transaction if the data copy hangs * up on a pagefault (e.g., from an NFS server mapping). */ abuf = dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl), blksz); ASSERT(abuf != NULL); ASSERT(arc_buf_size(abuf) == blksz); if ((error = zfs_uiocopy(abuf->b_data, blksz, UIO_WRITE, uio, &nbytes))) { dmu_return_arcbuf(abuf); break; } ASSERT3S(nbytes, ==, blksz); } else { nbytes = MIN(n, (DMU_MAX_ACCESS >> 1) - P2PHASE(woff, blksz)); if (pfbytes < nbytes) { if (zfs_uio_prefaultpages(nbytes, uio)) { error = SET_ERROR(EFAULT); break; } pfbytes = nbytes; } } /* * Start a transaction. */ dmu_tx_t *tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); dmu_buf_impl_t *db = (dmu_buf_impl_t *)sa_get_db(zp->z_sa_hdl); DB_DNODE_ENTER(db); dmu_tx_hold_write_by_dnode(tx, DB_DNODE(db), woff, nbytes); DB_DNODE_EXIT(db); zfs_sa_upgrade_txholds(tx, zp); error = dmu_tx_assign(tx, DMU_TX_WAIT); if (error) { dmu_tx_abort(tx); if (abuf != NULL) dmu_return_arcbuf(abuf); break; } /* * NB: We must call zfs_clear_setid_bits_if_necessary before * committing the transaction! */ /* * If rangelock_enter() over-locked we grow the blocksize * and then reduce the lock range. This will only happen * on the first iteration since rangelock_reduce() will * shrink down lr_length to the appropriate size. */ if (lr->lr_length == UINT64_MAX) { zfs_grow_blocksize(zp, blksz, tx); zfs_rangelock_reduce(lr, woff, n); } dmu_flags_t dflags = DMU_READ_PREFETCH; if (ioflag & O_DIRECT) dflags |= DMU_UNCACHEDIO; if (uio->uio_extflg & UIO_DIRECT) dflags |= DMU_DIRECTIO; ssize_t tx_bytes; if (abuf == NULL) { tx_bytes = zfs_uio_resid(uio); zfs_uio_fault_disable(uio, B_TRUE); error = dmu_write_uio_dbuf(sa_get_db(zp->z_sa_hdl), uio, nbytes, tx, dflags); zfs_uio_fault_disable(uio, B_FALSE); #ifdef __linux__ if (error == EFAULT) { zfs_clear_setid_bits_if_necessary(zfsvfs, zp, cr, &clear_setid_bits_txg, tx); dmu_tx_commit(tx); /* * Account for partial writes before * continuing the loop. * Update needs to occur before the next * zfs_uio_prefaultpages, or prefaultpages may * error, and we may break the loop early. */ n -= tx_bytes - zfs_uio_resid(uio); pfbytes -= tx_bytes - zfs_uio_resid(uio); continue; } #endif /* * On FreeBSD, EFAULT should be propagated back to the * VFS, which will handle faulting and will retry. */ if (error != 0 && error != EFAULT) { zfs_clear_setid_bits_if_necessary(zfsvfs, zp, cr, &clear_setid_bits_txg, tx); dmu_tx_commit(tx); break; } tx_bytes -= zfs_uio_resid(uio); } else { /* * Thus, we're writing a full block at a block-aligned * offset and extending the file past EOF. * * dmu_assign_arcbuf_by_dbuf() will directly assign the * arc buffer to a dbuf. */ error = dmu_assign_arcbuf_by_dbuf( sa_get_db(zp->z_sa_hdl), woff, abuf, tx, dflags); if (error != 0) { /* * XXX This might not be necessary if * dmu_assign_arcbuf_by_dbuf is guaranteed * to be atomic. */ zfs_clear_setid_bits_if_necessary(zfsvfs, zp, cr, &clear_setid_bits_txg, tx); dmu_return_arcbuf(abuf); dmu_tx_commit(tx); break; } ASSERT3S(nbytes, <=, zfs_uio_resid(uio)); zfs_uioskip(uio, nbytes); tx_bytes = nbytes; } /* * There is a window where a file's pages can be mmap'ed after * zfs_setup_direct() is called. This is due to the fact that * the rangelock in this function is acquired after calling * zfs_setup_direct(). This is done so that * zfs_uio_prefaultpages() does not attempt to fault in pages * on Linux for Direct I/O requests. This is not necessary as * the pages are pinned in memory and can not be faulted out. * Ideally, the rangelock would be held before calling * zfs_setup_direct() and zfs_uio_prefaultpages(); however, * this can lead to a deadlock as zfs_getpage() also acquires * the rangelock as a RL_WRITER and prefaulting the pages can * lead to zfs_getpage() being called. * * In the case of the pages being mapped after * zfs_setup_direct() is called, the call to update_pages() * will still be made to make sure there is consistency between * the ARC and the Linux page cache. This is an ufortunate * situation as the data will be read back into the ARC after * the Direct I/O write has completed, but this is the penality * for writing to a mmap'ed region of a file using Direct I/O. */ if (tx_bytes && zn_has_cached_data(zp, woff, woff + tx_bytes - 1)) { update_pages(zp, woff, tx_bytes, zfsvfs->z_os); } /* * If we made no progress, we're done. If we made even * partial progress, update the znode and ZIL accordingly. */ if (tx_bytes == 0) { (void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs), (void *)&zp->z_size, sizeof (uint64_t), tx); dmu_tx_commit(tx); ASSERT(error != 0); break; } zfs_clear_setid_bits_if_necessary(zfsvfs, zp, cr, &clear_setid_bits_txg, tx); zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime); /* * Update the file size (zp_size) if it has changed; * account for possible concurrent updates. */ while ((end_size = zp->z_size) < zfs_uio_offset(uio)) { (void) atomic_cas_64(&zp->z_size, end_size, zfs_uio_offset(uio)); ASSERT(error == 0 || error == EFAULT); } /* * If we are replaying and eof is non zero then force * the file size to the specified eof. Note, there's no * concurrency during replay. */ if (zfsvfs->z_replay && zfsvfs->z_replay_eof != 0) zp->z_size = zfsvfs->z_replay_eof; error1 = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx); if (error1 != 0) /* Avoid clobbering EFAULT. */ error = error1; /* * NB: During replay, the TX_SETATTR record logged by * zfs_clear_setid_bits_if_necessary must precede any of * the TX_WRITE records logged here. */ zfs_log_write(zilog, tx, TX_WRITE, zp, woff, tx_bytes, commit, uio->uio_extflg & UIO_DIRECT ? B_TRUE : B_FALSE, NULL, NULL); dmu_tx_commit(tx); /* * Direct I/O was deferred in order to grow the first block. * At this point it can be re-enabled for subsequent writes. */ if (o_direct_defer) { ASSERT(ioflag & O_DIRECT); uio->uio_extflg |= UIO_DIRECT; o_direct_defer = B_FALSE; } if (error != 0) break; ASSERT3S(tx_bytes, ==, nbytes); n -= nbytes; pfbytes -= nbytes; } if (o_direct_defer) { ASSERT(ioflag & O_DIRECT); uio->uio_extflg |= UIO_DIRECT; o_direct_defer = B_FALSE; } zfs_znode_update_vfs(zp); zfs_rangelock_exit(lr); /* * Cleanup for Direct I/O if requested. */ if (uio->uio_extflg & UIO_DIRECT) zfs_uio_free_dio_pages(uio, UIO_WRITE); /* * If we're in replay mode, or we made no progress, or the * uio data is inaccessible return an error. Otherwise, it's * at least a partial write, so it's successful. */ if (zfsvfs->z_replay || zfs_uio_resid(uio) == start_resid || error == EFAULT) { zfs_exit(zfsvfs, FTAG); return (error); } if (commit) zil_commit(zilog, zp->z_id); int64_t nwritten = start_resid - zfs_uio_resid(uio); dataset_kstats_update_write_kstats(&zfsvfs->z_kstat, nwritten); zfs_exit(zfsvfs, FTAG); return (0); } /* * Rewrite a range of file as-is without modification. * * IN: zp - znode of file to be rewritten. * off - Offset of the range to rewrite. * len - Length of the range to rewrite. * flags - Random rewrite parameters. * arg - flags-specific argument. * * RETURN: 0 if success * error code if failure */ int zfs_rewrite(znode_t *zp, uint64_t off, uint64_t len, uint64_t flags, uint64_t arg) { int error; - if (flags != 0 || arg != 0) + if ((flags & ~ZFS_REWRITE_PHYSICAL) != 0 || arg != 0) return (SET_ERROR(EINVAL)); zfsvfs_t *zfsvfs = ZTOZSB(zp); if ((error = zfs_enter_verify_zp(zfsvfs, zp, FTAG)) != 0) return (error); + /* Check if physical rewrite is allowed */ + spa_t *spa = zfsvfs->z_os->os_spa; + if ((flags & ZFS_REWRITE_PHYSICAL) && + !spa_feature_is_enabled(spa, SPA_FEATURE_PHYSICAL_REWRITE)) { + zfs_exit(zfsvfs, FTAG); + return (SET_ERROR(ENOTSUP)); + } + if (zfs_is_readonly(zfsvfs)) { zfs_exit(zfsvfs, FTAG); return (SET_ERROR(EROFS)); } if (off >= zp->z_size) { zfs_exit(zfsvfs, FTAG); return (0); } if (len == 0 || len > zp->z_size - off) len = zp->z_size - off; /* Flush any mmap()'d data to disk */ if (zn_has_cached_data(zp, off, off + len - 1)) zn_flush_cached_data(zp, B_TRUE); zfs_locked_range_t *lr; lr = zfs_rangelock_enter(&zp->z_rangelock, off, len, RL_WRITER); const uint64_t uid = KUID_TO_SUID(ZTOUID(zp)); const uint64_t gid = KGID_TO_SGID(ZTOGID(zp)); const uint64_t projid = zp->z_projid; dmu_buf_impl_t *db = (dmu_buf_impl_t *)sa_get_db(zp->z_sa_hdl); DB_DNODE_ENTER(db); dnode_t *dn = DB_DNODE(db); uint64_t n, noff = off, nr = 0, nw = 0; while (len > 0) { /* * Rewrite only actual data, skipping any holes. This might * be inaccurate for dirty files, but we don't really care. */ if (noff == off) { /* Find next data in the file. */ error = dnode_next_offset(dn, 0, &noff, 1, 1, 0); if (error || noff >= off + len) { if (error == ESRCH) /* No more data. */ error = 0; break; } ASSERT3U(noff, >=, off); len -= noff - off; off = noff; /* Find where the data end. */ error = dnode_next_offset(dn, DNODE_FIND_HOLE, &noff, 1, 1, 0); if (error != 0) noff = off + len; } ASSERT3U(noff, >, off); if (zfs_id_overblockquota(zfsvfs, DMU_USERUSED_OBJECT, uid) || zfs_id_overblockquota(zfsvfs, DMU_GROUPUSED_OBJECT, gid) || (projid != ZFS_DEFAULT_PROJID && zfs_id_overblockquota(zfsvfs, DMU_PROJECTUSED_OBJECT, projid))) { error = SET_ERROR(EDQUOT); break; } n = MIN(MIN(len, noff - off), DMU_MAX_ACCESS / 2 - P2PHASE(off, zp->z_blksz)); dmu_tx_t *tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_write_by_dnode(tx, dn, off, n); error = dmu_tx_assign(tx, DMU_TX_WAIT); if (error) { dmu_tx_abort(tx); break; } /* Mark all dbufs within range as dirty to trigger rewrite. */ dmu_buf_t **dbp; int numbufs; error = dmu_buf_hold_array_by_dnode(dn, off, n, TRUE, FTAG, &numbufs, &dbp, DMU_READ_PREFETCH | DMU_UNCACHEDIO); if (error) { dmu_tx_commit(tx); break; } for (int i = 0; i < numbufs; i++) { nr += dbp[i]->db_size; if (dmu_buf_is_dirty(dbp[i], tx)) continue; nw += dbp[i]->db_size; - dmu_buf_will_dirty(dbp[i], tx); + if (flags & ZFS_REWRITE_PHYSICAL) + dmu_buf_will_rewrite(dbp[i], tx); + else + dmu_buf_will_dirty(dbp[i], tx); } dmu_buf_rele_array(dbp, numbufs, FTAG); dmu_tx_commit(tx); len -= n; off += n; if (issig()) { error = SET_ERROR(EINTR); break; } } DB_DNODE_EXIT(db); dataset_kstats_update_read_kstats(&zfsvfs->z_kstat, nr); dataset_kstats_update_write_kstats(&zfsvfs->z_kstat, nw); zfs_rangelock_exit(lr); zfs_exit(zfsvfs, FTAG); return (error); } int zfs_getsecattr(znode_t *zp, vsecattr_t *vsecp, int flag, cred_t *cr) { zfsvfs_t *zfsvfs = ZTOZSB(zp); int error; boolean_t skipaclchk = (flag & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE; if ((error = zfs_enter_verify_zp(zfsvfs, zp, FTAG)) != 0) return (error); error = zfs_getacl(zp, vsecp, skipaclchk, cr); zfs_exit(zfsvfs, FTAG); return (error); } int zfs_setsecattr(znode_t *zp, vsecattr_t *vsecp, int flag, cred_t *cr) { zfsvfs_t *zfsvfs = ZTOZSB(zp); int error; boolean_t skipaclchk = (flag & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE; zilog_t *zilog; if ((error = zfs_enter_verify_zp(zfsvfs, zp, FTAG)) != 0) return (error); zilog = zfsvfs->z_log; error = zfs_setacl(zp, vsecp, skipaclchk, cr); if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); zfs_exit(zfsvfs, FTAG); return (error); } /* * Get the optimal alignment to ensure direct IO can be performed without * incurring any RMW penalty on write. If direct IO is not enabled for this * file, returns an error. */ int zfs_get_direct_alignment(znode_t *zp, uint64_t *alignp) { zfsvfs_t *zfsvfs = ZTOZSB(zp); if (!zfs_dio_enabled || zfsvfs->z_os->os_direct == ZFS_DIRECT_DISABLED) return (SET_ERROR(EOPNOTSUPP)); /* * If the file has multiple blocks, then its block size is fixed * forever, and so is the ideal alignment. * * If however it only has a single block, then we want to return the * max block size it could possibly grown to (ie, the dataset * recordsize). We do this so that a program querying alignment * immediately after the file is created gets a value that won't change * once the file has grown into the second block and beyond. * * Because we don't have a count of blocks easily available here, we * check if the apparent file size is smaller than its current block * size (meaning, the file hasn't yet grown into the current block * size) and then, check if the block size is smaller than the dataset * maximum (meaning, if the file grew past the current block size, the * block size could would be increased). */ if (zp->z_size <= zp->z_blksz && zp->z_blksz < zfsvfs->z_max_blksz) *alignp = MAX(zfsvfs->z_max_blksz, PAGE_SIZE); else *alignp = MAX(zp->z_blksz, PAGE_SIZE); return (0); } #ifdef ZFS_DEBUG static int zil_fault_io = 0; #endif static void zfs_get_done(zgd_t *zgd, int error); /* * Get data to generate a TX_WRITE intent log record. */ int zfs_get_data(void *arg, uint64_t gen, lr_write_t *lr, char *buf, struct lwb *lwb, zio_t *zio) { zfsvfs_t *zfsvfs = arg; objset_t *os = zfsvfs->z_os; znode_t *zp; uint64_t object = lr->lr_foid; uint64_t offset = lr->lr_offset; uint64_t size = lr->lr_length; zgd_t *zgd; int error = 0; uint64_t zp_gen; ASSERT3P(lwb, !=, NULL); ASSERT3U(size, !=, 0); /* * Nothing to do if the file has been removed */ if (zfs_zget(zfsvfs, object, &zp) != 0) return (SET_ERROR(ENOENT)); if (zp->z_unlinked) { /* * Release the vnode asynchronously as we currently have the * txg stopped from syncing. */ zfs_zrele_async(zp); return (SET_ERROR(ENOENT)); } /* check if generation number matches */ if (sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zfsvfs), &zp_gen, sizeof (zp_gen)) != 0) { zfs_zrele_async(zp); return (SET_ERROR(EIO)); } if (zp_gen != gen) { zfs_zrele_async(zp); return (SET_ERROR(ENOENT)); } zgd = kmem_zalloc(sizeof (zgd_t), KM_SLEEP); zgd->zgd_lwb = lwb; zgd->zgd_private = zp; /* * Write records come in two flavors: immediate and indirect. * For small writes it's cheaper to store the data with the * log record (immediate); for large writes it's cheaper to * sync the data and get a pointer to it (indirect) so that * we don't have to write the data twice. */ if (buf != NULL) { /* immediate write */ zgd->zgd_lr = zfs_rangelock_enter(&zp->z_rangelock, offset, size, RL_READER); /* test for truncation needs to be done while range locked */ if (offset >= zp->z_size) { error = SET_ERROR(ENOENT); } else { error = dmu_read(os, object, offset, size, buf, DMU_READ_NO_PREFETCH | DMU_KEEP_CACHING); } ASSERT(error == 0 || error == ENOENT); } else { /* indirect write */ ASSERT3P(zio, !=, NULL); /* * Have to lock the whole block to ensure when it's * written out and its checksum is being calculated * that no one can change the data. We need to re-check * blocksize after we get the lock in case it's changed! */ for (;;) { uint64_t blkoff; size = zp->z_blksz; blkoff = ISP2(size) ? P2PHASE(offset, size) : offset; offset -= blkoff; zgd->zgd_lr = zfs_rangelock_enter(&zp->z_rangelock, offset, size, RL_READER); if (zp->z_blksz == size) break; offset += blkoff; zfs_rangelock_exit(zgd->zgd_lr); } /* test for truncation needs to be done while range locked */ if (lr->lr_offset >= zp->z_size) error = SET_ERROR(ENOENT); #ifdef ZFS_DEBUG if (zil_fault_io) { error = SET_ERROR(EIO); zil_fault_io = 0; } #endif dmu_buf_t *dbp; if (error == 0) error = dmu_buf_hold_noread(os, object, offset, zgd, &dbp); if (error == 0) { zgd->zgd_db = dbp; dmu_buf_impl_t *db = (dmu_buf_impl_t *)dbp; boolean_t direct_write = B_FALSE; mutex_enter(&db->db_mtx); dbuf_dirty_record_t *dr = dbuf_find_dirty_eq(db, lr->lr_common.lrc_txg); if (dr != NULL && dr->dt.dl.dr_diowrite) direct_write = B_TRUE; mutex_exit(&db->db_mtx); /* * All Direct I/O writes will have already completed and * the block pointer can be immediately stored in the * log record. */ if (direct_write) { /* * A Direct I/O write always covers an entire * block. */ ASSERT3U(dbp->db_size, ==, zp->z_blksz); lr->lr_blkptr = dr->dt.dl.dr_overridden_by; zfs_get_done(zgd, 0); return (0); } blkptr_t *bp = &lr->lr_blkptr; zgd->zgd_bp = bp; ASSERT3U(dbp->db_offset, ==, offset); ASSERT3U(dbp->db_size, ==, size); error = dmu_sync(zio, lr->lr_common.lrc_txg, zfs_get_done, zgd); ASSERT(error || lr->lr_length <= size); /* * On success, we need to wait for the write I/O * initiated by dmu_sync() to complete before we can * release this dbuf. We will finish everything up * in the zfs_get_done() callback. */ if (error == 0) return (0); if (error == EALREADY) { lr->lr_common.lrc_txtype = TX_WRITE2; /* * TX_WRITE2 relies on the data previously * written by the TX_WRITE that caused * EALREADY. We zero out the BP because * it is the old, currently-on-disk BP. */ zgd->zgd_bp = NULL; BP_ZERO(bp); error = 0; } } } zfs_get_done(zgd, error); return (error); } static void zfs_get_done(zgd_t *zgd, int error) { (void) error; znode_t *zp = zgd->zgd_private; if (zgd->zgd_db) dmu_buf_rele(zgd->zgd_db, zgd); zfs_rangelock_exit(zgd->zgd_lr); /* * Release the vnode asynchronously as we currently have the * txg stopped from syncing. */ zfs_zrele_async(zp); kmem_free(zgd, sizeof (zgd_t)); } static int zfs_enter_two(zfsvfs_t *zfsvfs1, zfsvfs_t *zfsvfs2, const char *tag) { int error; /* Swap. Not sure if the order of zfs_enter()s is important. */ if (zfsvfs1 > zfsvfs2) { zfsvfs_t *tmpzfsvfs; tmpzfsvfs = zfsvfs2; zfsvfs2 = zfsvfs1; zfsvfs1 = tmpzfsvfs; } error = zfs_enter(zfsvfs1, tag); if (error != 0) return (error); if (zfsvfs1 != zfsvfs2) { error = zfs_enter(zfsvfs2, tag); if (error != 0) { zfs_exit(zfsvfs1, tag); return (error); } } return (0); } static void zfs_exit_two(zfsvfs_t *zfsvfs1, zfsvfs_t *zfsvfs2, const char *tag) { zfs_exit(zfsvfs1, tag); if (zfsvfs1 != zfsvfs2) zfs_exit(zfsvfs2, tag); } /* * We split each clone request in chunks that can fit into a single ZIL * log entry. Each ZIL log entry can fit 130816 bytes for a block cloning * operation (see zil_max_log_data() and zfs_log_clone_range()). This gives * us room for storing 1022 block pointers. * * On success, the function return the number of bytes copied in *lenp. * Note, it doesn't return how much bytes are left to be copied. * On errors which are caused by any file system limitations or * brt limitations `EINVAL` is returned. In the most cases a user * requested bad parameters, it could be possible to clone the file but * some parameters don't match the requirements. */ int zfs_clone_range(znode_t *inzp, uint64_t *inoffp, znode_t *outzp, uint64_t *outoffp, uint64_t *lenp, cred_t *cr) { zfsvfs_t *inzfsvfs, *outzfsvfs; objset_t *inos, *outos; zfs_locked_range_t *inlr, *outlr; dmu_buf_impl_t *db; dmu_tx_t *tx; zilog_t *zilog; uint64_t inoff, outoff, len, done; uint64_t outsize, size; int error; int count = 0; sa_bulk_attr_t bulk[3]; uint64_t mtime[2], ctime[2]; uint64_t uid, gid, projid; blkptr_t *bps; size_t maxblocks, nbps; uint_t inblksz; uint64_t clear_setid_bits_txg = 0; uint64_t last_synced_txg = 0; inoff = *inoffp; outoff = *outoffp; len = *lenp; done = 0; inzfsvfs = ZTOZSB(inzp); outzfsvfs = ZTOZSB(outzp); /* * We need to call zfs_enter() potentially on two different datasets, * so we need a dedicated function for that. */ error = zfs_enter_two(inzfsvfs, outzfsvfs, FTAG); if (error != 0) return (error); inos = inzfsvfs->z_os; outos = outzfsvfs->z_os; /* * Both source and destination have to belong to the same storage pool. */ if (dmu_objset_spa(inos) != dmu_objset_spa(outos)) { zfs_exit_two(inzfsvfs, outzfsvfs, FTAG); return (SET_ERROR(EXDEV)); } /* * outos and inos belongs to the same storage pool. * see a few lines above, only one check. */ if (!spa_feature_is_enabled(dmu_objset_spa(outos), SPA_FEATURE_BLOCK_CLONING)) { zfs_exit_two(inzfsvfs, outzfsvfs, FTAG); return (SET_ERROR(EOPNOTSUPP)); } ASSERT(!outzfsvfs->z_replay); /* * Block cloning from an unencrypted dataset into an encrypted * dataset and vice versa is not supported. */ if (inos->os_encrypted != outos->os_encrypted) { zfs_exit_two(inzfsvfs, outzfsvfs, FTAG); return (SET_ERROR(EXDEV)); } /* * Cloning across encrypted datasets is possible only if they * share the same master key. */ if (inos != outos && inos->os_encrypted && !dmu_objset_crypto_key_equal(inos, outos)) { zfs_exit_two(inzfsvfs, outzfsvfs, FTAG); return (SET_ERROR(EXDEV)); } error = zfs_verify_zp(inzp); if (error == 0) error = zfs_verify_zp(outzp); if (error != 0) { zfs_exit_two(inzfsvfs, outzfsvfs, FTAG); return (error); } /* * We don't copy source file's flags that's why we don't allow to clone * files that are in quarantine. */ if (inzp->z_pflags & ZFS_AV_QUARANTINED) { zfs_exit_two(inzfsvfs, outzfsvfs, FTAG); return (SET_ERROR(EACCES)); } if (inoff >= inzp->z_size) { *lenp = 0; zfs_exit_two(inzfsvfs, outzfsvfs, FTAG); return (0); } if (len > inzp->z_size - inoff) { len = inzp->z_size - inoff; } if (len == 0) { *lenp = 0; zfs_exit_two(inzfsvfs, outzfsvfs, FTAG); return (0); } /* * Callers might not be able to detect properly that we are read-only, * so check it explicitly here. */ if (zfs_is_readonly(outzfsvfs)) { zfs_exit_two(inzfsvfs, outzfsvfs, FTAG); return (SET_ERROR(EROFS)); } /* * If immutable or not appending then return EPERM. * Intentionally allow ZFS_READONLY through here. * See zfs_zaccess_common() */ if ((outzp->z_pflags & ZFS_IMMUTABLE) != 0) { zfs_exit_two(inzfsvfs, outzfsvfs, FTAG); return (SET_ERROR(EPERM)); } /* * No overlapping if we are cloning within the same file. */ if (inzp == outzp) { if (inoff < outoff + len && outoff < inoff + len) { zfs_exit_two(inzfsvfs, outzfsvfs, FTAG); return (SET_ERROR(EINVAL)); } } /* Flush any mmap()'d data to disk */ if (zn_has_cached_data(inzp, inoff, inoff + len - 1)) zn_flush_cached_data(inzp, B_TRUE); /* * Maintain predictable lock order. */ if (inzp < outzp || (inzp == outzp && inoff < outoff)) { inlr = zfs_rangelock_enter(&inzp->z_rangelock, inoff, len, RL_READER); outlr = zfs_rangelock_enter(&outzp->z_rangelock, outoff, len, RL_WRITER); } else { outlr = zfs_rangelock_enter(&outzp->z_rangelock, outoff, len, RL_WRITER); inlr = zfs_rangelock_enter(&inzp->z_rangelock, inoff, len, RL_READER); } inblksz = inzp->z_blksz; /* * We cannot clone into a file with different block size if we can't * grow it (block size is already bigger, has more than one block, or * not locked for growth). There are other possible reasons for the * grow to fail, but we cover what we can before opening transaction * and the rest detect after we try to do it. */ if (inblksz < outzp->z_blksz) { error = SET_ERROR(EINVAL); goto unlock; } if (inblksz != outzp->z_blksz && (outzp->z_size > outzp->z_blksz || outlr->lr_length != UINT64_MAX)) { error = SET_ERROR(EINVAL); goto unlock; } /* * Block size must be power-of-2 if destination offset != 0. * There can be no multiple blocks of non-power-of-2 size. */ if (outoff != 0 && !ISP2(inblksz)) { error = SET_ERROR(EINVAL); goto unlock; } /* * Offsets and len must be at block boundries. */ if ((inoff % inblksz) != 0 || (outoff % inblksz) != 0) { error = SET_ERROR(EINVAL); goto unlock; } /* * Length must be multipe of blksz, except for the end of the file. */ if ((len % inblksz) != 0 && (len < inzp->z_size - inoff || len < outzp->z_size - outoff)) { error = SET_ERROR(EINVAL); goto unlock; } /* * If we are copying only one block and it is smaller than recordsize * property, do not allow destination to grow beyond one block if it * is not there yet. Otherwise the destination will get stuck with * that block size forever, that can be as small as 512 bytes, no * matter how big the destination grow later. */ if (len <= inblksz && inblksz < outzfsvfs->z_max_blksz && outzp->z_size <= inblksz && outoff + len > inblksz) { error = SET_ERROR(EINVAL); goto unlock; } error = zn_rlimit_fsize(outoff + len); if (error != 0) { goto unlock; } if (inoff >= MAXOFFSET_T || outoff >= MAXOFFSET_T) { error = SET_ERROR(EFBIG); goto unlock; } SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(outzfsvfs), NULL, &mtime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(outzfsvfs), NULL, &ctime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(outzfsvfs), NULL, &outzp->z_size, 8); zilog = outzfsvfs->z_log; maxblocks = zil_max_log_data(zilog, sizeof (lr_clone_range_t)) / sizeof (bps[0]); uid = KUID_TO_SUID(ZTOUID(outzp)); gid = KGID_TO_SGID(ZTOGID(outzp)); projid = outzp->z_projid; bps = vmem_alloc(sizeof (bps[0]) * maxblocks, KM_SLEEP); /* * Clone the file in reasonable size chunks. Each chunk is cloned * in a separate transaction; this keeps the intent log records small * and allows us to do more fine-grained space accounting. */ while (len > 0) { size = MIN(inblksz * maxblocks, len); if (zfs_id_overblockquota(outzfsvfs, DMU_USERUSED_OBJECT, uid) || zfs_id_overblockquota(outzfsvfs, DMU_GROUPUSED_OBJECT, gid) || (projid != ZFS_DEFAULT_PROJID && zfs_id_overblockquota(outzfsvfs, DMU_PROJECTUSED_OBJECT, projid))) { error = SET_ERROR(EDQUOT); break; } nbps = maxblocks; last_synced_txg = spa_last_synced_txg(dmu_objset_spa(inos)); error = dmu_read_l0_bps(inos, inzp->z_id, inoff, size, bps, &nbps); if (error != 0) { /* * If we are trying to clone a block that was created * in the current transaction group, the error will be * EAGAIN here. Based on zfs_bclone_wait_dirty either * return a shortened range to the caller so it can * fallback, or wait for the next TXG and check again. */ if (error == EAGAIN && zfs_bclone_wait_dirty) { txg_wait_flag_t wait_flags = spa_get_failmode(dmu_objset_spa(inos)) == ZIO_FAILURE_MODE_CONTINUE ? TXG_WAIT_SUSPEND : 0; error = txg_wait_synced_flags( dmu_objset_pool(inos), last_synced_txg + 1, wait_flags); if (error == 0) continue; ASSERT3U(error, ==, ESHUTDOWN); error = SET_ERROR(EIO); } break; } /* * Start a transaction. */ tx = dmu_tx_create(outos); dmu_tx_hold_sa(tx, outzp->z_sa_hdl, B_FALSE); db = (dmu_buf_impl_t *)sa_get_db(outzp->z_sa_hdl); DB_DNODE_ENTER(db); dmu_tx_hold_clone_by_dnode(tx, DB_DNODE(db), outoff, size, inblksz); DB_DNODE_EXIT(db); zfs_sa_upgrade_txholds(tx, outzp); error = dmu_tx_assign(tx, DMU_TX_WAIT); if (error != 0) { dmu_tx_abort(tx); break; } /* * Copy source znode's block size. This is done only if the * whole znode is locked (see zfs_rangelock_cb()) and only * on the first iteration since zfs_rangelock_reduce() will * shrink down lr_length to the appropriate size. */ if (outlr->lr_length == UINT64_MAX) { zfs_grow_blocksize(outzp, inblksz, tx); /* * Block growth may fail for many reasons we can not * predict here. If it happen the cloning is doomed. */ if (inblksz != outzp->z_blksz) { error = SET_ERROR(EINVAL); dmu_tx_commit(tx); break; } /* * Round range lock up to the block boundary, so we * prevent appends until we are done. */ zfs_rangelock_reduce(outlr, outoff, ((len - 1) / inblksz + 1) * inblksz); } error = dmu_brt_clone(outos, outzp->z_id, outoff, size, tx, bps, nbps); if (error != 0) { dmu_tx_commit(tx); break; } if (zn_has_cached_data(outzp, outoff, outoff + size - 1)) { update_pages(outzp, outoff, size, outos); } zfs_clear_setid_bits_if_necessary(outzfsvfs, outzp, cr, &clear_setid_bits_txg, tx); zfs_tstamp_update_setup(outzp, CONTENT_MODIFIED, mtime, ctime); /* * Update the file size (zp_size) if it has changed; * account for possible concurrent updates. */ while ((outsize = outzp->z_size) < outoff + size) { (void) atomic_cas_64(&outzp->z_size, outsize, outoff + size); } error = sa_bulk_update(outzp->z_sa_hdl, bulk, count, tx); zfs_log_clone_range(zilog, tx, TX_CLONE_RANGE, outzp, outoff, size, inblksz, bps, nbps); dmu_tx_commit(tx); if (error != 0) break; inoff += size; outoff += size; len -= size; done += size; if (issig()) { error = SET_ERROR(EINTR); break; } } vmem_free(bps, sizeof (bps[0]) * maxblocks); zfs_znode_update_vfs(outzp); unlock: zfs_rangelock_exit(outlr); zfs_rangelock_exit(inlr); if (done > 0) { /* * If we have made at least partial progress, reset the error. */ error = 0; ZFS_ACCESSTIME_STAMP(inzfsvfs, inzp); if (outos->os_sync == ZFS_SYNC_ALWAYS) { zil_commit(zilog, outzp->z_id); } *inoffp += done; *outoffp += done; *lenp = done; } else { /* * If we made no progress, there must be a good reason. * EOF is handled explicitly above, before the loop. */ ASSERT3S(error, !=, 0); } zfs_exit_two(inzfsvfs, outzfsvfs, FTAG); return (error); } /* * Usual pattern would be to call zfs_clone_range() from zfs_replay_clone(), * but we cannot do that, because when replaying we don't have source znode * available. This is why we need a dedicated replay function. */ int zfs_clone_range_replay(znode_t *zp, uint64_t off, uint64_t len, uint64_t blksz, const blkptr_t *bps, size_t nbps) { zfsvfs_t *zfsvfs; dmu_buf_impl_t *db; dmu_tx_t *tx; int error; int count = 0; sa_bulk_attr_t bulk[3]; uint64_t mtime[2], ctime[2]; ASSERT3U(off, <, MAXOFFSET_T); ASSERT3U(len, >, 0); ASSERT3U(nbps, >, 0); zfsvfs = ZTOZSB(zp); ASSERT(spa_feature_is_enabled(dmu_objset_spa(zfsvfs->z_os), SPA_FEATURE_BLOCK_CLONING)); if ((error = zfs_enter_verify_zp(zfsvfs, zp, FTAG)) != 0) return (error); ASSERT(zfsvfs->z_replay); ASSERT(!zfs_is_readonly(zfsvfs)); if ((off % blksz) != 0) { zfs_exit(zfsvfs, FTAG); return (SET_ERROR(EINVAL)); } SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zfsvfs), NULL, &zp->z_size, 8); /* * Start a transaction. */ tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); db = (dmu_buf_impl_t *)sa_get_db(zp->z_sa_hdl); DB_DNODE_ENTER(db); dmu_tx_hold_clone_by_dnode(tx, DB_DNODE(db), off, len, blksz); DB_DNODE_EXIT(db); zfs_sa_upgrade_txholds(tx, zp); error = dmu_tx_assign(tx, DMU_TX_WAIT); if (error != 0) { dmu_tx_abort(tx); zfs_exit(zfsvfs, FTAG); return (error); } if (zp->z_blksz < blksz) zfs_grow_blocksize(zp, blksz, tx); dmu_brt_clone(zfsvfs->z_os, zp->z_id, off, len, tx, bps, nbps); zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime); if (zp->z_size < off + len) zp->z_size = off + len; error = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx); /* * zil_replaying() not only check if we are replaying ZIL, but also * updates the ZIL header to record replay progress. */ VERIFY(zil_replaying(zfsvfs->z_log, tx)); dmu_tx_commit(tx); zfs_znode_update_vfs(zp); zfs_exit(zfsvfs, FTAG); return (error); } EXPORT_SYMBOL(zfs_access); EXPORT_SYMBOL(zfs_fsync); EXPORT_SYMBOL(zfs_holey); EXPORT_SYMBOL(zfs_read); EXPORT_SYMBOL(zfs_write); EXPORT_SYMBOL(zfs_getsecattr); EXPORT_SYMBOL(zfs_setsecattr); EXPORT_SYMBOL(zfs_clone_range); EXPORT_SYMBOL(zfs_clone_range_replay); ZFS_MODULE_PARAM(zfs_vnops, zfs_vnops_, read_chunk_size, U64, ZMOD_RW, "Bytes to read per chunk"); ZFS_MODULE_PARAM(zfs, zfs_, bclone_enabled, INT, ZMOD_RW, "Enable block cloning"); ZFS_MODULE_PARAM(zfs, zfs_, bclone_wait_dirty, INT, ZMOD_RW, "Wait for dirty blocks when cloning"); ZFS_MODULE_PARAM(zfs, zfs_, dio_enabled, INT, ZMOD_RW, "Enable Direct I/O"); ZFS_MODULE_PARAM(zfs, zfs_, dio_strict, INT, ZMOD_RW, "Return errors on misaligned Direct I/O"); diff --git a/module/zfs/zio.c b/module/zfs/zio.c index 41e0dc0004df..218aec6093e2 100644 --- a/module/zfs/zio.c +++ b/module/zfs/zio.c @@ -1,5962 +1,5988 @@ // SPDX-License-Identifier: CDDL-1.0 /* * 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 https://opensource.org/licenses/CDDL-1.0. * 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, 2022 by Delphix. All rights reserved. * Copyright (c) 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2019, 2023, 2024, 2025, Klara, Inc. * Copyright (c) 2019, Allan Jude * Copyright (c) 2021, Datto, Inc. * Copyright (c) 2021, 2024 by George Melikov. 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 /* * ========================================================================== * I/O type descriptions * ========================================================================== */ const char *const 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_flush", "z_trim" }; int zio_dva_throttle_enabled = B_TRUE; static int zio_deadman_log_all = B_FALSE; /* * ========================================================================== * I/O kmem caches * ========================================================================== */ static kmem_cache_t *zio_cache; static 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) static uint64_t zio_buf_cache_allocs[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; static uint64_t zio_buf_cache_frees[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; #endif /* Mark IOs as "slow" if they take longer than 30 seconds */ static uint_t 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. */ /* defer frees starting in this pass */ uint_t zfs_sync_pass_deferred_free = 2; /* don't compress starting in this pass */ static uint_t zfs_sync_pass_dont_compress = 8; /* rewrite new bps starting in this pass */ static uint_t zfs_sync_pass_rewrite = 2; /* * 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; static int zio_requeue_io_start_cut_in_line = 1; #ifdef ZFS_DEBUG static const int zio_buf_debug_limit = 16384; #else static const int zio_buf_debug_limit = 0; #endif typedef struct zio_stats { kstat_named_t ziostat_total_allocations; kstat_named_t ziostat_alloc_class_fallbacks; kstat_named_t ziostat_gang_writes; kstat_named_t ziostat_gang_multilevel; } zio_stats_t; static zio_stats_t zio_stats = { { "total_allocations", KSTAT_DATA_UINT64 }, { "alloc_class_fallbacks", KSTAT_DATA_UINT64 }, { "gang_writes", KSTAT_DATA_UINT64 }, { "gang_multilevel", KSTAT_DATA_UINT64 }, }; struct { wmsum_t ziostat_total_allocations; wmsum_t ziostat_alloc_class_fallbacks; wmsum_t ziostat_gang_writes; wmsum_t ziostat_gang_multilevel; } ziostat_sums; #define ZIOSTAT_BUMP(stat) wmsum_add(&ziostat_sums.stat, 1); static kstat_t *zio_ksp; static inline void __zio_execute(zio_t *zio); static void zio_taskq_dispatch(zio_t *, zio_taskq_type_t, boolean_t); static int zio_kstats_update(kstat_t *ksp, int rw) { zio_stats_t *zs = ksp->ks_data; if (rw == KSTAT_WRITE) return (EACCES); zs->ziostat_total_allocations.value.ui64 = wmsum_value(&ziostat_sums.ziostat_total_allocations); zs->ziostat_alloc_class_fallbacks.value.ui64 = wmsum_value(&ziostat_sums.ziostat_alloc_class_fallbacks); zs->ziostat_gang_writes.value.ui64 = wmsum_value(&ziostat_sums.ziostat_gang_writes); zs->ziostat_gang_multilevel.value.ui64 = wmsum_value(&ziostat_sums.ziostat_gang_multilevel); return (0); } void zio_init(void) { size_t c; 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); wmsum_init(&ziostat_sums.ziostat_total_allocations, 0); wmsum_init(&ziostat_sums.ziostat_alloc_class_fallbacks, 0); wmsum_init(&ziostat_sums.ziostat_gang_writes, 0); wmsum_init(&ziostat_sums.ziostat_gang_multilevel, 0); zio_ksp = kstat_create("zfs", 0, "zio_stats", "misc", KSTAT_TYPE_NAMED, sizeof (zio_stats) / sizeof (kstat_named_t), KSTAT_FLAG_VIRTUAL); if (zio_ksp != NULL) { zio_ksp->ks_data = &zio_stats; zio_ksp->ks_update = zio_kstats_update; kstat_install(zio_ksp); } for (c = 0; c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; c++) { size_t size = (c + 1) << SPA_MINBLOCKSHIFT; size_t align, cflags, data_cflags; char name[32]; /* * Create cache for each half-power of 2 size, starting from * SPA_MINBLOCKSIZE. It should give us memory space efficiency * of ~7/8, sufficient for transient allocations mostly using * these caches. */ size_t p2 = size; while (!ISP2(p2)) p2 &= p2 - 1; if (!IS_P2ALIGNED(size, p2 / 2)) continue; #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; #endif if (IS_P2ALIGNED(size, PAGESIZE)) align = PAGESIZE; else align = 1 << (highbit64(size ^ (size - 1)) - 1); cflags = (zio_exclude_metadata || size > zio_buf_debug_limit) ? KMC_NODEBUG : 0; data_cflags = KMC_NODEBUG; if (abd_size_alloc_linear(size)) { cflags |= KMC_RECLAIMABLE; data_cflags |= KMC_RECLAIMABLE; } if (cflags == data_cflags) { /* * Resulting kmem caches would be identical. * Save memory by creating only one. */ (void) snprintf(name, sizeof (name), "zio_buf_comb_%lu", (ulong_t)size); zio_buf_cache[c] = kmem_cache_create(name, size, align, NULL, NULL, NULL, NULL, NULL, cflags); zio_data_buf_cache[c] = zio_buf_cache[c]; continue; } (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, NULL, 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 n = SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; #if defined(ZFS_DEBUG) && !defined(_KERNEL) for (size_t i = 0; i < n; i++) { if (zio_buf_cache_allocs[i] != zio_buf_cache_frees[i]) (void) printf("zio_fini: [%d] %llu != %llu\n", (int)((i + 1) << SPA_MINBLOCKSHIFT), (long long unsigned)zio_buf_cache_allocs[i], (long long unsigned)zio_buf_cache_frees[i]); } #endif /* * The same kmem cache can show up multiple times in both zio_buf_cache * and zio_data_buf_cache. Do a wasteful but trivially correct scan to * sort it out. */ for (size_t i = 0; i < n; i++) { kmem_cache_t *cache = zio_buf_cache[i]; if (cache == NULL) continue; for (size_t j = i; j < n; j++) { if (cache == zio_buf_cache[j]) zio_buf_cache[j] = NULL; if (cache == zio_data_buf_cache[j]) zio_data_buf_cache[j] = NULL; } kmem_cache_destroy(cache); } for (size_t i = 0; i < n; i++) { kmem_cache_t *cache = zio_data_buf_cache[i]; if (cache == NULL) continue; for (size_t j = i; j < n; j++) { if (cache == zio_data_buf_cache[j]) zio_data_buf_cache[j] = NULL; } kmem_cache_destroy(cache); } for (size_t i = 0; i < n; i++) { VERIFY3P(zio_buf_cache[i], ==, NULL); VERIFY3P(zio_data_buf_cache[i], ==, NULL); } if (zio_ksp != NULL) { kstat_delete(zio_ksp); zio_ksp = NULL; } wmsum_fini(&ziostat_sums.ziostat_total_allocations); wmsum_fini(&ziostat_sums.ziostat_alloc_class_fallbacks); wmsum_fini(&ziostat_sums.ziostat_gang_writes); wmsum_fini(&ziostat_sums.ziostat_gang_multilevel); kmem_cache_destroy(zio_link_cache); kmem_cache_destroy(zio_cache); zio_inject_fini(); lz4_fini(); } /* * ========================================================================== * Allocate and free I/O buffers * ========================================================================== */ #if defined(ZFS_DEBUG) && defined(_KERNEL) #define ZFS_ZIO_BUF_CANARY 1 #endif #ifdef ZFS_ZIO_BUF_CANARY static const ulong_t zio_buf_canary = (ulong_t)0xdeadc0dedead210b; /* * Use empty space after the buffer to detect overflows. * * Since zio_init() creates kmem caches only for certain set of buffer sizes, * allocations of different sizes may have some unused space after the data. * Filling part of that space with a known pattern on allocation and checking * it on free should allow us to detect some buffer overflows. */ static void zio_buf_put_canary(ulong_t *p, size_t size, kmem_cache_t **cache, size_t c) { size_t off = P2ROUNDUP(size, sizeof (ulong_t)); ulong_t *canary = p + off / sizeof (ulong_t); size_t asize = (c + 1) << SPA_MINBLOCKSHIFT; if (c + 1 < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT && cache[c] == cache[c + 1]) asize = (c + 2) << SPA_MINBLOCKSHIFT; for (; off < asize; canary++, off += sizeof (ulong_t)) *canary = zio_buf_canary; } static void zio_buf_check_canary(ulong_t *p, size_t size, kmem_cache_t **cache, size_t c) { size_t off = P2ROUNDUP(size, sizeof (ulong_t)); ulong_t *canary = p + off / sizeof (ulong_t); size_t asize = (c + 1) << SPA_MINBLOCKSHIFT; if (c + 1 < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT && cache[c] == cache[c + 1]) asize = (c + 2) << SPA_MINBLOCKSHIFT; for (; off < asize; canary++, off += sizeof (ulong_t)) { if (unlikely(*canary != zio_buf_canary)) { PANIC("ZIO buffer overflow %p (%zu) + %zu %#lx != %#lx", p, size, (canary - p) * sizeof (ulong_t), *canary, zio_buf_canary); } } } #endif /* * 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 void *p = kmem_cache_alloc(zio_buf_cache[c], KM_PUSHPAGE); #ifdef ZFS_ZIO_BUF_CANARY zio_buf_put_canary(p, size, zio_buf_cache, c); #endif return (p); } /* * 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); void *p = kmem_cache_alloc(zio_data_buf_cache[c], KM_PUSHPAGE); #ifdef ZFS_ZIO_BUF_CANARY zio_buf_put_canary(p, size, zio_data_buf_cache, c); #endif return (p); } 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 #ifdef ZFS_ZIO_BUF_CANARY zio_buf_check_canary(buf, size, zio_buf_cache, c); #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); #ifdef ZFS_ZIO_BUF_CANARY zio_buf_check_canary(buf, size, zio_data_buf_cache, c); #endif kmem_cache_free(zio_data_buf_cache[c], buf); } static void zio_abd_free(void *abd, size_t size) { (void) 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) { int ret = zio_decompress_data(BP_GET_COMPRESS(zio->io_bp), zio->io_abd, data, zio->io_size, size, &zio->io_prop.zp_complevel); 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. */ abd_t *abd = abd_alloc_linear(lsize, B_TRUE); ret = zio_decompress_data(BP_GET_COMPRESS(bp), zio->io_abd, abd, zio->io_size, lsize, &zio->io_prop.zp_complevel); if (ret != 0) { abd_free(abd); ret = SET_ERROR(EIO); goto error; } ret = zio_crypt_do_indirect_mac_checksum_abd(B_FALSE, abd, lsize, BP_SHOULD_BYTESWAP(bp), mac); abd_free(abd); } 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 when this was done, * we had run out of bits in what is now zio_flag_t. Future cleanup * could make this a flag bit. */ 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, BP_GET_PHYSICAL_BIRTH(zio->io_bp)); (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); } static void zio_add_child_impl(zio_t *pio, zio_t *cio, boolean_t first) { /* * 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); /* Parent should not have READY stage if child doesn't have it. */ IMPLY((cio->io_pipeline & ZIO_STAGE_READY) == 0 && (cio->io_child_type != ZIO_CHILD_VDEV), (pio->io_pipeline & ZIO_STAGE_READY) == 0); zio_link_t *zl = kmem_cache_alloc(zio_link_cache, KM_SLEEP); zl->zl_parent = pio; zl->zl_child = cio; mutex_enter(&pio->io_lock); if (first) ASSERT(list_is_empty(&cio->io_parent_list)); else mutex_enter(&cio->io_lock); ASSERT(pio->io_state[ZIO_WAIT_DONE] == 0); uint64_t *countp = pio->io_children[cio->io_child_type]; for (int w = 0; w < ZIO_WAIT_TYPES; w++) countp[w] += !cio->io_state[w]; list_insert_head(&pio->io_child_list, zl); list_insert_head(&cio->io_parent_list, zl); if (!first) mutex_exit(&cio->io_lock); mutex_exit(&pio->io_lock); } void zio_add_child(zio_t *pio, zio_t *cio) { zio_add_child_impl(pio, cio, B_FALSE); } static void zio_add_child_first(zio_t *pio, zio_t *cio) { zio_add_child_impl(pio, cio, B_TRUE); } 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); 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_post |= zio->io_post; 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. We do this if the parent's zio type matches the child's * type, or if it's a zio_null() with no done callback, and so * has no actual work to do. Otherwise dispatch the parent zio * in its own taskq. * * 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 && (pio->io_type == zio->io_type || (pio->io_type == ZIO_TYPE_NULL && !pio->io_done))) { *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, zio_flag_t 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); memset(zio, 0, 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) { if (type != ZIO_TYPE_WRITE || zio->io_child_type == ZIO_CHILD_DDT) { zio->io_bp_copy = *bp; zio->io_bp = &zio->io_bp_copy; /* so caller can free */ } else { zio->io_bp = (blkptr_t *)bp; } zio->io_bp_orig = *bp; 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; if (flags & ZIO_FLAG_PREALLOCATED) { BP_ZERO_DVAS(zio->io_bp); BP_SET_BIRTH(zio->io_bp, 0, 0); } } 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_allocator = ZIO_ALLOCATOR_NONE; zio->io_state[ZIO_WAIT_READY] = (stage >= ZIO_STAGE_READY) || (pipeline & ZIO_STAGE_READY) == 0; zio->io_state[ZIO_WAIT_DONE] = (stage >= ZIO_STAGE_DONE); if (zb != NULL) zio->io_bookmark = *zb; if (pio != 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_first(pio, zio); } taskq_init_ent(&zio->io_tqent); return (zio); } 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 intended to be between others. Provides synchronization at READY * and DONE pipeline stages and calls the respective callbacks. */ zio_t * zio_null(zio_t *pio, spa_t *spa, vdev_t *vd, zio_done_func_t *done, void *private, zio_flag_t 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 intended to be a root of a tree. Unlike null ZIO does not have a * READY pipeline stage (is ready on creation), so it should not be used * as child of any ZIO that may need waiting for grandchildren READY stage * (any other ZIO type). */ zio_t * zio_root(spa_t *spa, zio_done_func_t *done, void *private, zio_flag_t flags) { zio_t *zio; zio = zio_create(NULL, spa, 0, NULL, NULL, 0, 0, done, private, ZIO_TYPE_NULL, ZIO_PRIORITY_NOW, flags, NULL, 0, NULL, ZIO_STAGE_OPEN, ZIO_ROOT_PIPELINE); return (zio); } 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); zfs_dbgmsg("bad blkptr at %px: " "DVA[0]=%#llx/%#llx " "DVA[1]=%#llx/%#llx " "DVA[2]=%#llx/%#llx " "prop=%#llx " "prop2=%#llx " "pad=%#llx " "phys_birth=%#llx " "birth=%#llx " "fill=%#llx " "cksum=%#llx/%#llx/%#llx/%#llx", bp, (long long)bp->blk_dva[0].dva_word[0], (long long)bp->blk_dva[0].dva_word[1], (long long)bp->blk_dva[1].dva_word[0], (long long)bp->blk_dva[1].dva_word[1], (long long)bp->blk_dva[2].dva_word[0], (long long)bp->blk_dva[2].dva_word[1], (long long)bp->blk_prop, (long long)bp->blk_prop2, (long long)bp->blk_pad, (long long)BP_GET_RAW_PHYSICAL_BIRTH(bp), (long long)BP_GET_LOGICAL_BIRTH(bp), (long long)bp->blk_fill, (long long)bp->blk_cksum.zc_word[0], (long long)bp->blk_cksum.zc_word[1], (long long)bp->blk_cksum.zc_word[2], (long long)bp->blk_cksum.zc_word[3]); switch (blk_verify) { case BLK_VERIFY_HALT: 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 0 is returned. The zfs_blkptr_verify * argument controls the behavior when an invalid field is detected. * * Values for blk_verify_flag: * BLK_VERIFY_ONLY: evaluate the block * BLK_VERIFY_LOG: evaluate the block and log problems * BLK_VERIFY_HALT: call zfs_panic_recover on error * * Values for blk_config_flag: * BLK_CONFIG_HELD: caller holds SCL_VDEV for writer * BLK_CONFIG_NEEDED: caller holds no config lock, SCL_VDEV will be * obtained for reader * BLK_CONFIG_SKIP: skip checks which require SCL_VDEV, for better * performance */ int zfs_blkptr_verify(spa_t *spa, const blkptr_t *bp, enum blk_config_flag blk_config, enum blk_verify_flag blk_verify) { int errors = 0; if (unlikely(!DMU_OT_IS_VALID(BP_GET_TYPE(bp)))) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %px has invalid TYPE %llu", bp, (longlong_t)BP_GET_TYPE(bp)); } if (unlikely(BP_GET_COMPRESS(bp) >= ZIO_COMPRESS_FUNCTIONS)) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %px has invalid COMPRESS %llu", bp, (longlong_t)BP_GET_COMPRESS(bp)); } if (unlikely(BP_GET_LSIZE(bp) > SPA_MAXBLOCKSIZE)) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %px has invalid LSIZE %llu", bp, (longlong_t)BP_GET_LSIZE(bp)); } if (BP_IS_EMBEDDED(bp)) { if (unlikely(BPE_GET_ETYPE(bp) >= NUM_BP_EMBEDDED_TYPES)) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %px has invalid ETYPE %llu", bp, (longlong_t)BPE_GET_ETYPE(bp)); } if (unlikely(BPE_GET_PSIZE(bp) > BPE_PAYLOAD_SIZE)) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %px has invalid PSIZE %llu", bp, (longlong_t)BPE_GET_PSIZE(bp)); } return (errors ? ECKSUM : 0); } else if (BP_IS_HOLE(bp)) { /* * Holes are allowed (expected, even) to have no DVAs, no * checksum, and no psize. */ return (errors ? ECKSUM : 0); } else if (unlikely(!DVA_IS_VALID(&bp->blk_dva[0]))) { /* Non-hole, non-embedded BPs _must_ have at least one DVA */ errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %px has no valid DVAs", bp); } if (unlikely(BP_GET_CHECKSUM(bp) >= ZIO_CHECKSUM_FUNCTIONS)) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %px has invalid CHECKSUM %llu", bp, (longlong_t)BP_GET_CHECKSUM(bp)); } if (unlikely(BP_GET_PSIZE(bp) > SPA_MAXBLOCKSIZE)) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %px has invalid PSIZE %llu", bp, (longlong_t)BP_GET_PSIZE(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 (unlikely(!spa->spa_trust_config)) return (errors ? ECKSUM : 0); switch (blk_config) { case BLK_CONFIG_HELD: ASSERT(spa_config_held(spa, SCL_VDEV, RW_WRITER)); break; case BLK_CONFIG_NEEDED: spa_config_enter(spa, SCL_VDEV, bp, RW_READER); break; case BLK_CONFIG_NEEDED_TRY: if (!spa_config_tryenter(spa, SCL_VDEV, bp, RW_READER)) return (EBUSY); break; case BLK_CONFIG_SKIP: return (errors ? ECKSUM : 0); default: panic("invalid blk_config %u", blk_config); } /* * Pool-specific checks. * * Note: it would be nice to verify that the logical birth * and 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++) { const dva_t *dva = &bp->blk_dva[i]; uint64_t vdevid = DVA_GET_VDEV(dva); if (unlikely(vdevid >= spa->spa_root_vdev->vdev_children)) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %px DVA %u has invalid VDEV %llu", bp, i, (longlong_t)vdevid); continue; } vdev_t *vd = spa->spa_root_vdev->vdev_child[vdevid]; if (unlikely(vd == NULL)) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %px DVA %u has invalid VDEV %llu", bp, i, (longlong_t)vdevid); continue; } if (unlikely(vd->vdev_ops == &vdev_hole_ops)) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %px 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(dva); uint64_t asize = DVA_GET_ASIZE(dva); if (DVA_GET_GANG(dva)) asize = vdev_gang_header_asize(vd); if (unlikely(offset + asize > vd->vdev_asize)) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %px DVA %u has invalid OFFSET %llu", bp, i, (longlong_t)offset); } } if (blk_config == BLK_CONFIG_NEEDED || blk_config == BLK_CONFIG_NEEDED_TRY) spa_config_exit(spa, SCL_VDEV, bp); return (errors ? ECKSUM : 0); } boolean_t zfs_dva_valid(spa_t *spa, const dva_t *dva, const blkptr_t *bp) { (void) 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 (DVA_GET_GANG(dva)) asize = vdev_gang_header_asize(vd); 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, zio_flag_t flags, const zbookmark_phys_t *zb) { zio_t *zio; zio = zio_create(pio, spa, BP_GET_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 *done, void *private, zio_priority_t priority, zio_flag_t flags, const zbookmark_phys_t *zb) { zio_t *zio; enum zio_stage pipeline = zp->zp_direct_write == B_TRUE ? ZIO_DIRECT_WRITE_PIPELINE : (flags & ZIO_FLAG_DDT_CHILD) ? ZIO_DDT_CHILD_WRITE_PIPELINE : ZIO_WRITE_PIPELINE; zio = zio_create(pio, spa, txg, bp, data, lsize, psize, done, private, ZIO_TYPE_WRITE, priority, flags, NULL, 0, zb, ZIO_STAGE_OPEN, pipeline); zio->io_ready = ready; zio->io_children_ready = children_ready; 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, zio_flag_t 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, int gang_copies, boolean_t nopwrite, boolean_t brtwrite) { 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)); ASSERT(!brtwrite || !nopwrite); /* * 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_brtwrite = brtwrite; zio->io_prop.zp_copies = copies; zio->io_prop.zp_gang_copies = gang_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, BLK_CONFIG_NEEDED, 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; /* * 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)) || brt_maybe_exists(spa, bp)) { metaslab_check_free(spa, bp); 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, zio_flag_t 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) || brt_maybe_exists(spa, bp)) { /* * GANG, DEDUP and BRT blocks can induce a read (for the gang * block header, the DDT or the BRT), 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, zio_flag_t flags) { zio_t *zio; (void) zfs_blkptr_verify(spa, bp, (flags & ZIO_FLAG_CONFIG_WRITER) ? BLK_CONFIG_HELD : BLK_CONFIG_NEEDED, 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(BP_GET_LOGICAL_BIRTH(&spa->spa_uberblock.ub_rootbp), <, spa_min_claim_txg(spa)); ASSERT(txg == spa_min_claim_txg(spa) || txg == 0); 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_trim(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, zio_done_func_t *done, void *private, zio_priority_t priority, zio_flag_t 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, zio_flag_t 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, zio_flag_t 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, zio_flag_t 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; /* * We never allow the mirror VDEV to attempt reading from any * additional data copies after the first Direct I/O checksum * verify failure. This is to avoid bad data being written out * through the mirror during self healing. See comment in * vdev_mirror_io_done() for more details. */ ASSERT0(pio->io_post & ZIO_POST_DIO_CHKSUM_ERR); } else if (type == ZIO_TYPE_WRITE && pio->io_prop.zp_direct_write == B_TRUE) { /* * By default we only will verify checksums for Direct I/O * writes for Linux. FreeBSD is able to place user pages under * write protection before issuing them to the ZIO pipeline. * * Checksum validation errors will only be reported through * the top-level VDEV, which is set by this child ZIO. */ ASSERT3P(bp, !=, NULL); ASSERT3U(pio->io_child_type, ==, ZIO_CHILD_LOGICAL); pipeline |= ZIO_STAGE_DIO_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_ALLOC_THROTTLED && (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_ALLOC_THROTTLED; } 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); 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, zio_flag_t 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); } /* * Send a flush command to the given vdev. Unlike most zio creation functions, * the flush zios are issued immediately. You can wait on pio to pause until * the flushes complete. */ void zio_flush(zio_t *pio, vdev_t *vd) { const zio_flag_t flags = ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_DONT_RETRY; if (vd->vdev_nowritecache) return; if (vd->vdev_children == 0) { zio_nowait(zio_create(pio, vd->vdev_spa, 0, NULL, NULL, 0, 0, NULL, NULL, ZIO_TYPE_FLUSH, ZIO_PRIORITY_NOW, flags, vd, 0, NULL, ZIO_STAGE_OPEN, ZIO_FLUSH_PIPELINE)); } else { for (uint64_t c = 0; c < vd->vdev_children; c++) zio_flush(pio, vd->vdev_child[c]); } } 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; } } /* * Round provided allocation size up to a value that can be allocated * by at least some vdev(s) in the pool with minimum or no additional * padding and without extra space usage on others */ static uint64_t zio_roundup_alloc_size(spa_t *spa, uint64_t size) { if (size > spa->spa_min_alloc) return (roundup(size, spa->spa_gcd_alloc)); return (spa->spa_min_alloc); } size_t zio_get_compression_max_size(enum zio_compress compress, uint64_t gcd_alloc, uint64_t min_alloc, size_t s_len) { size_t d_len; /* minimum 12.5% must be saved (legacy value, may be changed later) */ d_len = s_len - (s_len >> 3); /* ZLE can't use exactly d_len bytes, it needs more, so ignore it */ if (compress == ZIO_COMPRESS_ZLE) return (d_len); d_len = d_len - d_len % gcd_alloc; if (d_len < min_alloc) return (BPE_PAYLOAD_SIZE); return (d_len); } /* * ========================================================================== * 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)); } 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_GET_BIRTH(bp) != zio->io_txg); *bp = *zio->io_bp_override; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (zp->zp_brtwrite) return (zio); ASSERT(!BP_GET_DEDUP(zio->io_bp_override)); 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; uint32_t 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_GET_BIRTH(bp) == 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) || BP_IS_GANG(bp) || MIN(zp->zp_copies, 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)) { abd_t *cabd = NULL; if (abd_cmp_zero(zio->io_abd, lsize) == 0) psize = 0; else if (compress == ZIO_COMPRESS_EMPTY) psize = lsize; else psize = zio_compress_data(compress, zio->io_abd, &cabd, lsize, zio_get_compression_max_size(compress, spa->spa_gcd_alloc, spa->spa_min_alloc, lsize), zp->zp_complevel); if (psize == 0) { compress = ZIO_COMPRESS_OFF; } else if (psize >= lsize) { compress = ZIO_COMPRESS_OFF; if (cabd != NULL) abd_free(cabd); } else if (psize <= BPE_PAYLOAD_SIZE && !zp->zp_encrypt && zp->zp_level == 0 && !DMU_OT_HAS_FILL(zp->zp_type) && spa_feature_is_enabled(spa, SPA_FEATURE_EMBEDDED_DATA)) { void *cbuf = abd_borrow_buf_copy(cabd, lsize); 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); abd_return_buf(cabd, cbuf, lsize); abd_free(cabd); BP_SET_LOGICAL_BIRTH(bp, zio->io_txg); zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; ASSERT(spa_feature_is_active(spa, SPA_FEATURE_EMBEDDED_DATA)); return (zio); } else { /* * Round compressed size up to the minimum allocation * size 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. */ size_t rounded = (size_t)zio_roundup_alloc_size(spa, psize); if (rounded >= lsize) { compress = ZIO_COMPRESS_OFF; abd_free(cabd); psize = lsize; } else { abd_zero_off(cabd, psize, rounded - psize); psize = rounded; zio_push_transform(zio, cabd, 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. */ if (abd_cmp_zero(zio->io_abd, lsize) == 0) { psize = 0; compress = ZIO_COMPRESS_OFF; } else { psize = lsize; } } else if (zio->io_flags & ZIO_FLAG_RAW_COMPRESS && !(zio->io_flags & ZIO_FLAG_RAW_ENCRYPT)) { /* * If we are raw receiving an encrypted dataset we should not * take this codepath because it will change the on-disk block * and decryption will fail. */ size_t rounded = MIN((size_t)zio_roundup_alloc_size(spa, psize), lsize); if (rounded != psize) { abd_t *cdata = abd_alloc_linear(rounded, B_TRUE); abd_zero_off(cdata, psize, rounded - psize); abd_copy_off(cdata, zio->io_abd, 0, 0, psize); psize = rounded; zio_push_transform(zio, cdata, psize, rounded, NULL); } } 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_GET_BIRTH(bp) == 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 (BP_GET_LOGICAL_BIRTH(&zio->io_bp_orig) != 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; /* * 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 or cut the line otherwise. */ if (zio->io_priority == ZIO_PRIORITY_SYNC_WRITE) { if (spa->spa_zio_taskq[t][q + 1].stqs_count != 0) q++; else cutinline = B_TRUE; } ASSERT3U(q, <, ZIO_TASKQ_TYPES); spa_taskq_dispatch(spa, t, q, zio_execute, zio, cutinline); } 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) { ASSERT((zio->io_type != ZIO_TYPE_WRITE) || ZIO_HAS_ALLOCATOR(zio)); zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_FALSE); return (NULL); } void zio_interrupt(void *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; int ticks = MAX(1, NSEC_TO_TICK(diff)); clock_t expire_at_tick = ddi_get_lbolt() + ticks; DTRACE_PROBE3(zio__delay__hit, zio_t *, zio, hrtime_t, now, hrtime_t, diff); tid = taskq_dispatch_delay(system_taskq, 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; uint64_t failmode = spa_get_deadman_failmode(pio->io_spa); 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; 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%llx 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, (u_longlong_t)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, (u_longlong_t)pio->io_flags, pio->io_stage, pio->io_pipeline, pio->io_pipeline_trace, (u_longlong_t)zb->zb_objset, (u_longlong_t)zb->zb_object, (u_longlong_t)zb->zb_level, (u_longlong_t)zb->zb_blkid, (u_longlong_t)pio->io_offset, (u_longlong_t)pio->io_size, pio->io_error); (void) zfs_ereport_post(FM_EREPORT_ZFS_DEADMAN, pio->io_spa, vd, zb, pio, 0); } if (vd != NULL && vd->vdev_ops->vdev_op_leaf && list_is_empty(&pio->io_child_list) && failmode == ZIO_FAILURE_MODE_CONTINUE && taskq_empty_ent(&pio->io_tqent) && pio->io_queue_state == ZIO_QS_ACTIVE) { pio->io_error = EINTR; 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, const 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(void *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); #else (void) zio; #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(); if (zio->io_type == ZIO_TYPE_WRITE) { spa_select_allocator(zio); } __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 && list_is_empty(&zio->io_parent_list)) { 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; pio = spa->spa_async_zio_root[CPU_SEQID_UNSTABLE]; zio_add_child(pio, zio); } ASSERT0(zio->io_queued_timestamp); zio->io_queued_timestamp = gethrtime(); if (zio->io_type == ZIO_TYPE_WRITE) { spa_select_allocator(zio); } __zio_execute(zio); } /* * ========================================================================== * Reexecute, cancel, or suspend/resume failed I/O * ========================================================================== */ static void zio_reexecute(void *arg) { zio_t *pio = arg; zio_t *cio, *cio_next, *gio; 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); mutex_enter(&pio->io_lock); pio->io_flags = pio->io_orig_flags; pio->io_stage = pio->io_orig_stage; pio->io_pipeline = pio->io_orig_pipeline; pio->io_post = 0; pio->io_flags |= ZIO_FLAG_REEXECUTED; pio->io_pipeline_trace = 0; pio->io_error = 0; pio->io_state[ZIO_WAIT_READY] = (pio->io_stage >= ZIO_STAGE_READY) || (pio->io_pipeline & ZIO_STAGE_READY) == 0; pio->io_state[ZIO_WAIT_DONE] = (pio->io_stage >= ZIO_STAGE_DONE); /* * It's possible for a failed ZIO to be a descendant of more than one * ZIO tree. When reexecuting it, we have to be sure to add its wait * states to all parent wait counts. * * Those parents, in turn, may have other children that are currently * active, usually because they've already been reexecuted after * resuming. Those children may be executing and may call * zio_notify_parent() at the same time as we're updating our parent's * counts. To avoid races while updating the counts, we take * gio->io_lock before each update. */ zio_link_t *zl = NULL; while ((gio = zio_walk_parents(pio, &zl)) != NULL) { mutex_enter(&gio->io_lock); for (int w = 0; w < ZIO_WAIT_TYPES; w++) { gio->io_children[pio->io_child_type][w] += !pio->io_state[w]; } mutex_exit(&gio->io_lock); } 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'. */ zl = NULL; for (cio = zio_walk_children(pio, &zl); cio != NULL; cio = cio_next) { cio_next = zio_walk_children(pio, &zl); 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)); if (reason != ZIO_SUSPEND_MMP) { cmn_err(CE_WARN, "Pool '%s' has encountered an uncorrectable " "I/O failure and has been suspended.", 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); txg_wait_kick(spa->spa_dsl_pool); } int zio_resume(spa_t *spa) { zio_t *pio; /* * Reexecute all previously suspended i/o. */ mutex_enter(&spa->spa_suspend_lock); if (spa->spa_suspended != ZIO_SUSPEND_NONE) cmn_err(CE_WARN, "Pool '%s' was suspended and is being " "resumed. Failed I/O will be retried.", spa_name(spa)); 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 a gang header and up to gbh_nblkptrs(size) * gang members. The gang header is like an indirect block: it's an array * of block pointers, though the header has a small tail (a zio_eck_t) * that stores an embedded checksum. It is allocated using only a single * sector as the requested size, and hence is allocatable regardless of * fragmentation. Its size is determined by the smallest allocatable * asize of the vdevs it was allocated on. 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_free(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, gn->gn_gangblocksize); zio = zio_rewrite(pio, pio->io_spa, pio->io_txg, bp, gbh_abd, gn->gn_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_free(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); } static zio_t * zio_free_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, abd_t *data, uint64_t offset) { (void) gn, (void) data, (void) 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); } static zio_t * zio_claim_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, abd_t *data, uint64_t offset) { (void) gn, (void) data, (void) 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, uint64_t gangblocksize) { zio_gang_node_t *gn; ASSERT(*gnpp == NULL); gn = kmem_zalloc(sizeof (*gn) + (gbh_nblkptrs(gangblocksize) * sizeof (gn)), KM_SLEEP); gn->gn_gangblocksize = gn->gn_allocsize = gangblocksize; gn->gn_gbh = zio_buf_alloc(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 < gbh_nblkptrs(gn->gn_allocsize); g++) ASSERT(gn->gn_child[g] == NULL); zio_buf_free(gn->gn_gbh, gn->gn_allocsize); kmem_free(gn, sizeof (*gn) + (gbh_nblkptrs(gn->gn_allocsize) * 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 < gbh_nblkptrs(gn->gn_allocsize); 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) { uint64_t gangblocksize = UINT64_MAX; if (spa_feature_is_active(gio->io_spa, SPA_FEATURE_DYNAMIC_GANG_HEADER)) { spa_config_enter(gio->io_spa, SCL_VDEV, FTAG, RW_READER); for (int dva = 0; dva < BP_GET_NDVAS(bp); dva++) { vdev_t *vd = vdev_lookup_top(gio->io_spa, DVA_GET_VDEV(&bp->blk_dva[dva])); uint64_t psize = vdev_gang_header_psize(vd); gangblocksize = MIN(gangblocksize, psize); } spa_config_exit(gio->io_spa, SCL_VDEV, FTAG); } else { gangblocksize = SPA_OLD_GANGBLOCKSIZE; } ASSERT3U(gangblocksize, !=, UINT64_MAX); zio_gang_node_t *gn = zio_gang_node_alloc(gnpp, gangblocksize); abd_t *gbh_abd = abd_get_from_buf(gn->gn_gbh, gangblocksize); ASSERT(gio->io_gang_leader == gio); ASSERT(BP_IS_GANG(bp)); zio_nowait(zio_read(gio, gio->io_spa, bp, gbh_abd, 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(list_is_empty(&zio->io_child_list)); 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); /* * If this was an old-style gangblock, the gangblocksize should have * been updated in zio_checksum_error to reflect that. */ ASSERT3U(gbh_eck(gn->gn_gbh, gn->gn_gangblocksize)->zec_magic, ==, ZEC_MAGIC); abd_free(zio->io_abd); for (int g = 0; g < gbh_nblkptrs(gn->gn_gangblocksize); g++) { blkptr_t *gbp = gbh_bp(gn->gn_gbh, 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) { ASSERT3U(gbh_eck(gn->gn_gbh, gn->gn_gangblocksize)->zec_magic, ==, ZEC_MAGIC); for (int g = 0; g < gbh_nblkptrs(gn->gn_gangblocksize); g++) { blkptr_t *gbp = gbh_bp(gn->gn_gbh, 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_gang_inherit_allocator(zio_t *pio, zio_t *cio) { cio->io_allocator = pio->io_allocator; } 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; /* * If we're getting direct-invoked from zio_write_gang_block(), * the bp_orig will be set. */ ASSERT(BP_IS_HOLE(&zio->io_bp_orig) || zio->io_flags & ZIO_FLAG_PREALLOCATED); 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)); VERIFY3U(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_free(zio->io_abd); } static void zio_update_feature(void *arg, dmu_tx_t *tx) { spa_t *spa = dmu_tx_pool(tx)->dp_spa; spa_feature_incr(spa, (spa_feature_t)(uintptr_t)arg, tx); } static zio_t * zio_write_gang_block(zio_t *pio, metaslab_class_t *mc) { spa_t *spa = pio->io_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; zio_prop_t zp; int error; boolean_t has_data = !(pio->io_flags & ZIO_FLAG_NODATA); /* * Store multiple copies of the GBH, so that we can still traverse * all the data (e.g. to free or scrub) even if a block is damaged. * This value respects the redundant_metadata property. */ int gbh_copies = gio->io_prop.zp_gang_copies; if (gbh_copies == 0) { /* * This should only happen in the case where we're filling in * DDT entries for a parent that wants more copies than the DDT * has. In that case, we cannot gang without creating a mixed * blkptr, which is illegal. */ ASSERT3U(gio->io_child_type, ==, ZIO_CHILD_DDT); pio->io_error = EAGAIN; return (pio); } ASSERT3S(gbh_copies, >, 0); ASSERT3S(gbh_copies, <=, SPA_DVAS_PER_BP); ASSERT(ZIO_HAS_ALLOCATOR(pio)); int flags = METASLAB_GANG_HEADER; if (pio->io_flags & ZIO_FLAG_ALLOC_THROTTLED) { ASSERT(pio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(has_data); flags |= METASLAB_ASYNC_ALLOC; } uint64_t gangblocksize = SPA_OLD_GANGBLOCKSIZE; uint64_t candidate = gangblocksize; error = metaslab_alloc_range(spa, mc, gangblocksize, gangblocksize, bp, gbh_copies, txg, pio == gio ? NULL : gio->io_bp, flags, &pio->io_alloc_list, pio->io_allocator, pio, &candidate); if (error) { pio->io_error = error; return (pio); } if (spa_feature_is_active(spa, SPA_FEATURE_DYNAMIC_GANG_HEADER)) gangblocksize = candidate; 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, gangblocksize); gbh = gn->gn_gbh; memset(gbh, 0, gangblocksize); gbh_abd = abd_get_from_buf(gbh, gangblocksize); /* * Create the gang header. */ zio = zio_rewrite(pio, spa, txg, bp, gbh_abd, gangblocksize, zio_write_gang_done, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); zio_gang_inherit_allocator(pio, zio); if (pio->io_flags & ZIO_FLAG_ALLOC_THROTTLED) { boolean_t more; VERIFY(metaslab_class_throttle_reserve(mc, zio->io_allocator, gbh_copies, zio->io_size, B_TRUE, &more)); zio->io_flags |= ZIO_FLAG_ALLOC_THROTTLED; } /* * Create and nowait the gang children. First, we try to do * opportunistic allocations. If that fails to generate enough * space, we fall back to normal zio_write calls for nested gang. */ int g; boolean_t any_failed = B_FALSE; for (g = 0; resid != 0; g++) { flags &= METASLAB_ASYNC_ALLOC; flags |= METASLAB_GANG_CHILD; 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 = zp.zp_storage_type = DMU_OT_NONE; zp.zp_level = 0; zp.zp_copies = gio->io_prop.zp_copies; zp.zp_gang_copies = gio->io_prop.zp_gang_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; zp.zp_direct_write = B_FALSE; memset(zp.zp_salt, 0, ZIO_DATA_SALT_LEN); memset(zp.zp_iv, 0, ZIO_DATA_IV_LEN); memset(zp.zp_mac, 0, ZIO_DATA_MAC_LEN); uint64_t min_size = zio_roundup_alloc_size(spa, resid / (gbh_nblkptrs(gangblocksize) - g)); min_size = MIN(min_size, resid); bp = &((blkptr_t *)gbh)[g]; zio_alloc_list_t cio_list; metaslab_trace_init(&cio_list); uint64_t allocated_size = UINT64_MAX; error = metaslab_alloc_range(spa, mc, min_size, resid, bp, gio->io_prop.zp_copies, txg, NULL, flags, &cio_list, zio->io_allocator, NULL, &allocated_size); boolean_t allocated = error == 0; any_failed |= !allocated; uint64_t psize = allocated ? MIN(resid, allocated_size) : min_size; ASSERT3U(psize, >=, min_size); zio_t *cio = zio_write(zio, spa, txg, bp, has_data ? abd_get_offset(pio->io_abd, pio->io_size - resid) : NULL, psize, psize, &zp, zio_write_gang_member_ready, NULL, zio_write_gang_done, &gn->gn_child[g], pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio) | (allocated ? ZIO_FLAG_PREALLOCATED : 0), &pio->io_bookmark); resid -= psize; zio_gang_inherit_allocator(zio, cio); if (allocated) { metaslab_trace_move(&cio_list, &cio->io_alloc_list); metaslab_group_alloc_increment_all(spa, &cio->io_bp_orig, zio->io_allocator, flags, psize, cio); } /* * We do not reserve for the child writes, since we already * reserved for the parent. Unreserve though will be called * for individual children. We can do this since sum of all * child's physical sizes is equal to parent's physical size. * It would not work for potentially bigger allocation sizes. */ zio_nowait(cio); } /* * If we used more gang children than the old limit, we must already be * using the new headers. No need to update anything, just move on. * * Otherwise, we might be in a case where we need to turn on the new * feature, so we check that. We enable the new feature if we didn't * manage to fit everything into 3 gang children and we could have * written more than that. */ if (g > gbh_nblkptrs(SPA_OLD_GANGBLOCKSIZE)) { ASSERT(spa_feature_is_active(spa, SPA_FEATURE_DYNAMIC_GANG_HEADER)); } else if (any_failed && candidate > SPA_OLD_GANGBLOCKSIZE && spa_feature_is_enabled(spa, SPA_FEATURE_DYNAMIC_GANG_HEADER) && !spa_feature_is_active(spa, SPA_FEATURE_DYNAMIC_GANG_HEADER)) { dmu_tx_t *tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg + 1); dsl_sync_task_nowait(spa->spa_dsl_pool, zio_update_feature, (void *)SPA_FEATURE_DYNAMIC_GANG_HEADER, tx); dmu_tx_commit(tx); } /* * Set pio's pipeline to just wait for zio to finish. */ pio->io_pipeline = ZIO_INTERLOCK_PIPELINE; 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_IS_HOLE(bp)); 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); ASSERT3U(bp->blk_prop, ==, bp_orig->blk_prop); /* * 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); for (int d = 0; d < BP_GET_NDVAS(bp_orig); d++) { vdev_t *tvd = vdev_lookup_top(zio->io_spa, DVA_GET_VDEV(&bp_orig->blk_dva[d])); 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); } /* * ========================================================================== * Block Reference Table * ========================================================================== */ static zio_t * zio_brt_free(zio_t *zio) { blkptr_t *bp; bp = zio->io_bp; if (BP_GET_LEVEL(bp) > 0 || BP_IS_METADATA(bp) || !brt_maybe_exists(zio->io_spa, bp)) { return (zio); } if (!brt_entry_decref(zio->io_spa, bp)) { /* * This isn't the last reference, so we cannot free * the data yet. */ zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; } return (zio); } /* * ========================================================================== * Dedup * ========================================================================== */ static void zio_ddt_child_read_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; ddt_t *ddt; ddt_entry_t *dde = zio->io_private; zio_t *pio = zio_unique_parent(zio); mutex_enter(&pio->io_lock); ddt = ddt_select(zio->io_spa, bp); if (zio->io_error == 0) { ddt_phys_variant_t v = ddt_phys_select(ddt, dde, bp); /* this phys variant doesn't need repair */ ddt_phys_clear(dde->dde_phys, v); } if (zio->io_error == 0 && dde->dde_io->dde_repair_abd == NULL) dde->dde_io->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_variant_t v_self = ddt_phys_select(ddt, dde, bp); ddt_univ_phys_t *ddp = dde->dde_phys; blkptr_t blk; ASSERT(zio->io_vsd == NULL); zio->io_vsd = dde; if (v_self == DDT_PHYS_NONE) return (zio); /* issue I/O for the other copies */ for (int p = 0; p < DDT_NPHYS(ddt); p++) { ddt_phys_variant_t v = DDT_PHYS_VARIANT(ddt, p); if (ddt_phys_birth(ddp, v) == 0 || v == v_self) continue; ddt_bp_create(ddt->ddt_checksum, &dde->dde_key, ddp, v, &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_io->dde_repair_abd != NULL) { abd_copy(zio->io_abd, dde->dde_io->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 = 0; p < DDT_NPHYS(ddt); p++) { if (DDT_PHYS_IS_DITTO(ddt, p)) continue; if (dde->dde_io == NULL) continue; zio_t *lio = dde->dde_io->dde_lead_zio[p]; if (lio == NULL) continue; if (do_raw) return (lio->io_size != zio->io_size || abd_cmp(zio->io_abd, lio->io_abd) != 0); return (lio->io_orig_size != zio->io_orig_size || abd_cmp(zio->io_orig_abd, lio->io_orig_abd) != 0); } for (int p = 0; p < DDT_NPHYS(ddt); p++) { ddt_phys_variant_t v = DDT_PHYS_VARIANT(ddt, p); uint64_t phys_birth = ddt_phys_birth(dde->dde_phys, v); if (phys_birth != 0 && do_raw) { blkptr_t blk = *zio->io_bp; uint64_t psize; abd_t *tmpabd; int error; ddt_bp_fill(dde->dde_phys, v, &blk, 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 (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(dde->dde_phys, v, &blk, 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_done(zio_t *zio) { ddt_t *ddt = ddt_select(zio->io_spa, zio->io_bp); ddt_entry_t *dde = zio->io_private; zio_link_t *zl = NULL; ASSERT3P(zio_walk_parents(zio, &zl), !=, NULL); int p = DDT_PHYS_FOR_COPIES(ddt, zio->io_prop.zp_copies); ddt_phys_variant_t v = DDT_PHYS_VARIANT(ddt, p); ddt_univ_phys_t *ddp = dde->dde_phys; ddt_enter(ddt); /* we're the lead, so once we're done there's no one else outstanding */ if (dde->dde_io->dde_lead_zio[p] == zio) dde->dde_io->dde_lead_zio[p] = NULL; ddt_univ_phys_t *orig = &dde->dde_io->dde_orig_phys; if (zio->io_error != 0) { /* * The write failed, so we're about to abort the entire IO * chain. We need to revert the entry back to what it was at * the last time it was successfully extended. */ ddt_phys_unextend(ddp, orig, v); ddt_phys_clear(orig, v); ddt_exit(ddt); return; } /* * Add references for all dedup writes that were waiting on the * physical one, skipping any other physical writes that are waiting. */ zio_t *pio; zl = NULL; while ((pio = zio_walk_parents(zio, &zl)) != NULL) { if (!(pio->io_flags & ZIO_FLAG_DDT_CHILD)) ddt_phys_addref(ddp, v); } /* * We've successfully added new DVAs to the entry. Clear the saved * state or, if there's still outstanding IO, remember it so we can * revert to a known good state if that IO fails. */ if (dde->dde_io->dde_lead_zio[p] == NULL) ddt_phys_clear(orig, v); else ddt_phys_copy(orig, ddp, v); ddt_exit(ddt); } static void zio_ddt_child_write_ready(zio_t *zio) { ddt_t *ddt = ddt_select(zio->io_spa, zio->io_bp); ddt_entry_t *dde = zio->io_private; zio_link_t *zl = NULL; ASSERT3P(zio_walk_parents(zio, &zl), !=, NULL); int p = DDT_PHYS_FOR_COPIES(ddt, zio->io_prop.zp_copies); ddt_phys_variant_t v = DDT_PHYS_VARIANT(ddt, p); if (ddt_phys_is_gang(dde->dde_phys, v)) { for (int i = 0; i < BP_GET_NDVAS(zio->io_bp); i++) { dva_t *d = &zio->io_bp->blk_dva[i]; metaslab_group_alloc_decrement(zio->io_spa, DVA_GET_VDEV(d), zio->io_allocator, METASLAB_ASYNC_ALLOC, zio->io_size, zio); } zio->io_error = EAGAIN; } if (zio->io_error != 0) return; ddt_enter(ddt); ddt_phys_extend(dde->dde_phys, v, zio->io_bp); zio_t *pio; zl = NULL; while ((pio = zio_walk_parents(zio, &zl)) != NULL) { if (!(pio->io_flags & ZIO_FLAG_DDT_CHILD)) ddt_bp_fill(dde->dde_phys, v, pio->io_bp, zio->io_txg); } 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; ddt_t *ddt = ddt_select(spa, bp); ddt_entry_t *dde; 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))); /* * Deduplication will not take place for Direct I/O writes. The * ddt_tree will be emptied in syncing context. Direct I/O writes take * place in the open-context. Direct I/O write can not attempt to * modify the ddt_tree while issuing out a write. */ ASSERT3B(zio->io_prop.zp_direct_write, ==, B_FALSE); ddt_enter(ddt); /* * Search DDT for matching entry. Skip DVAs verification here, since * they can go only from override, and once we get here the override * pointer can't have "D" flag to be confused with pruned DDT entries. */ IMPLY(zio->io_bp_override, !BP_GET_DEDUP(zio->io_bp_override)); dde = ddt_lookup(ddt, bp, B_FALSE); if (dde == NULL) { /* DDT size is over its quota so no new entries */ zp->zp_dedup = B_FALSE; BP_SET_DEDUP(bp, B_FALSE); if (zio->io_bp_override == NULL) zio->io_pipeline = ZIO_WRITE_PIPELINE; ddt_exit(ddt); return (zio); } 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); } int p = DDT_PHYS_FOR_COPIES(ddt, zp->zp_copies); ddt_phys_variant_t v = DDT_PHYS_VARIANT(ddt, p); ddt_univ_phys_t *ddp = dde->dde_phys; /* * In the common cases, at this point we have a regular BP with no * allocated DVAs, and the corresponding DDT entry for its checksum. * Our goal is to fill the BP with enough DVAs to satisfy its copies= * requirement. * * One of three things needs to happen to fulfill this: * * - if the DDT entry has enough DVAs to satisfy the BP, we just copy * them out of the entry and return; * * - if the DDT entry has no DVAs (ie its brand new), then we have to * issue the write as normal so that DVAs can be allocated and the * data land on disk. We then copy the DVAs into the DDT entry on * return. * * - if the DDT entry has some DVAs, but too few, we have to issue the * write, adjusted to have allocate fewer copies. When it returns, we * add the new DVAs to the DDT entry, and update the BP to have the * full amount it originally requested. * * In all cases, if there's already a writing IO in flight, we need to * defer the action until after the write is done. If our action is to * write, we need to adjust our request for additional DVAs to match * what will be in the DDT entry after it completes. In this way every * IO can be guaranteed to recieve enough DVAs simply by joining the * end of the chain and letting the sequence play out. */ /* * Number of DVAs in the DDT entry. If the BP is encrypted we ignore * the third one as normal. */ int have_dvas = ddt_phys_dva_count(ddp, v, BP_IS_ENCRYPTED(bp)); IMPLY(have_dvas == 0, ddt_phys_birth(ddp, v) == 0); boolean_t is_ganged = ddt_phys_is_gang(ddp, v); /* Number of DVAs requested by the IO. */ uint8_t need_dvas = zp->zp_copies; /* Number of DVAs in outstanding writes for this dde. */ uint8_t parent_dvas = 0; /* * What we do next depends on whether or not there's IO outstanding that * will update this entry. */ if (dde->dde_io == NULL || dde->dde_io->dde_lead_zio[p] == NULL) { /* * No IO outstanding, so we only need to worry about ourselves. */ /* * Override BPs bring their own DVAs and their own problems. */ if (zio->io_bp_override) { /* * For a brand-new entry, all the work has been done * for us, and we can just fill it out from the provided * block and leave. */ if (have_dvas == 0) { ASSERT(BP_GET_BIRTH(bp) == txg); ASSERT(BP_EQUAL(bp, zio->io_bp_override)); ddt_phys_extend(ddp, v, bp); ddt_phys_addref(ddp, v); ddt_exit(ddt); return (zio); } /* * If we already have this entry, then we want to treat * it like a regular write. To do this we just wipe * them out and proceed like a regular write. * * Even if there are some DVAs in the entry, we still * have to clear them out. We can't use them to fill * out the dedup entry, as they are all referenced * together by a bp already on disk, and will be freed * as a group. */ BP_ZERO_DVAS(bp); BP_SET_BIRTH(bp, 0, 0); } /* * If there are enough DVAs in the entry to service our request, * then we can just use them as-is. */ if (have_dvas >= need_dvas) { + /* + * For rewrite operations, try preserving the original + * logical birth time. If the result matches the + * original BP, this becomes a NOP. + */ + if (zp->zp_rewrite) { + uint64_t orig_logical_birth = + BP_GET_LOGICAL_BIRTH(&zio->io_bp_orig); + ddt_bp_fill(ddp, v, bp, orig_logical_birth); + if (BP_EQUAL(bp, &zio->io_bp_orig)) { + /* We can skip accounting. */ + zio->io_flags |= ZIO_FLAG_NOPWRITE; + ddt_exit(ddt); + return (zio); + } + } + ddt_bp_fill(ddp, v, bp, txg); ddt_phys_addref(ddp, v); ddt_exit(ddt); return (zio); } /* * Otherwise, we have to issue IO to fill the entry up to the * amount we need. */ need_dvas -= have_dvas; } else { /* * There's a write in-flight. If there's already enough DVAs on * the entry, then either there were already enough to start * with, or the in-flight IO is between READY and DONE, and so * has extended the entry with new DVAs. Either way, we don't * need to do anything, we can just slot in behind it. */ if (zio->io_bp_override) { /* * If there's a write out, then we're soon going to * have our own copies of this block, so clear out the * override block and treat it as a regular dedup * write. See comment above. */ BP_ZERO_DVAS(bp); BP_SET_BIRTH(bp, 0, 0); } if (have_dvas >= need_dvas) { /* * A minor point: there might already be enough * committed DVAs in the entry to service our request, * but we don't know which are completed and which are * allocated but not yet written. In this case, should * the IO for the new DVAs fail, we will be on the end * of the IO chain and will also recieve an error, even * though our request could have been serviced. * * This is an extremely rare case, as it requires the * original block to be copied with a request for a * larger number of DVAs, then copied again requesting * the same (or already fulfilled) number of DVAs while * the first request is active, and then that first * request errors. In return, the logic required to * catch and handle it is complex. For now, I'm just * not going to bother with it. */ /* * We always fill the bp here as we may have arrived * after the in-flight write has passed READY, and so * missed out. */ ddt_bp_fill(ddp, v, bp, txg); zio_add_child(zio, dde->dde_io->dde_lead_zio[p]); ddt_exit(ddt); return (zio); } /* * There's not enough in the entry yet, so we need to look at * the write in-flight and see how many DVAs it will have once * it completes. * * The in-flight write has potentially had its copies request * reduced (if we're filling out an existing entry), so we need * to reach in and get the original write to find out what it is * expecting. * * Note that the parent of the lead zio will always have the * highest zp_copies of any zio in the chain, because ones that * can be serviced without additional IO are always added to * the back of the chain. */ zio_link_t *zl = NULL; zio_t *pio = zio_walk_parents(dde->dde_io->dde_lead_zio[p], &zl); ASSERT(pio); parent_dvas = pio->io_prop.zp_copies; if (parent_dvas >= need_dvas) { zio_add_child(zio, dde->dde_io->dde_lead_zio[p]); ddt_exit(ddt); return (zio); } /* * Still not enough, so we will need to issue to get the * shortfall. */ need_dvas -= parent_dvas; } if (is_ganged) { zp->zp_dedup = B_FALSE; BP_SET_DEDUP(bp, B_FALSE); zio->io_pipeline = ZIO_WRITE_PIPELINE; ddt_exit(ddt); return (zio); } /* * We need to write. We will create a new write with the copies * property adjusted to match the number of DVAs we need to need to * grow the DDT entry by to satisfy the request. */ zio_prop_t czp = *zp; if (have_dvas > 0 || parent_dvas > 0) { czp.zp_copies = need_dvas; czp.zp_gang_copies = 0; } else { ASSERT3U(czp.zp_copies, ==, need_dvas); } zio_t *cio = zio_write(zio, spa, txg, bp, zio->io_orig_abd, zio->io_orig_size, zio->io_orig_size, &czp, zio_ddt_child_write_ready, 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); /* * We are the new lead zio, because our parent has the highest * zp_copies that has been requested for this entry so far. */ ddt_alloc_entry_io(dde); if (dde->dde_io->dde_lead_zio[p] == NULL) { /* * First time out, take a copy of the stable entry to revert * to if there's an error (see zio_ddt_child_write_done()) */ ddt_phys_copy(&dde->dde_io->dde_orig_phys, dde->dde_phys, v); } else { /* * Make the existing chain our child, because it cannot * complete until we have. */ zio_add_child(cio, dde->dde_io->dde_lead_zio[p]); } dde->dde_io->dde_lead_zio[p] = cio; ddt_exit(ddt); zio_nowait(cio); return (zio); } static 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 = NULL; 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) { ddt_phys_variant_t v = ddt_phys_select(ddt, dde, bp); if (v != DDT_PHYS_NONE) ddt_phys_decref(dde->dde_phys, v); } ddt_exit(ddt); /* * When no entry was found, it must have been pruned, * so we can free it now instead of decrementing the * refcount in the DDT. */ if (!dde) { BP_SET_DEDUP(bp, 0); zio->io_pipeline |= ZIO_STAGE_DVA_FREE; } return (zio); } /* * ========================================================================== * Allocate and free blocks * ========================================================================== */ static zio_t * zio_io_to_allocate(metaslab_class_allocator_t *mca, boolean_t *more) { zio_t *zio; ASSERT(MUTEX_HELD(&mca->mca_lock)); zio = avl_first(&mca->mca_tree); if (zio == NULL) { *more = B_FALSE; return (NULL); } ASSERT(IO_IS_ALLOCATING(zio)); ASSERT(ZIO_HAS_ALLOCATOR(zio)); /* * Try to place a reservation for this zio. If we're unable to * reserve then we throttle. */ if (!metaslab_class_throttle_reserve(zio->io_metaslab_class, zio->io_allocator, zio->io_prop.zp_copies, zio->io_size, B_FALSE, more)) { return (NULL); } zio->io_flags |= ZIO_FLAG_ALLOC_THROTTLED; avl_remove(&mca->mca_tree, zio); ASSERT3U(zio->io_stage, <, ZIO_STAGE_DVA_ALLOCATE); if (avl_is_empty(&mca->mca_tree)) *more = B_FALSE; return (zio); } static zio_t * zio_dva_throttle(zio_t *zio) { spa_t *spa = zio->io_spa; zio_t *nio; metaslab_class_t *mc; boolean_t more; /* * If not already chosen, choose an appropriate allocation class. */ mc = zio->io_metaslab_class; if (mc == NULL) mc = spa_preferred_class(spa, zio); 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_type == ZIO_TYPE_WRITE); ASSERT(ZIO_HAS_ALLOCATOR(zio)); ASSERT(zio->io_child_type > ZIO_CHILD_GANG); ASSERT3U(zio->io_queued_timestamp, >, 0); ASSERT(zio->io_stage == ZIO_STAGE_DVA_THROTTLE); zio->io_metaslab_class = mc; metaslab_class_allocator_t *mca = &mc->mc_allocator[zio->io_allocator]; mutex_enter(&mca->mca_lock); avl_add(&mca->mca_tree, zio); nio = zio_io_to_allocate(mca, &more); mutex_exit(&mca->mca_lock); return (nio); } static void zio_allocate_dispatch(metaslab_class_t *mc, int allocator) { metaslab_class_allocator_t *mca = &mc->mc_allocator[allocator]; zio_t *zio; boolean_t more; do { mutex_enter(&mca->mca_lock); zio = zio_io_to_allocate(mca, &more); mutex_exit(&mca->mca_lock); 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); } while (more); } static zio_t * zio_dva_allocate(zio_t *zio) { spa_t *spa = zio->io_spa; metaslab_class_t *mc, *newmc; 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; } if (zio->io_flags & ZIO_FLAG_PREALLOCATED) { ASSERT3U(zio->io_child_type, ==, ZIO_CHILD_GANG); memcpy(zio->io_bp->blk_dva, zio->io_bp_orig.blk_dva, 3 * sizeof (dva_t)); BP_SET_LOGICAL_BIRTH(zio->io_bp, BP_GET_LOGICAL_BIRTH(&zio->io_bp_orig)); BP_SET_PHYSICAL_BIRTH(zio->io_bp, BP_GET_RAW_PHYSICAL_BIRTH(&zio->io_bp_orig)); return (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)); 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, choose an appropriate allocation class. */ mc = zio->io_metaslab_class; if (mc == NULL) { mc = spa_preferred_class(spa, zio); zio->io_metaslab_class = mc; } ZIOSTAT_BUMP(ziostat_total_allocations); again: /* * Try allocating the block in the usual metaslab class. * If that's full, allocate it in some other class(es). * If that's full, allocate as a gang block, * and if all are full, the allocation fails (which shouldn't happen). * * Note that we do not fall back on embedded slog (ZIL) space, to * preserve unfragmented slog space, which is critical for decent * sync write performance. If a log allocation fails, we will fall * back to spa_sync() which is abysmal for performance. */ ASSERT(ZIO_HAS_ALLOCATOR(zio)); error = metaslab_alloc(spa, mc, zio->io_size, bp, zio->io_prop.zp_copies, zio->io_txg, NULL, flags, &zio->io_alloc_list, zio->io_allocator, zio); /* * When the dedup or special class is spilling into the normal class, * there can still be significant space available due to deferred * frees that are in-flight. We track the txg when this occurred and * back off adding new DDT entries for a few txgs to allow the free * blocks to be processed. */ if (error == ENOSPC && spa->spa_dedup_class_full_txg != zio->io_txg && (mc == spa_dedup_class(spa) || (mc == spa_special_class(spa) && !spa_has_dedup(spa) && spa_special_has_ddt(spa)))) { spa->spa_dedup_class_full_txg = zio->io_txg; zfs_dbgmsg("%s[%llu]: %s class spilling, req size %llu, " "%llu allocated of %llu", spa_name(spa), (u_longlong_t)zio->io_txg, metaslab_class_get_name(mc), (u_longlong_t)zio->io_size, (u_longlong_t)metaslab_class_get_alloc(mc), (u_longlong_t)metaslab_class_get_space(mc)); } /* * Fall back to some other class when this one is full. */ if (error == ENOSPC && (newmc = spa_preferred_class(spa, zio)) != mc) { /* * If we are holding old class reservation, drop it. * Dispatch the next ZIO(s) there if some are waiting. */ if (zio->io_flags & ZIO_FLAG_ALLOC_THROTTLED) { if (metaslab_class_throttle_unreserve(mc, zio->io_allocator, zio->io_prop.zp_copies, zio->io_size)) { zio_allocate_dispatch(zio->io_metaslab_class, zio->io_allocator); } zio->io_flags &= ~ZIO_FLAG_ALLOC_THROTTLED; } if (zfs_flags & ZFS_DEBUG_METASLAB_ALLOC) { zfs_dbgmsg("%s: metaslab allocation failure in %s " "class, trying fallback to %s class: zio %px, " "size %llu, error %d", spa_name(spa), metaslab_class_get_name(mc), metaslab_class_get_name(newmc), zio, (u_longlong_t)zio->io_size, error); } zio->io_metaslab_class = mc = newmc; ZIOSTAT_BUMP(ziostat_alloc_class_fallbacks); /* * If the new class uses throttling, return to that pipeline * stage. Otherwise just do another allocation attempt. */ 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)) { zio->io_stage = ZIO_STAGE_DVA_THROTTLE >> 1; return (zio); } goto again; } if (error == ENOSPC && zio->io_size > spa->spa_min_alloc) { if (zfs_flags & ZFS_DEBUG_METASLAB_ALLOC) { zfs_dbgmsg("%s: metaslab allocation failure, " "trying ganging: zio %px, size %llu, error %d", spa_name(spa), zio, (u_longlong_t)zio->io_size, error); } ZIOSTAT_BUMP(ziostat_gang_writes); if (flags & METASLAB_GANG_CHILD) ZIOSTAT_BUMP(ziostat_gang_multilevel); return (zio_write_gang_block(zio, mc)); } if (error != 0) { if (error != ENOSPC || (zfs_flags & ZFS_DEBUG_METASLAB_ALLOC)) { zfs_dbgmsg("%s: metaslab allocation failure: zio %px, " "size %llu, error %d", spa_name(spa), zio, (u_longlong_t)zio->io_size, error); } zio->io_error = error; + } else if (zio->io_prop.zp_rewrite) { + /* + * For rewrite operations, preserve the logical birth time + * but set the physical birth time to the current txg. + */ + uint64_t logical_birth = BP_GET_LOGICAL_BIRTH(&zio->io_bp_orig); + ASSERT3U(logical_birth, <=, zio->io_txg); + BP_SET_BIRTH(zio->io_bp, logical_birth, zio->io_txg); + BP_SET_REWRITE(zio->io_bp, 1); } 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_GET_BIRTH(bp) == 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_GET_BIRTH(bp), B_TRUE); } if (gn != NULL) { for (int g = 0; g < gbh_nblkptrs(gn->gn_gangblocksize); g++) { zio_dva_unallocate(zio, gn->gn_child[g], gbh_bp(gn->gn_gbh, 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; 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. */ int flags = METASLAB_ZIL; int allocator = (uint_t)cityhash1(os->os_dsl_dataset->ds_object) % spa->spa_alloc_count; ZIOSTAT_BUMP(ziostat_total_allocations); /* Try log class (dedicated slog devices) first */ error = metaslab_alloc(spa, spa_log_class(spa), size, new_bp, 1, txg, NULL, flags, &io_alloc_list, allocator, NULL); *slog = (error == 0); /* Try special_embedded_log class (reserved on special vdevs) */ if (error != 0) { error = metaslab_alloc(spa, spa_special_embedded_log_class(spa), size, new_bp, 1, txg, NULL, flags, &io_alloc_list, allocator, NULL); } /* Try special class (general special vdev allocation) */ if (error != 0) { error = metaslab_alloc(spa, spa_special_class(spa), size, new_bp, 1, txg, NULL, flags, &io_alloc_list, allocator, NULL); } /* Try embedded_log class (reserved on normal vdevs) */ if (error != 0) { error = metaslab_alloc(spa, spa_embedded_log_class(spa), size, new_bp, 1, txg, NULL, flags, &io_alloc_list, allocator, NULL); } /* Finally fall back to normal class */ if (error != 0) { ZIOSTAT_BUMP(ziostat_alloc_class_fallbacks); error = metaslab_alloc(spa, spa_normal_class(spa), size, new_bp, 1, txg, NULL, flags, &io_alloc_list, allocator, NULL); } 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), (u_longlong_t)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_noalloc) { 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. * * Leaf DTL_PARTIAL can be empty when a legitimate write comes from * a dRAID spare vdev. For example, when a dRAID spare is first * used, its spare blocks need to be written to but the leaf vdev's * of such blocks can have empty DTL_PARTIAL. * * There seemed no clean way to allow such writes while bypassing * spurious ones. At this point, just avoid all bypassing for dRAID * for correctness. */ 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 && vd->vdev_top->vdev_ops != &vdev_draid_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); } /* * Select the next best leaf I/O to process. Distributed spares are * excluded since they dispatch the I/O directly to a leaf vdev after * applying the dRAID mapping. */ if (vd->vdev_ops->vdev_op_leaf && vd->vdev_ops != &vdev_draid_spare_ops && (zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE || zio->io_type == ZIO_TYPE_TRIM)) { 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(); if (zio_handle_device_injection(vd, zio, ENOSYS) != 0) { /* * "no-op" injections return success, but do no actual * work. Just return it. */ zio_delay_interrupt(zio); return (NULL); } } 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_FLUSH || 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 && vd->vdev_ops != &vdev_draid_spare_ops) { if (zio->io_type != ZIO_TYPE_FLUSH) vdev_queue_io_done(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_FLUSH && 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 && vd->vdev_remove_wanted == B_FALSE) 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); } void zio_vsd_default_cksum_report(zio_t *zio, zio_cksum_report_t *zcr) { 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 a Direct I/O operation has a checksum verify error then this I/O * should not attempt to be issued again. */ if (zio->io_post & ZIO_POST_DIO_CHKSUM_ERR) { if (zio->io_type == ZIO_TYPE_WRITE) { ASSERT3U(zio->io_child_type, ==, ZIO_CHILD_LOGICAL); ASSERT3U(zio->io_error, ==, EIO); } zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; return (zio); } 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_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) { vdev_dbgmsg(vd, "zio_vdev_io_assess(zio=%px) setting " "cant_write=TRUE due to write failure with ENXIO", zio); vd->vdev_cant_write = B_TRUE; } /* * If a cache flush returns ENOTSUP 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, and * then we act like the flush succeeded. */ if (zio->io_error == ENOTSUP && zio->io_type == ZIO_TYPE_FLUSH && vd != NULL) { vd->vdev_nowritecache = B_TRUE; zio->io_error = 0; } if (zio->io_error) zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; 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); ASSERT3U(zio->io_prop.zp_checksum, ==, ZIO_CHECKSUM_LABEL); } ASSERT0(zio->io_post & ZIO_POST_DIO_CHKSUM_ERR); IMPLY(zio->io_flags & ZIO_FLAG_DIO_READ, !(zio->io_flags & ZIO_FLAG_SPECULATIVE)); if ((error = zio_checksum_error(zio, &info)) != 0) { zio->io_error = error; if (error == ECKSUM && !(zio->io_flags & ZIO_FLAG_SPECULATIVE)) { if (zio->io_flags & ZIO_FLAG_DIO_READ) { zio->io_post |= ZIO_POST_DIO_CHKSUM_ERR; zio_t *pio = zio_unique_parent(zio); /* * Any Direct I/O read that has a checksum * error must be treated as suspicous as the * contents of the buffer could be getting * manipulated while the I/O is taking place. * * The checksum verify error will only be * reported here for disk and file VDEV's and * will be reported on those that the failure * occurred on. Other types of VDEV's report the * verify failure in their own code paths. */ if (pio->io_child_type == ZIO_CHILD_LOGICAL) { zio_dio_chksum_verify_error_report(zio); } } else { mutex_enter(&zio->io_vd->vdev_stat_lock); zio->io_vd->vdev_stat.vs_checksum_errors++; mutex_exit(&zio->io_vd->vdev_stat_lock); (void) zfs_ereport_start_checksum(zio->io_spa, zio->io_vd, &zio->io_bookmark, zio, zio->io_offset, zio->io_size, &info); } } } return (zio); } static zio_t * zio_dio_checksum_verify(zio_t *zio) { zio_t *pio = zio_unique_parent(zio); int error; ASSERT3P(zio->io_vd, !=, NULL); ASSERT3P(zio->io_bp, !=, NULL); ASSERT3U(zio->io_child_type, ==, ZIO_CHILD_VDEV); ASSERT3U(zio->io_type, ==, ZIO_TYPE_WRITE); ASSERT3B(pio->io_prop.zp_direct_write, ==, B_TRUE); ASSERT3U(pio->io_child_type, ==, ZIO_CHILD_LOGICAL); if (zfs_vdev_direct_write_verify == 0 || zio->io_error != 0) goto out; if ((error = zio_checksum_error(zio, NULL)) != 0) { zio->io_error = error; if (error == ECKSUM) { zio->io_post |= ZIO_POST_DIO_CHKSUM_ERR; zio_dio_chksum_verify_error_report(zio); } } out: 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; } /* * Report Direct I/O checksum verify error and create ZED event. */ void zio_dio_chksum_verify_error_report(zio_t *zio) { ASSERT(zio->io_post & ZIO_POST_DIO_CHKSUM_ERR); if (zio->io_child_type == ZIO_CHILD_LOGICAL) return; mutex_enter(&zio->io_vd->vdev_stat_lock); zio->io_vd->vdev_stat.vs_dio_verify_errors++; mutex_exit(&zio->io_vd->vdev_stat_lock); if (zio->io_type == ZIO_TYPE_WRITE) { /* * Convert checksum error for writes into EIO. */ zio->io_error = SET_ERROR(EIO); /* * Report dio_verify_wr ZED event. */ (void) zfs_ereport_post(FM_EREPORT_ZFS_DIO_VERIFY_WR, zio->io_spa, zio->io_vd, &zio->io_bookmark, zio, 0); } else { /* * Report dio_verify_rd ZED event. */ (void) zfs_ereport_post(FM_EREPORT_ZFS_DIO_VERIFY_RD, zio->io_spa, zio->io_vd, &zio->io_bookmark, zio, 0); } } /* * ========================================================================== * 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_LOGICAL_BIT | ZIO_CHILD_GANG_BIT | ZIO_CHILD_DDT_BIT, ZIO_WAIT_READY)) { return (NULL); } if (zio->io_ready) { ASSERT(IO_IS_ALLOCATING(zio)); ASSERT(BP_GET_BIRTH(bp) == 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); } #ifdef ZFS_DEBUG if (bp != NULL && bp != &zio->io_bp_copy) zio->io_bp_copy = *bp; #endif if (zio->io_error != 0) { zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (zio->io_flags & ZIO_FLAG_ALLOC_THROTTLED) { ASSERT(IO_IS_ALLOCATING(zio)); ASSERT(zio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(zio->io_metaslab_class != NULL); ASSERT(ZIO_HAS_ALLOCATOR(zio)); /* * We were unable to allocate anything, unreserve and * issue the next I/O to allocate. */ if (metaslab_class_throttle_unreserve( zio->io_metaslab_class, zio->io_allocator, zio->io_prop.zp_copies, zio->io_size)) { zio_allocate_dispatch(zio->io_metaslab_class, 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 != NULL && 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 *pio = zio_unique_parent(zio); vdev_t *vd = zio->io_vd; int flags = METASLAB_ASYNC_ALLOC; const void *tag = pio; uint64_t size = pio->io_size; 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_ALLOC_THROTTLED); /* * 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 first use their parent as tag. * We set the size to match the original allocation call for that case. */ if (pio->io_child_type == ZIO_CHILD_GANG && (pio->io_flags & ZIO_FLAG_IO_REWRITE)) { tag = zio_unique_parent(pio); size = SPA_OLD_GANGBLOCKSIZE; } ASSERT(IO_IS_ALLOCATING(pio) || (pio->io_child_type == ZIO_CHILD_GANG && (pio->io_flags & ZIO_FLAG_IO_REWRITE))); ASSERT(ZIO_HAS_ALLOCATOR(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); ASSERT(zio->io_metaslab_class->mc_alloc_throttle_enabled); metaslab_group_alloc_decrement(zio->io_spa, vd->vdev_id, pio->io_allocator, flags, size, tag); if (metaslab_class_throttle_unreserve(pio->io_metaslab_class, pio->io_allocator, 1, pio->io_size)) { zio_allocate_dispatch(zio->io_metaslab_class, 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_ALLOC_THROTTLED && zio->io_child_type == ZIO_CHILD_VDEV) zio_dva_throttle_done(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(memcmp(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 (adata != NULL && 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 (adata != NULL && 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) && !(zio->io_post & ZIO_POST_DIO_CHKSUM_ERR)) { 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->io_post & ZIO_POST_DIO_CHKSUM_ERR) && 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, BP_GET_PHYSICAL_BIRTH(zio->io_bp)); (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) { /* * A DDT child tried to create a mixed gang/non-gang BP. We're * going to have to just retry as a non-dedup IO. */ if (zio->io_error == EAGAIN && IO_IS_ALLOCATING(zio) && zio->io_prop.zp_dedup) { zio->io_post |= ZIO_POST_REEXECUTE; zio->io_prop.zp_dedup = B_FALSE; } /* * 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) && !(zio->io_post & ZIO_POST_DIO_CHKSUM_ERR)) { if (zio->io_error != ENOSPC) zio->io_post |= ZIO_POST_REEXECUTE; else zio->io_post |= ZIO_POST_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_post |= ZIO_POST_SUSPEND; if (!(zio->io_flags & ZIO_FLAG_CANFAIL) && !(zio->io_post & (ZIO_POST_REEXECUTE|ZIO_POST_SUSPEND))) zio->io_post |= ZIO_POST_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_post & (ZIO_POST_REEXECUTE|ZIO_POST_SUSPEND))) && 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_post & ZIO_POST_SUSPEND)) zio->io_post &= ~ZIO_POST_SUSPEND; if (zio->io_post & (ZIO_POST_REEXECUTE|ZIO_POST_SUSPEND)) { /* * A Direct I/O operation that has a checksum verify error * should not attempt to reexecute. Instead, the error should * just be propagated back. */ ASSERT0(zio->io_post & ZIO_POST_DIO_CHKSUM_ERR); /* * 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_post & ZIO_POST_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_post & ZIO_POST_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 { ASSERT(zio->io_post & ZIO_POST_REEXECUTE); /* * Reexecution is potentially a huge amount of work. * Hand it off to the otherwise-unused claim taskq. */ spa_taskq_dispatch(zio->io_spa, ZIO_TYPE_CLAIM, ZIO_TASKQ_ISSUE, zio_reexecute, zio, B_FALSE); } return (NULL); } ASSERT(list_is_empty(&zio->io_child_list)); ASSERT0(zio->io_post & ZIO_POST_REEXECUTE); ASSERT0(zio->io_post & ZIO_POST_SUSPEND); 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); } /* * 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_brt_free, 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_dio_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++; ASSERT0(last_block->zb_level); /* 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); } /* * This function is similar to zbookmark_subtree_completed(), but returns true * if subtree_root is equal or ahead of last_block, i.e. still to be done. */ boolean_t zbookmark_subtree_tbd(const dnode_phys_t *dnp, const zbookmark_phys_t *subtree_root, const zbookmark_phys_t *last_block) { ASSERT0(last_block->zb_level); if (dnp == NULL) return (B_FALSE); return (zbookmark_compare(dnp->dn_datablkszsec, dnp->dn_indblkshift, 1ULL << (DNODE_BLOCK_SHIFT - SPA_MINBLOCKSHIFT), 0, subtree_root, 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); 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, UINT, ZMOD_RW, "Defer frees starting in this pass"); ZFS_MODULE_PARAM(zfs, zfs_, sync_pass_dont_compress, UINT, ZMOD_RW, "Don't compress starting in this pass"); ZFS_MODULE_PARAM(zfs, zfs_, sync_pass_rewrite, UINT, 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"); diff --git a/tests/runfiles/common.run b/tests/runfiles/common.run index deca3c05b073..9fad8946f4f3 100644 --- a/tests/runfiles/common.run +++ b/tests/runfiles/common.run @@ -1,1117 +1,1117 @@ # SPDX-License-Identifier: CDDL-1.0 # # This file and its contents are supplied under the terms of the # Common Development and Distribution License ("CDDL"), version 1.0. # You may only use this file in accordance with the terms of version # 1.0 of the CDDL. # # A full copy of the text of the CDDL should have accompanied this # source. A copy of the CDDL is also available via the Internet at # http://www.illumos.org/license/CDDL. # # This run file contains all of the common functional tests. When # adding a new test consider also adding it to the sanity.run file # if the new test runs to completion in only a few seconds. # # Approximate run time: 4-5 hours # [DEFAULT] pre = setup quiet = False pre_user = root user = root timeout = 600 post_user = root post = cleanup failsafe_user = root failsafe = callbacks/zfs_failsafe tags = ['functional'] [tests/functional/acl/off] tests = ['dosmode', 'posixmode'] tags = ['functional', 'acl'] [tests/functional/alloc_class] tests = ['alloc_class_001_pos', 'alloc_class_002_neg', 'alloc_class_003_pos', 'alloc_class_004_pos', 'alloc_class_005_pos', 'alloc_class_006_pos', 'alloc_class_007_pos', 'alloc_class_008_pos', 'alloc_class_009_pos', 'alloc_class_010_pos', 'alloc_class_011_neg', 'alloc_class_012_pos', 'alloc_class_013_pos', 'alloc_class_016_pos'] tags = ['functional', 'alloc_class'] [tests/functional/append] tests = ['file_append', 'threadsappend_001_pos'] tags = ['functional', 'append'] [tests/functional/arc] tests = ['dbufstats_001_pos', 'dbufstats_002_pos', 'dbufstats_003_pos', 'arcstats_runtime_tuning'] tags = ['functional', 'arc'] [tests/functional/atime] tests = ['atime_001_pos', 'atime_002_neg', 'root_atime_off', 'root_atime_on'] tags = ['functional', 'atime'] [tests/functional/bclone] tests = ['bclone_crossfs_corner_cases_limited', 'bclone_crossfs_data', 'bclone_crossfs_embedded', 'bclone_crossfs_hole', 'bclone_diffprops_all', 'bclone_diffprops_checksum', 'bclone_diffprops_compress', 'bclone_diffprops_copies', 'bclone_diffprops_recordsize', 'bclone_prop_sync', 'bclone_samefs_corner_cases_limited', 'bclone_samefs_data', 'bclone_samefs_embedded', 'bclone_samefs_hole'] tags = ['functional', 'bclone'] timeout = 7200 [tests/functional/block_cloning] tests = ['block_cloning_clone_mmap_cached', 'block_cloning_copyfilerange', 'block_cloning_copyfilerange_partial', 'block_cloning_copyfilerange_fallback', 'block_cloning_disabled_copyfilerange', 'block_cloning_copyfilerange_cross_dataset', 'block_cloning_cross_enc_dataset', 'block_cloning_copyfilerange_fallback_same_txg', 'block_cloning_replay', 'block_cloning_replay_encrypted', 'block_cloning_lwb_buffer_overflow', 'block_cloning_clone_mmap_write', 'block_cloning_rlimit_fsize', 'block_cloning_large_offset'] tags = ['functional', 'block_cloning'] [tests/functional/bootfs] tests = ['bootfs_001_pos', 'bootfs_002_neg', 'bootfs_003_pos', 'bootfs_004_neg', 'bootfs_005_neg', 'bootfs_006_pos', 'bootfs_007_pos', 'bootfs_008_pos'] tags = ['functional', 'bootfs'] [tests/functional/btree] tests = ['btree_positive', 'btree_negative'] tags = ['functional', 'btree'] pre = post = [tests/functional/cache] tests = ['cache_001_pos', 'cache_002_pos', 'cache_003_pos', 'cache_004_neg', 'cache_005_neg', 'cache_006_pos', 'cache_007_neg', 'cache_008_neg', 'cache_009_pos', 'cache_010_pos', 'cache_011_pos', 'cache_012_pos'] tags = ['functional', 'cache'] [tests/functional/cachefile] tests = ['cachefile_001_pos', 'cachefile_002_pos', 'cachefile_003_pos', 'cachefile_004_pos'] tags = ['functional', 'cachefile'] [tests/functional/casenorm] tests = ['case_all_values', 'norm_all_values', 'mixed_create_failure', 'sensitive_none_lookup', 'sensitive_none_delete', 'sensitive_formd_lookup', 'sensitive_formd_delete', 'insensitive_none_lookup', 'insensitive_none_delete', 'insensitive_formd_lookup', 'insensitive_formd_delete', 'mixed_none_lookup', 'mixed_none_lookup_ci', 'mixed_none_delete', 'mixed_formd_lookup', 'mixed_formd_lookup_ci', 'mixed_formd_delete'] tags = ['functional', 'casenorm'] [tests/functional/channel_program/lua_core] tests = ['tst.args_to_lua', 'tst.divide_by_zero', 'tst.exists', 'tst.encryption', 'tst.integer_illegal', 'tst.integer_overflow', 'tst.language_functions_neg', 'tst.language_functions_pos', 'tst.large_prog', 'tst.libraries', 'tst.memory_limit', 'tst.nested_neg', 'tst.nested_pos', 'tst.nvlist_to_lua', 'tst.recursive_neg', 'tst.recursive_pos', 'tst.return_large', 'tst.return_nvlist_neg', 'tst.return_nvlist_pos', 'tst.return_recursive_table', 'tst.stack_gsub', 'tst.timeout'] tags = ['functional', 'channel_program', 'lua_core'] [tests/functional/channel_program/synctask_core] tests = ['tst.destroy_fs', 'tst.destroy_snap', 'tst.get_count_and_limit', 'tst.get_index_props', 'tst.get_mountpoint', 'tst.get_neg', 'tst.get_number_props', 'tst.get_string_props', 'tst.get_type', 'tst.get_userquota', 'tst.get_written', 'tst.inherit', 'tst.list_bookmarks', 'tst.list_children', 'tst.list_clones', 'tst.list_holds', 'tst.list_snapshots', 'tst.list_system_props', 'tst.list_user_props', 'tst.parse_args_neg','tst.promote_conflict', 'tst.promote_multiple', 'tst.promote_simple', 'tst.rollback_mult', 'tst.rollback_one', 'tst.set_props', 'tst.snapshot_destroy', 'tst.snapshot_neg', 'tst.snapshot_recursive', 'tst.snapshot_rename', 'tst.snapshot_simple', 'tst.bookmark.create', 'tst.bookmark.copy', 'tst.clone', 'tst.terminate_by_signal' ] tags = ['functional', 'channel_program', 'synctask_core'] [tests/functional/checksum] tests = ['run_edonr_test', 'run_sha2_test', 'run_skein_test', 'run_blake3_test', 'filetest_001_pos', 'filetest_002_pos'] tags = ['functional', 'checksum'] [tests/functional/clean_mirror] tests = [ 'clean_mirror_001_pos', 'clean_mirror_002_pos', 'clean_mirror_003_pos', 'clean_mirror_004_pos'] tags = ['functional', 'clean_mirror'] [tests/functional/cli_root/json] tests = ['json_sanity'] tags = ['functional', 'cli_root', 'json'] [tests/functional/cli_root/zinject] tests = ['zinject_args', 'zinject_counts', 'zinject_probe'] pre = post = tags = ['functional', 'cli_root', 'zinject'] [tests/functional/cli_root/zdb] tests = ['zdb_002_pos', 'zdb_003_pos', 'zdb_004_pos', 'zdb_005_pos', 'zdb_006_pos', 'zdb_args_neg', 'zdb_args_pos', 'zdb_block_size_histogram', 'zdb_checksum', 'zdb_decompress', 'zdb_display_block', 'zdb_encrypted', 'zdb_label_checksum', 'zdb_object_range_neg', 'zdb_object_range_pos', 'zdb_objset_id', 'zdb_decompress_zstd', 'zdb_recover', 'zdb_recover_2', 'zdb_backup', 'zdb_tunables'] pre = post = tags = ['functional', 'cli_root', 'zdb'] timeout = 1200 [tests/functional/cli_root/zfs] tests = ['zfs_001_neg', 'zfs_002_pos'] tags = ['functional', 'cli_root', 'zfs'] [tests/functional/cli_root/zfs_bookmark] tests = ['zfs_bookmark_cliargs'] tags = ['functional', 'cli_root', 'zfs_bookmark'] [tests/functional/cli_root/zfs_change-key] tests = ['zfs_change-key', 'zfs_change-key_child', 'zfs_change-key_format', 'zfs_change-key_inherit', 'zfs_change-key_load', 'zfs_change-key_location', 'zfs_change-key_pbkdf2iters', 'zfs_change-key_clones'] tags = ['functional', 'cli_root', 'zfs_change-key'] [tests/functional/cli_root/zfs_clone] tests = ['zfs_clone_001_neg', 'zfs_clone_002_pos', 'zfs_clone_003_pos', 'zfs_clone_004_pos', 'zfs_clone_005_pos', 'zfs_clone_006_pos', 'zfs_clone_007_pos', 'zfs_clone_008_neg', 'zfs_clone_009_neg', 'zfs_clone_010_pos', 'zfs_clone_encrypted', 'zfs_clone_deeply_nested', 'zfs_clone_rm_nested'] tags = ['functional', 'cli_root', 'zfs_clone'] [tests/functional/cli_root/zfs_copies] tests = ['zfs_copies_001_pos', 'zfs_copies_002_pos', 'zfs_copies_003_pos', 'zfs_copies_004_neg', 'zfs_copies_005_neg', 'zfs_copies_006_pos'] tags = ['functional', 'cli_root', 'zfs_copies'] [tests/functional/cli_root/zfs_create] tests = ['zfs_create_001_pos', 'zfs_create_002_pos', 'zfs_create_003_pos', 'zfs_create_004_pos', 'zfs_create_005_pos', 'zfs_create_006_pos', 'zfs_create_007_pos', 'zfs_create_008_neg', 'zfs_create_009_neg', 'zfs_create_010_neg', 'zfs_create_011_pos', 'zfs_create_012_pos', 'zfs_create_013_pos', 'zfs_create_014_pos', 'zfs_create_encrypted', 'zfs_create_crypt_combos', 'zfs_create_dryrun', 'zfs_create_nomount', 'zfs_create_verbose'] tags = ['functional', 'cli_root', 'zfs_create'] [tests/functional/cli_root/zpool_prefetch] tests = ['zpool_prefetch_001_pos'] tags = ['functional', 'cli_root', 'zpool_prefetch'] [tests/functional/cli_root/zfs_destroy] tests = ['zfs_clone_livelist_condense_and_disable', 'zfs_clone_livelist_condense_races', 'zfs_clone_livelist_dedup', 'zfs_destroy_001_pos', 'zfs_destroy_002_pos', 'zfs_destroy_003_pos', 'zfs_destroy_004_pos', 'zfs_destroy_005_neg', 'zfs_destroy_006_neg', 'zfs_destroy_007_neg', 'zfs_destroy_008_pos', 'zfs_destroy_009_pos', 'zfs_destroy_010_pos', 'zfs_destroy_011_pos', 'zfs_destroy_012_pos', 'zfs_destroy_013_neg', 'zfs_destroy_014_pos', 'zfs_destroy_015_pos', 'zfs_destroy_016_pos', 'zfs_destroy_clone_livelist', 'zfs_destroy_dev_removal', 'zfs_destroy_dev_removal_condense'] tags = ['functional', 'cli_root', 'zfs_destroy'] [tests/functional/cli_root/zfs_diff] tests = ['zfs_diff_changes', 'zfs_diff_cliargs', 'zfs_diff_timestamp', 'zfs_diff_types', 'zfs_diff_encrypted', 'zfs_diff_mangle'] tags = ['functional', 'cli_root', 'zfs_diff'] [tests/functional/cli_root/zfs_get] tests = ['zfs_get_001_pos', 'zfs_get_002_pos', 'zfs_get_003_pos', 'zfs_get_004_pos', 'zfs_get_005_neg', 'zfs_get_006_neg', 'zfs_get_007_neg', 'zfs_get_008_pos', 'zfs_get_009_pos', 'zfs_get_010_neg'] tags = ['functional', 'cli_root', 'zfs_get'] [tests/functional/cli_root/zfs_ids_to_path] tests = ['zfs_ids_to_path_001_pos'] tags = ['functional', 'cli_root', 'zfs_ids_to_path'] [tests/functional/cli_root/zfs_inherit] tests = ['zfs_inherit_001_neg', 'zfs_inherit_002_neg', 'zfs_inherit_003_pos', 'zfs_inherit_mountpoint'] tags = ['functional', 'cli_root', 'zfs_inherit'] [tests/functional/cli_root/zfs_load-key] tests = ['zfs_load-key', 'zfs_load-key_all', 'zfs_load-key_file', 'zfs_load-key_https', 'zfs_load-key_location', 'zfs_load-key_noop', 'zfs_load-key_recursive'] tags = ['functional', 'cli_root', 'zfs_load-key'] [tests/functional/cli_root/zfs_mount] tests = ['zfs_mount_001_pos', 'zfs_mount_002_pos', 'zfs_mount_003_pos', 'zfs_mount_004_pos', 'zfs_mount_005_pos', 'zfs_mount_007_pos', 'zfs_mount_009_neg', 'zfs_mount_010_neg', 'zfs_mount_011_neg', 'zfs_mount_012_pos', 'zfs_mount_all_001_pos', 'zfs_mount_encrypted', 'zfs_mount_remount', 'zfs_mount_all_fail', 'zfs_mount_all_mountpoints', 'zfs_mount_test_race', 'zfs_mount_recursive'] tags = ['functional', 'cli_root', 'zfs_mount'] [tests/functional/cli_root/zfs_program] tests = ['zfs_program_json'] tags = ['functional', 'cli_root', 'zfs_program'] [tests/functional/cli_root/zfs_promote] tests = ['zfs_promote_001_pos', 'zfs_promote_002_pos', 'zfs_promote_003_pos', 'zfs_promote_004_pos', 'zfs_promote_005_pos', 'zfs_promote_006_neg', 'zfs_promote_007_neg', 'zfs_promote_008_pos', 'zfs_promote_encryptionroot'] tags = ['functional', 'cli_root', 'zfs_promote'] [tests/functional/cli_root/zfs_property] tests = ['zfs_written_property_001_pos'] tags = ['functional', 'cli_root', 'zfs_property'] [tests/functional/cli_root/zfs_receive] tests = ['zfs_receive_001_pos', 'zfs_receive_002_pos', 'zfs_receive_003_pos', 'zfs_receive_004_neg', 'zfs_receive_005_neg', 'zfs_receive_006_pos', 'zfs_receive_007_neg', 'zfs_receive_008_pos', 'zfs_receive_009_neg', 'zfs_receive_010_pos', 'zfs_receive_011_pos', 'zfs_receive_012_pos', 'zfs_receive_013_pos', 'zfs_receive_014_pos', 'zfs_receive_015_pos', 'zfs_receive_016_pos', 'receive-o-x_props_override', 'receive-o-x_props_aliases', 'zfs_receive_from_encrypted', 'zfs_receive_to_encrypted', 'zfs_receive_raw', 'zfs_receive_raw_incremental', 'zfs_receive_-e', 'zfs_receive_raw_-d', 'zfs_receive_from_zstd', 'zfs_receive_new_props', 'zfs_receive_-wR-encrypted-mix', 'zfs_receive_corrective', 'zfs_receive_compressed_corrective', 'zfs_receive_large_block_corrective'] tags = ['functional', 'cli_root', 'zfs_receive'] [tests/functional/cli_root/zfs_rename] tests = ['zfs_rename_001_pos', 'zfs_rename_002_pos', 'zfs_rename_003_pos', 'zfs_rename_004_neg', 'zfs_rename_005_neg', 'zfs_rename_006_pos', 'zfs_rename_007_pos', 'zfs_rename_008_pos', 'zfs_rename_009_neg', 'zfs_rename_010_neg', 'zfs_rename_011_pos', 'zfs_rename_012_neg', 'zfs_rename_013_pos', 'zfs_rename_014_neg', 'zfs_rename_encrypted_child', 'zfs_rename_to_encrypted', 'zfs_rename_mountpoint', 'zfs_rename_nounmount'] tags = ['functional', 'cli_root', 'zfs_rename'] [tests/functional/cli_root/zfs_reservation] tests = ['zfs_reservation_001_pos', 'zfs_reservation_002_pos'] tags = ['functional', 'cli_root', 'zfs_reservation'] [tests/functional/cli_root/zfs_rewrite] -tests = ['zfs_rewrite'] +tests = ['zfs_rewrite', 'zfs_rewrite_physical'] tags = ['functional', 'cli_root', 'zfs_rewrite'] [tests/functional/cli_root/zfs_rollback] tests = ['zfs_rollback_001_pos', 'zfs_rollback_002_pos', 'zfs_rollback_003_neg', 'zfs_rollback_004_neg'] tags = ['functional', 'cli_root', 'zfs_rollback'] [tests/functional/cli_root/zfs_send] tests = ['zfs_send_001_pos', 'zfs_send_002_pos', 'zfs_send_003_pos', 'zfs_send_004_neg', 'zfs_send_005_pos', 'zfs_send_006_pos', 'zfs_send_007_pos', 'zfs_send_encrypted', 'zfs_send_encrypted_unloaded', 'zfs_send_raw', 'zfs_send_sparse', 'zfs_send-b', 'zfs_send_skip_missing'] tags = ['functional', 'cli_root', 'zfs_send'] [tests/functional/cli_root/zfs_set] tests = ['cache_001_pos', 'cache_002_neg', 'canmount_001_pos', 'canmount_002_pos', 'canmount_003_pos', 'canmount_004_pos', 'checksum_001_pos', 'compression_001_pos', 'mountpoint_001_pos', 'mountpoint_002_pos', 'reservation_001_neg', 'user_property_002_pos', 'share_mount_001_neg', 'snapdir_001_pos', 'onoffs_001_pos', 'user_property_001_pos', 'user_property_003_neg', 'readonly_001_pos', 'user_property_004_pos', 'version_001_neg', 'zfs_set_001_neg', 'zfs_set_002_neg', 'zfs_set_003_neg', 'property_alias_001_pos', 'mountpoint_003_pos', 'ro_props_001_pos', 'zfs_set_keylocation', 'zfs_set_feature_activation', 'zfs_set_nomount'] tags = ['functional', 'cli_root', 'zfs_set'] [tests/functional/cli_root/zfs_share] tests = ['zfs_share_001_pos', 'zfs_share_002_pos', 'zfs_share_003_pos', 'zfs_share_004_pos', 'zfs_share_006_pos', 'zfs_share_008_neg', 'zfs_share_010_neg', 'zfs_share_011_pos', 'zfs_share_concurrent_shares', 'zfs_share_after_mount'] tags = ['functional', 'cli_root', 'zfs_share'] [tests/functional/cli_root/zfs_snapshot] tests = ['zfs_snapshot_001_neg', 'zfs_snapshot_002_neg', 'zfs_snapshot_003_neg', 'zfs_snapshot_004_neg', 'zfs_snapshot_005_neg', 'zfs_snapshot_006_pos', 'zfs_snapshot_007_neg', 'zfs_snapshot_008_neg', 'zfs_snapshot_009_pos'] tags = ['functional', 'cli_root', 'zfs_snapshot'] [tests/functional/cli_root/zfs_unload-key] tests = ['zfs_unload-key', 'zfs_unload-key_all', 'zfs_unload-key_recursive'] tags = ['functional', 'cli_root', 'zfs_unload-key'] [tests/functional/cli_root/zfs_unmount] tests = ['zfs_unmount_001_pos', 'zfs_unmount_002_pos', 'zfs_unmount_003_pos', 'zfs_unmount_004_pos', 'zfs_unmount_005_pos', 'zfs_unmount_006_pos', 'zfs_unmount_007_neg', 'zfs_unmount_008_neg', 'zfs_unmount_009_pos', 'zfs_unmount_all_001_pos', 'zfs_unmount_nested', 'zfs_unmount_unload_keys'] tags = ['functional', 'cli_root', 'zfs_unmount'] [tests/functional/cli_root/zfs_unshare] tests = ['zfs_unshare_001_pos', 'zfs_unshare_002_pos', 'zfs_unshare_003_pos', 'zfs_unshare_004_neg', 'zfs_unshare_005_neg', 'zfs_unshare_006_pos', 'zfs_unshare_007_pos'] tags = ['functional', 'cli_root', 'zfs_unshare'] [tests/functional/cli_root/zfs_upgrade] tests = ['zfs_upgrade_001_pos', 'zfs_upgrade_002_pos', 'zfs_upgrade_003_pos', 'zfs_upgrade_004_pos', 'zfs_upgrade_005_pos', 'zfs_upgrade_006_neg', 'zfs_upgrade_007_neg'] tags = ['functional', 'cli_root', 'zfs_upgrade'] [tests/functional/cli_root/zfs_wait] tests = ['zfs_wait_deleteq', 'zfs_wait_getsubopt'] tags = ['functional', 'cli_root', 'zfs_wait'] [tests/functional/cli_root/zhack] tests = ['zhack_label_repair_001', 'zhack_label_repair_002', 'zhack_label_repair_003', 'zhack_label_repair_004'] pre = post = tags = ['functional', 'cli_root', 'zhack'] [tests/functional/cli_root/zpool] tests = ['zpool_001_neg', 'zpool_002_pos', 'zpool_003_pos', 'zpool_colors'] tags = ['functional', 'cli_root', 'zpool'] [tests/functional/cli_root/zpool_add] tests = ['zpool_add_001_pos', 'zpool_add_002_pos', 'zpool_add_003_pos', 'zpool_add_004_pos', 'zpool_add_006_pos', 'zpool_add_007_neg', 'zpool_add_008_neg', 'zpool_add_009_neg', 'zpool_add_010_pos', 'add-o_ashift', 'add_prop_ashift', 'zpool_add_dryrun_output'] tags = ['functional', 'cli_root', 'zpool_add'] [tests/functional/cli_root/zpool_attach] tests = ['zpool_attach_001_neg', 'attach-o_ashift'] tags = ['functional', 'cli_root', 'zpool_attach'] [tests/functional/cli_root/zpool_clear] tests = ['zpool_clear_001_pos', 'zpool_clear_002_neg', 'zpool_clear_003_neg', 'zpool_clear_readonly'] tags = ['functional', 'cli_root', 'zpool_clear'] [tests/functional/cli_root/zpool_create] tests = ['zpool_create_001_pos', 'zpool_create_002_pos', 'zpool_create_003_pos', 'zpool_create_004_pos', 'zpool_create_005_pos', 'zpool_create_006_pos', 'zpool_create_007_neg', 'zpool_create_008_pos', 'zpool_create_009_neg', 'zpool_create_010_neg', 'zpool_create_011_neg', 'zpool_create_012_neg', 'zpool_create_014_neg', 'zpool_create_015_neg', 'zpool_create_017_neg', 'zpool_create_018_pos', 'zpool_create_019_pos', 'zpool_create_020_pos', 'zpool_create_021_pos', 'zpool_create_022_pos', 'zpool_create_023_neg', 'zpool_create_024_pos', 'zpool_create_encrypted', 'zpool_create_crypt_combos', 'zpool_create_draid_001_pos', 'zpool_create_draid_002_pos', 'zpool_create_draid_003_pos', 'zpool_create_draid_004_pos', 'zpool_create_features_001_pos', 'zpool_create_features_002_pos', 'zpool_create_features_003_pos', 'zpool_create_features_004_neg', 'zpool_create_features_005_pos', 'zpool_create_features_006_pos', 'zpool_create_features_007_pos', 'zpool_create_features_008_pos', 'zpool_create_features_009_pos', 'create-o_ashift', 'zpool_create_tempname', 'zpool_create_dryrun_output'] tags = ['functional', 'cli_root', 'zpool_create'] [tests/functional/cli_root/zpool_destroy] tests = ['zpool_destroy_001_pos', 'zpool_destroy_002_pos', 'zpool_destroy_003_neg'] pre = post = tags = ['functional', 'cli_root', 'zpool_destroy'] [tests/functional/cli_root/zpool_detach] tests = ['zpool_detach_001_neg'] tags = ['functional', 'cli_root', 'zpool_detach'] [tests/functional/cli_root/zpool_events] tests = ['zpool_events_clear', 'zpool_events_cliargs', 'zpool_events_follow', 'zpool_events_poolname', 'zpool_events_errors', 'zpool_events_duplicates', 'zpool_events_clear_retained'] tags = ['functional', 'cli_root', 'zpool_events'] [tests/functional/cli_root/zpool_export] tests = ['zpool_export_001_pos', 'zpool_export_002_pos', 'zpool_export_003_neg', 'zpool_export_004_pos', 'zpool_export_parallel_pos', 'zpool_export_parallel_admin'] tags = ['functional', 'cli_root', 'zpool_export'] [tests/functional/cli_root/zpool_get] tests = ['zpool_get_001_pos', 'zpool_get_002_pos', 'zpool_get_003_pos', 'zpool_get_004_neg', 'zpool_get_005_pos', 'vdev_get_001_pos', 'vdev_get_all'] tags = ['functional', 'cli_root', 'zpool_get'] [tests/functional/cli_root/zpool_history] tests = ['zpool_history_001_neg', 'zpool_history_002_pos'] tags = ['functional', 'cli_root', 'zpool_history'] [tests/functional/cli_root/zpool_import] tests = ['zpool_import_001_pos', 'zpool_import_002_pos', 'zpool_import_003_pos', 'zpool_import_004_pos', 'zpool_import_005_pos', 'zpool_import_006_pos', 'zpool_import_007_pos', 'zpool_import_008_pos', 'zpool_import_009_neg', 'zpool_import_010_pos', 'zpool_import_011_neg', 'zpool_import_012_pos', 'zpool_import_013_neg', 'zpool_import_014_pos', 'zpool_import_015_pos', 'zpool_import_016_pos', 'zpool_import_017_pos', 'zpool_import_features_001_pos', 'zpool_import_features_002_neg', 'zpool_import_features_003_pos', 'zpool_import_missing_001_pos', 'zpool_import_missing_002_pos', 'zpool_import_missing_003_pos', 'zpool_import_rename_001_pos', 'zpool_import_all_001_pos', 'zpool_import_encrypted', 'zpool_import_encrypted_load', 'zpool_import_errata3', 'zpool_import_errata4', 'import_cachefile_device_added', 'import_cachefile_device_removed', 'import_cachefile_device_replaced', 'import_cachefile_mirror_attached', 'import_cachefile_mirror_detached', 'import_cachefile_paths_changed', 'import_cachefile_shared_device', 'import_devices_missing', 'import_log_missing', 'import_paths_changed', 'import_rewind_config_changed', 'import_rewind_device_replaced', 'zpool_import_status', 'zpool_import_parallel_pos', 'zpool_import_parallel_neg', 'zpool_import_parallel_admin'] tags = ['functional', 'cli_root', 'zpool_import'] timeout = 1200 [tests/functional/cli_root/zpool_labelclear] tests = ['zpool_labelclear_active', 'zpool_labelclear_exported', 'zpool_labelclear_removed', 'zpool_labelclear_valid'] pre = post = tags = ['functional', 'cli_root', 'zpool_labelclear'] [tests/functional/cli_root/zpool_initialize] tests = ['zpool_initialize_attach_detach_add_remove', 'zpool_initialize_fault_export_import_online', 'zpool_initialize_import_export', 'zpool_initialize_multiple_pools', 'zpool_initialize_offline_export_import_online', 'zpool_initialize_online_offline', 'zpool_initialize_split', 'zpool_initialize_start_and_cancel_neg', 'zpool_initialize_start_and_cancel_pos', 'zpool_initialize_suspend_resume', 'zpool_initialize_uninit', 'zpool_initialize_unsupported_vdevs', 'zpool_initialize_verify_checksums', 'zpool_initialize_verify_initialized'] pre = tags = ['functional', 'cli_root', 'zpool_initialize'] [tests/functional/cli_root/zpool_offline] tests = ['zpool_offline_001_pos', 'zpool_offline_002_neg', 'zpool_offline_003_pos'] tags = ['functional', 'cli_root', 'zpool_offline'] [tests/functional/cli_root/zpool_online] tests = ['zpool_online_001_pos', 'zpool_online_002_neg'] tags = ['functional', 'cli_root', 'zpool_online'] [tests/functional/cli_root/zpool_reguid] tests = ['zpool_reguid_001_pos', 'zpool_reguid_002_neg'] tags = ['functional', 'cli_root', 'zpool_reguid'] [tests/functional/cli_root/zpool_remove] tests = ['zpool_remove_001_neg', 'zpool_remove_002_pos', 'zpool_remove_003_pos'] tags = ['functional', 'cli_root', 'zpool_remove'] [tests/functional/cli_root/zpool_replace] tests = ['zpool_replace_001_neg', 'replace-o_ashift', 'replace_prop_ashift'] tags = ['functional', 'cli_root', 'zpool_replace'] [tests/functional/cli_root/zpool_resilver] tests = ['zpool_resilver_bad_args', 'zpool_resilver_restart', 'zpool_resilver_concurrent'] tags = ['functional', 'cli_root', 'zpool_resilver'] [tests/functional/cli_root/zpool_scrub] tests = ['zpool_scrub_001_neg', 'zpool_scrub_002_pos', 'zpool_scrub_003_pos', 'zpool_scrub_004_pos', 'zpool_scrub_005_pos', 'zpool_scrub_encrypted_unloaded', 'zpool_scrub_print_repairing', 'zpool_scrub_offline_device', 'zpool_scrub_multiple_copies', 'zpool_scrub_multiple_pools', 'zpool_error_scrub_001_pos', 'zpool_error_scrub_002_pos', 'zpool_error_scrub_003_pos', 'zpool_error_scrub_004_pos', 'zpool_scrub_date_range_001'] tags = ['functional', 'cli_root', 'zpool_scrub'] [tests/functional/cli_root/zpool_set] tests = ['zpool_set_001_pos', 'zpool_set_002_neg', 'zpool_set_003_neg', 'zpool_set_ashift', 'zpool_set_features', 'vdev_set_001_pos', 'user_property_001_pos', 'user_property_002_neg', 'zpool_set_clear_userprop'] tags = ['functional', 'cli_root', 'zpool_set'] [tests/functional/cli_root/zpool_split] tests = ['zpool_split_cliargs', 'zpool_split_devices', 'zpool_split_encryption', 'zpool_split_props', 'zpool_split_vdevs', 'zpool_split_resilver', 'zpool_split_indirect', 'zpool_split_dryrun_output'] tags = ['functional', 'cli_root', 'zpool_split'] [tests/functional/cli_root/zpool_status] tests = ['zpool_status_001_pos', 'zpool_status_002_pos', 'zpool_status_003_pos', 'zpool_status_004_pos', 'zpool_status_005_pos', 'zpool_status_006_pos', 'zpool_status_007_pos', 'zpool_status_008_pos', 'zpool_status_features_001_pos'] tags = ['functional', 'cli_root', 'zpool_status'] [tests/functional/cli_root/zpool_sync] tests = ['zpool_sync_001_pos', 'zpool_sync_002_neg'] tags = ['functional', 'cli_root', 'zpool_sync'] [tests/functional/cli_root/zpool_trim] tests = ['zpool_trim_attach_detach_add_remove', 'zpool_trim_fault_export_import_online', 'zpool_trim_import_export', 'zpool_trim_multiple', 'zpool_trim_multiple_pools', 'zpool_trim_neg', 'zpool_trim_offline_export_import_online', 'zpool_trim_online_offline', 'zpool_trim_partial', 'zpool_trim_rate', 'zpool_trim_rate_neg', 'zpool_trim_secure', 'zpool_trim_split', 'zpool_trim_start_and_cancel_neg', 'zpool_trim_start_and_cancel_pos', 'zpool_trim_suspend_resume', 'zpool_trim_unsupported_vdevs', 'zpool_trim_verify_checksums', 'zpool_trim_verify_trimmed'] tags = ['functional', 'zpool_trim'] [tests/functional/cli_root/zpool_upgrade] tests = ['zpool_upgrade_001_pos', 'zpool_upgrade_002_pos', 'zpool_upgrade_003_pos', 'zpool_upgrade_004_pos', 'zpool_upgrade_005_neg', 'zpool_upgrade_006_neg', 'zpool_upgrade_007_pos', 'zpool_upgrade_008_pos', 'zpool_upgrade_009_neg', 'zpool_upgrade_features_001_pos'] tags = ['functional', 'cli_root', 'zpool_upgrade'] [tests/functional/cli_root/zpool_wait] tests = ['zpool_wait_discard', 'zpool_wait_freeing', 'zpool_wait_initialize_basic', 'zpool_wait_initialize_cancel', 'zpool_wait_initialize_flag', 'zpool_wait_multiple', 'zpool_wait_no_activity', 'zpool_wait_remove', 'zpool_wait_remove_cancel', 'zpool_wait_trim_basic', 'zpool_wait_trim_cancel', 'zpool_wait_trim_flag', 'zpool_wait_usage'] tags = ['functional', 'cli_root', 'zpool_wait'] [tests/functional/cli_root/zpool_wait/scan] tests = ['zpool_wait_replace_cancel', 'zpool_wait_rebuild', 'zpool_wait_resilver', 'zpool_wait_scrub_cancel', 'zpool_wait_replace', 'zpool_wait_scrub_basic', 'zpool_wait_scrub_flag'] tags = ['functional', 'cli_root', 'zpool_wait'] [tests/functional/cli_user/misc] tests = ['zdb_001_neg', 'zfs_001_neg', 'zfs_allow_001_neg', 'zfs_clone_001_neg', 'zfs_create_001_neg', 'zfs_destroy_001_neg', 'zfs_get_001_neg', 'zfs_inherit_001_neg', 'zfs_mount_001_neg', 'zfs_promote_001_neg', 'zfs_receive_001_neg', 'zfs_rename_001_neg', 'zfs_rollback_001_neg', 'zfs_send_001_neg', 'zfs_set_001_neg', 'zfs_share_001_neg', 'zfs_snapshot_001_neg', 'zfs_unallow_001_neg', 'zfs_unmount_001_neg', 'zfs_unshare_001_neg', 'zfs_upgrade_001_neg', 'zpool_001_neg', 'zpool_add_001_neg', 'zpool_attach_001_neg', 'zpool_clear_001_neg', 'zpool_create_001_neg', 'zpool_destroy_001_neg', 'zpool_detach_001_neg', 'zpool_export_001_neg', 'zpool_get_001_neg', 'zpool_history_001_neg', 'zpool_import_001_neg', 'zpool_import_002_neg', 'zpool_offline_001_neg', 'zpool_online_001_neg', 'zpool_remove_001_neg', 'zpool_replace_001_neg', 'zpool_scrub_001_neg', 'zpool_set_001_neg', 'zpool_status_001_neg', 'zpool_upgrade_001_neg', 'arcstat_001_pos', 'arc_summary_001_pos', 'arc_summary_002_neg', 'zpool_wait_privilege', 'zilstat_001_pos'] user = tags = ['functional', 'cli_user', 'misc'] [tests/functional/cli_user/zfs_list] tests = ['zfs_list_001_pos', 'zfs_list_002_pos', 'zfs_list_003_pos', 'zfs_list_004_neg', 'zfs_list_005_neg', 'zfs_list_007_pos', 'zfs_list_008_neg'] user = tags = ['functional', 'cli_user', 'zfs_list'] [tests/functional/cli_user/zpool_iostat] tests = ['zpool_iostat_001_neg', 'zpool_iostat_002_pos', 'zpool_iostat_003_neg', 'zpool_iostat_004_pos', 'zpool_iostat_005_pos', 'zpool_iostat_-c_disable', 'zpool_iostat_-c_homedir', 'zpool_iostat_-c_searchpath'] user = tags = ['functional', 'cli_user', 'zpool_iostat'] [tests/functional/cli_user/zpool_list] tests = ['zpool_list_001_pos', 'zpool_list_002_neg'] user = tags = ['functional', 'cli_user', 'zpool_list'] [tests/functional/cli_user/zpool_status] tests = ['zpool_status_003_pos', 'zpool_status_-c_disable', 'zpool_status_-c_homedir', 'zpool_status_-c_searchpath'] user = tags = ['functional', 'cli_user', 'zpool_status'] [tests/functional/compression] tests = ['compress_001_pos', 'compress_002_pos', 'compress_003_pos', 'l2arc_compressed_arc', 'l2arc_compressed_arc_disabled', 'l2arc_encrypted', 'l2arc_encrypted_no_compressed_arc'] tags = ['functional', 'compression'] [tests/functional/cp_files] tests = ['cp_files_001_pos', 'cp_files_002_pos', 'cp_stress'] tags = ['functional', 'cp_files'] [tests/functional/zap_shrink] tests = ['zap_shrink_001_pos'] tags = ['functional', 'zap_shrink'] [tests/functional/crtime] tests = ['crtime_001_pos' ] tags = ['functional', 'crtime'] [tests/functional/crypto] tests = ['icp_aes_ccm', 'icp_aes_gcm'] pre = post = tags = ['functional', 'crypto'] [tests/functional/ctime] tests = ['ctime_001_pos' ] tags = ['functional', 'ctime'] [tests/functional/deadman] tests = ['deadman_ratelimit', 'deadman_sync', 'deadman_zio'] pre = post = tags = ['functional', 'deadman'] [tests/functional/dedup] tests = ['dedup_fdt_create', 'dedup_fdt_import', 'dedup_fdt_pacing', 'dedup_legacy_create', 'dedup_legacy_import', 'dedup_legacy_fdt_upgrade', 'dedup_legacy_fdt_mixed', 'dedup_quota', 'dedup_prune', 'dedup_zap_shrink'] pre = post = tags = ['functional', 'dedup'] [tests/functional/delegate] tests = ['zfs_allow_001_pos', 'zfs_allow_002_pos', 'zfs_allow_003_pos', 'zfs_allow_004_pos', 'zfs_allow_005_pos', 'zfs_allow_006_pos', 'zfs_allow_007_pos', 'zfs_allow_008_pos', 'zfs_allow_009_neg', 'zfs_allow_010_pos', 'zfs_allow_011_neg', 'zfs_allow_012_neg', 'zfs_unallow_001_pos', 'zfs_unallow_002_pos', 'zfs_unallow_003_pos', 'zfs_unallow_004_pos', 'zfs_unallow_005_pos', 'zfs_unallow_006_pos', 'zfs_unallow_007_neg', 'zfs_unallow_008_neg'] tags = ['functional', 'delegate'] [tests/functional/direct] tests = ['dio_aligned_block', 'dio_async_always', 'dio_async_fio_ioengines', 'dio_compression', 'dio_dedup', 'dio_encryption', 'dio_grow_block', 'dio_max_recordsize', 'dio_mixed', 'dio_mmap', 'dio_overwrites', 'dio_property', 'dio_random', 'dio_read_verify', 'dio_recordsize', 'dio_unaligned_block', 'dio_unaligned_filesize'] tags = ['functional', 'direct'] [tests/functional/exec] tests = ['exec_001_pos', 'exec_002_neg'] tags = ['functional', 'exec'] [tests/functional/fadvise] tests = ['fadvise_willneed'] tags = ['functional', 'fadvise'] [tests/functional/failmode] tests = ['failmode_dmu_tx_wait', 'failmode_dmu_tx_continue'] tags = ['functional', 'failmode'] [tests/functional/fallocate] tests = ['fallocate_punch-hole'] tags = ['functional', 'fallocate'] [tests/functional/features/async_destroy] tests = ['async_destroy_001_pos'] tags = ['functional', 'features', 'async_destroy'] [tests/functional/features/large_dnode] tests = ['large_dnode_001_pos', 'large_dnode_003_pos', 'large_dnode_004_neg', 'large_dnode_005_pos', 'large_dnode_007_neg', 'large_dnode_009_pos'] tags = ['functional', 'features', 'large_dnode'] [tests/functional/gang_blocks] tests = ['gang_blocks_001_pos', 'gang_blocks_redundant', 'gang_blocks_ddt_copies', 'gang_blocks_dyn_header_pos', 'gang_blocks_dyn_header_neg', 'gang_blocks_dyn_multi'] tags = ['functional', 'gang_blocks'] [tests/functional/grow] pre = post = tests = ['grow_pool_001_pos', 'grow_replicas_001_pos'] tags = ['functional', 'grow'] [tests/functional/history] tests = ['history_001_pos', 'history_002_pos', 'history_003_pos', 'history_004_pos', 'history_005_neg', 'history_006_neg', 'history_007_pos', 'history_008_pos', 'history_009_pos', 'history_010_pos'] tags = ['functional', 'history'] [tests/functional/hkdf] pre = post = tests = ['hkdf_test'] tags = ['functional', 'hkdf'] [tests/functional/inheritance] tests = ['inherit_001_pos'] pre = tags = ['functional', 'inheritance'] [tests/functional/io] tests = ['mmap', 'posixaio', 'psync', 'sync'] tags = ['functional', 'io'] [tests/functional/inuse] tests = ['inuse_004_pos', 'inuse_005_pos', 'inuse_008_pos', 'inuse_009_pos'] post = tags = ['functional', 'inuse'] [tests/functional/large_files] tests = ['large_files_001_pos', 'large_files_002_pos'] tags = ['functional', 'large_files'] [tests/functional/limits] tests = ['filesystem_count', 'filesystem_limit', 'snapshot_count', 'snapshot_limit'] tags = ['functional', 'limits'] [tests/functional/link_count] tests = ['link_count_001', 'link_count_root_inode'] tags = ['functional', 'link_count'] [tests/functional/migration] tests = ['migration_001_pos', 'migration_002_pos', 'migration_003_pos', 'migration_004_pos', 'migration_005_pos', 'migration_006_pos', 'migration_007_pos', 'migration_008_pos', 'migration_009_pos', 'migration_010_pos', 'migration_011_pos', 'migration_012_pos'] tags = ['functional', 'migration'] [tests/functional/mmap] tests = ['mmap_mixed', 'mmap_read_001_pos', 'mmap_seek_001_pos', 'mmap_sync_001_pos', 'mmap_write_001_pos', 'mmap_ftruncate'] tags = ['functional', 'mmap'] [tests/functional/mount] tests = ['umount_001', 'umountall_001'] tags = ['functional', 'mount'] [tests/functional/mv_files] tests = ['mv_files_001_pos', 'mv_files_002_pos', 'random_creation'] tags = ['functional', 'mv_files'] [tests/functional/nestedfs] tests = ['nestedfs_001_pos'] tags = ['functional', 'nestedfs'] [tests/functional/no_space] tests = ['enospc_001_pos', 'enospc_002_pos', 'enospc_003_pos', 'enospc_df', 'enospc_ganging', 'enospc_rm'] tags = ['functional', 'no_space'] [tests/functional/nopwrite] tests = ['nopwrite_copies', 'nopwrite_mtime', 'nopwrite_negative', 'nopwrite_promoted_clone', 'nopwrite_recsize', 'nopwrite_sync', 'nopwrite_varying_compression', 'nopwrite_volume'] tags = ['functional', 'nopwrite'] [tests/functional/online_offline] tests = ['online_offline_001_pos', 'online_offline_002_neg', 'online_offline_003_neg'] tags = ['functional', 'online_offline'] [tests/functional/pool_checkpoint] tests = ['checkpoint_after_rewind', 'checkpoint_big_rewind', 'checkpoint_capacity', 'checkpoint_conf_change', 'checkpoint_discard', 'checkpoint_discard_busy', 'checkpoint_discard_many', 'checkpoint_indirect', 'checkpoint_invalid', 'checkpoint_lun_expsz', 'checkpoint_open', 'checkpoint_removal', 'checkpoint_rewind', 'checkpoint_ro_rewind', 'checkpoint_sm_scale', 'checkpoint_twice', 'checkpoint_vdev_add', 'checkpoint_zdb', 'checkpoint_zhack_feat'] tags = ['functional', 'pool_checkpoint'] timeout = 1800 [tests/functional/pool_names] tests = ['pool_names_001_pos', 'pool_names_002_neg'] pre = post = tags = ['functional', 'pool_names'] [tests/functional/projectquota] tests = ['defaultprojectquota_002_pos', 'defaultprojectquota_003_neg', 'defaultprojectquota_004_pos', 'defaultprojectquota_006_pos', 'defaultprojectquota_007_pos', 'projectquota_002_pos', 'projectquota_004_neg', 'projectquota_005_pos', 'projectquota_007_pos', 'projectquota_008_pos', 'projectquota_009_pos', 'projecttree_002_pos', 'projecttree_003_neg'] tags = ['functional', 'projectquota'] [tests/functional/poolversion] tests = ['poolversion_001_pos', 'poolversion_002_pos'] tags = ['functional', 'poolversion'] [tests/functional/pyzfs] tests = ['pyzfs_unittest'] pre = post = tags = ['functional', 'pyzfs'] [tests/functional/quota] tests = ['quota_001_pos', 'quota_002_pos', 'quota_003_pos', 'quota_004_pos', 'quota_005_pos', 'quota_006_neg'] tags = ['functional', 'quota'] [tests/functional/redacted_send] tests = ['redacted_compressed', 'redacted_contents', 'redacted_deleted', 'redacted_disabled_feature', 'redacted_embedded', 'redacted_holes', 'redacted_incrementals', 'redacted_largeblocks', 'redacted_many_clones', 'redacted_mixed_recsize', 'redacted_mounts', 'redacted_negative', 'redacted_origin', 'redacted_panic', 'redacted_props', 'redacted_resume', 'redacted_size', 'redacted_volume'] tags = ['functional', 'redacted_send'] [tests/functional/raidz] tests = ['raidz_001_neg', 'raidz_002_pos', 'raidz_expand_001_pos', 'raidz_expand_002_pos', 'raidz_expand_003_neg', 'raidz_expand_003_pos', 'raidz_expand_004_pos', 'raidz_expand_005_pos', 'raidz_expand_006_neg', 'raidz_expand_007_neg'] tags = ['functional', 'raidz'] timeout = 1200 [tests/functional/redundancy] tests = ['redundancy_draid', 'redundancy_draid1', 'redundancy_draid2', 'redundancy_draid3', 'redundancy_draid_damaged1', 'redundancy_draid_damaged2', 'redundancy_draid_spare1', 'redundancy_draid_spare2', 'redundancy_draid_spare3', 'redundancy_mirror', 'redundancy_raidz', 'redundancy_raidz1', 'redundancy_raidz2', 'redundancy_raidz3', 'redundancy_stripe'] tags = ['functional', 'redundancy'] timeout = 1200 [tests/functional/refquota] tests = ['refquota_001_pos', 'refquota_002_pos', 'refquota_003_pos', 'refquota_004_pos', 'refquota_005_pos', 'refquota_006_neg', 'refquota_007_neg', 'refquota_008_neg'] tags = ['functional', 'refquota'] [tests/functional/refreserv] tests = ['refreserv_001_pos', 'refreserv_002_pos', 'refreserv_003_pos', 'refreserv_004_pos', 'refreserv_005_pos', 'refreserv_multi_raidz', 'refreserv_raidz'] tags = ['functional', 'refreserv'] [tests/functional/removal] pre = tests = ['removal_all_vdev', 'removal_cancel', 'removal_check_space', 'removal_condense_export', 'removal_multiple_indirection', 'removal_nopwrite', 'removal_remap_deadlists', 'removal_resume_export', 'removal_sanity', 'removal_with_add', 'removal_with_create_fs', 'removal_with_dedup', 'removal_with_errors', 'removal_with_export', 'removal_with_indirect', 'removal_with_ganging', 'removal_with_faulted', 'removal_with_remove', 'removal_with_scrub', 'removal_with_send', 'removal_with_send_recv', 'removal_with_snapshot', 'removal_with_write', 'removal_with_zdb', 'remove_expanded', 'remove_mirror', 'remove_mirror_sanity', 'remove_raidz', 'remove_indirect', 'remove_attach_mirror', 'removal_reservation', 'removal_with_hole'] tags = ['functional', 'removal'] [tests/functional/rename_dirs] tests = ['rename_dirs_001_pos'] tags = ['functional', 'rename_dirs'] [tests/functional/replacement] tests = ['attach_import', 'attach_multiple', 'attach_rebuild', 'attach_resilver', 'detach', 'rebuild_disabled_feature', 'rebuild_multiple', 'rebuild_raidz', 'replace_import', 'replace_rebuild', 'replace_resilver', 'resilver_restart_001', 'resilver_restart_002', 'scrub_cancel'] tags = ['functional', 'replacement'] [tests/functional/reservation] tests = ['reservation_001_pos', 'reservation_002_pos', 'reservation_003_pos', 'reservation_004_pos', 'reservation_005_pos', 'reservation_006_pos', 'reservation_007_pos', 'reservation_008_pos', 'reservation_009_pos', 'reservation_010_pos', 'reservation_011_pos', 'reservation_012_pos', 'reservation_013_pos', 'reservation_014_pos', 'reservation_015_pos', 'reservation_016_pos', 'reservation_017_pos', 'reservation_018_pos', 'reservation_019_pos', 'reservation_020_pos', 'reservation_021_neg', 'reservation_022_pos'] tags = ['functional', 'reservation'] [tests/functional/rootpool] tests = ['rootpool_002_neg', 'rootpool_003_neg', 'rootpool_007_pos'] tags = ['functional', 'rootpool'] [tests/functional/rsend] tests = ['recv_dedup', 'recv_dedup_encrypted_zvol', 'rsend_001_pos', 'rsend_002_pos', 'rsend_003_pos', 'rsend_004_pos', 'rsend_005_pos', 'rsend_006_pos', 'rsend_007_pos', 'rsend_008_pos', 'rsend_009_pos', 'rsend_010_pos', 'rsend_011_pos', 'rsend_012_pos', 'rsend_013_pos', 'rsend_014_pos', 'rsend_016_neg', 'rsend_019_pos', 'rsend_020_pos', 'rsend_021_pos', 'rsend_022_pos', 'rsend_024_pos', 'rsend_025_pos', 'rsend_026_neg', 'rsend_027_pos', 'rsend_028_neg', 'rsend_029_neg', 'rsend_030_pos', 'rsend_031_pos', 'send-c_verify_ratio', 'send-c_verify_contents', 'send-c_props', 'send-c_incremental', 'send-c_volume', 'send-c_zstream_recompress', 'send-c_zstreamdump', 'send-c_lz4_disabled', 'send-c_recv_lz4_disabled', 'send-c_mixed_compression', 'send-c_stream_size_estimate', 'send-c_embedded_blocks', 'send-c_resume', 'send-cpL_varied_recsize', 'send-c_recv_dedup', 'send-L_toggle', 'send_encrypted_incremental', 'send_encrypted_freeobjects', 'send_encrypted_hierarchy', 'send_encrypted_props', 'send_encrypted_truncated_files', 'send_freeobjects', 'send_realloc_files', 'send_realloc_encrypted_files', 'send_spill_block', 'send_holds', 'send_hole_birth', 'send_mixed_raw', 'send-wR_encrypted_zvol', 'send_partial_dataset', 'send_invalid', 'send_doall', 'send_raw_spill_block', 'send_raw_ashift', 'send_raw_large_blocks', 'send_leak_keymaps'] tags = ['functional', 'rsend'] [tests/functional/scrub_mirror] tests = ['scrub_mirror_001_pos', 'scrub_mirror_002_pos', 'scrub_mirror_003_pos', 'scrub_mirror_004_pos'] tags = ['functional', 'scrub_mirror'] [tests/functional/slog] tests = ['slog_001_pos', 'slog_002_pos', 'slog_003_pos', 'slog_004_pos', 'slog_005_pos', 'slog_006_pos', 'slog_007_pos', 'slog_008_neg', 'slog_009_neg', 'slog_010_neg', 'slog_011_neg', 'slog_012_neg', 'slog_013_pos', 'slog_014_pos', 'slog_015_neg', 'slog_replay_fs_001', 'slog_replay_fs_002', 'slog_replay_volume', 'slog_016_pos'] tags = ['functional', 'slog'] [tests/functional/snapshot] tests = ['clone_001_pos', 'rollback_001_pos', 'rollback_002_pos', 'rollback_003_pos', 'snapshot_001_pos', 'snapshot_002_pos', 'snapshot_003_pos', 'snapshot_004_pos', 'snapshot_005_pos', 'snapshot_006_pos', 'snapshot_007_pos', 'snapshot_008_pos', 'snapshot_009_pos', 'snapshot_010_pos', 'snapshot_011_pos', 'snapshot_012_pos', 'snapshot_013_pos', 'snapshot_014_pos', 'snapshot_017_pos', 'snapshot_018_pos'] tags = ['functional', 'snapshot'] [tests/functional/snapused] tests = ['snapused_001_pos', 'snapused_002_pos', 'snapused_003_pos', 'snapused_004_pos', 'snapused_005_pos'] tags = ['functional', 'snapused'] [tests/functional/sparse] tests = ['sparse_001_pos'] tags = ['functional', 'sparse'] [tests/functional/stat] tests = ['stat_001_pos', 'statx_dioalign'] tags = ['functional', 'stat'] [tests/functional/suid] tests = ['suid_write_to_suid', 'suid_write_to_sgid', 'suid_write_to_suid_sgid', 'suid_write_to_none', 'suid_write_zil_replay'] tags = ['functional', 'suid'] [tests/functional/trim] tests = ['autotrim_integrity', 'autotrim_config', 'autotrim_trim_integrity', 'trim_integrity', 'trim_config', 'trim_l2arc'] tags = ['functional', 'trim'] [tests/functional/truncate] tests = ['truncate_001_pos', 'truncate_002_pos', 'truncate_timestamps'] tags = ['functional', 'truncate'] [tests/functional/upgrade] tests = ['upgrade_userobj_001_pos', 'upgrade_readonly_pool'] tags = ['functional', 'upgrade'] [tests/functional/userquota] tests = [ 'defaultuserquota_001_pos', 'defaultuserquota_002_pos', 'defaultuserquota_003_pos', 'defaultuserquota_004_neg', 'defaultuserquota_005_pos', 'defaultuserquota_006_pos', 'defaultuserquota_007_pos', 'defaultuserquota_008_pos', 'defaultuserquota_009_pos', 'defaultuserquota_010_neg', 'defaultuserquota_011_neg', 'defaultuserquota_012_neg', 'defaultuserquota_013_neg', 'userquota_001_pos', 'userquota_002_pos', 'userquota_003_pos', 'userquota_004_pos', 'userquota_005_neg', 'userquota_006_pos', 'userquota_007_pos', 'userquota_008_pos', 'userquota_009_pos', 'userquota_010_pos', 'userquota_011_pos', 'userquota_012_neg', 'userspace_001_pos', 'userspace_002_pos', 'userspace_004_pos', 'userspace_encrypted', 'userspace_send_encrypted', 'userspace_encrypted_13709'] tags = ['functional', 'userquota'] [tests/functional/vdev_disk:Linux] pre = post = tests = ['page_alignment'] tags = ['functional', 'vdev_disk'] [tests/functional/vdev_zaps] tests = ['vdev_zaps_001_pos', 'vdev_zaps_002_pos', 'vdev_zaps_003_pos', 'vdev_zaps_004_pos', 'vdev_zaps_005_pos', 'vdev_zaps_006_pos', 'vdev_zaps_007_pos'] tags = ['functional', 'vdev_zaps'] [tests/functional/write_dirs] tests = ['write_dirs_001_pos', 'write_dirs_002_pos'] tags = ['functional', 'write_dirs'] [tests/functional/xattr] tests = ['xattr_001_pos', 'xattr_002_neg', 'xattr_003_neg', 'xattr_004_pos', 'xattr_005_pos', 'xattr_006_pos', 'xattr_007_neg', 'xattr_011_pos', 'xattr_012_pos', 'xattr_013_pos', 'xattr_compat'] tags = ['functional', 'xattr'] [tests/functional/zvol/zvol_ENOSPC] tests = ['zvol_ENOSPC_001_pos'] tags = ['functional', 'zvol', 'zvol_ENOSPC'] [tests/functional/zvol/zvol_cli] tests = ['zvol_cli_001_pos', 'zvol_cli_002_pos', 'zvol_cli_003_neg'] tags = ['functional', 'zvol', 'zvol_cli'] [tests/functional/zvol/zvol_misc] tests = ['zvol_misc_002_pos', 'zvol_misc_hierarchy', 'zvol_misc_rename_inuse', 'zvol_misc_snapdev', 'zvol_misc_trim', 'zvol_misc_volmode', 'zvol_misc_zil'] tags = ['functional', 'zvol', 'zvol_misc'] [tests/functional/zvol/zvol_stress] tests = ['zvol_stress'] tags = ['functional', 'zvol', 'zvol_stress'] [tests/functional/zvol/zvol_swap] tests = ['zvol_swap_001_pos', 'zvol_swap_002_pos', 'zvol_swap_004_pos'] tags = ['functional', 'zvol', 'zvol_swap'] [tests/functional/libzfs] tests = ['many_fds', 'libzfs_input'] tags = ['functional', 'libzfs'] [tests/functional/log_spacemap] tests = ['log_spacemap_import_logs'] pre = post = tags = ['functional', 'log_spacemap'] [tests/functional/l2arc] tests = ['l2arc_arcstats_pos', 'l2arc_mfuonly_pos', 'l2arc_l2miss_pos', 'persist_l2arc_001_pos', 'persist_l2arc_002_pos', 'persist_l2arc_003_neg', 'persist_l2arc_004_pos', 'persist_l2arc_005_pos'] tags = ['functional', 'l2arc'] [tests/functional/zpool_influxdb] tests = ['zpool_influxdb'] tags = ['functional', 'zpool_influxdb'] diff --git a/tests/runfiles/sanity.run b/tests/runfiles/sanity.run index 732f252b52d2..7767c0c2d535 100644 --- a/tests/runfiles/sanity.run +++ b/tests/runfiles/sanity.run @@ -1,648 +1,648 @@ # SPDX-License-Identifier: CDDL-1.0 # # This file and its contents are supplied under the terms of the # Common Development and Distribution License ("CDDL"), version 1.0. # You may only use this file in accordance with the terms of version # 1.0 of the CDDL. # # A full copy of the text of the CDDL should have accompanied this # source. A copy of the CDDL is also available via the Internet at # http://www.illumos.org/license/CDDL. # # This run file contains a subset of functional tests which exercise # as much functionality as possible while still executing relatively # quickly. The included tests should take no more than a few seconds # each to run at most. This provides a convenient way to sanity test a # change before committing to a full test run which takes several hours. # # Approximate run time: 15 minutes # [DEFAULT] pre = setup quiet = False pre_user = root user = root timeout = 180 post_user = root post = cleanup failsafe_user = root failsafe = callbacks/zfs_failsafe tags = ['functional'] [tests/functional/acl/off] tests = ['posixmode'] tags = ['functional', 'acl'] [tests/functional/alloc_class] tests = ['alloc_class_003_pos', 'alloc_class_004_pos', 'alloc_class_005_pos', 'alloc_class_006_pos', 'alloc_class_008_pos', 'alloc_class_010_pos', 'alloc_class_011_neg'] tags = ['functional', 'alloc_class'] [tests/functional/arc] tests = ['dbufstats_001_pos', 'dbufstats_002_pos', 'arcstats_runtime_tuning'] tags = ['functional', 'arc'] [tests/functional/bootfs] tests = ['bootfs_004_neg', 'bootfs_007_pos'] tags = ['functional', 'bootfs'] [tests/functional/cache] tests = ['cache_004_neg', 'cache_005_neg', 'cache_007_neg', 'cache_010_pos'] tags = ['functional', 'cache'] [tests/functional/cachefile] tests = ['cachefile_001_pos', 'cachefile_002_pos', 'cachefile_003_pos', 'cachefile_004_pos'] tags = ['functional', 'cachefile'] [tests/functional/casenorm] tests = ['case_all_values', 'norm_all_values', 'sensitive_none_lookup', 'sensitive_none_delete', 'insensitive_none_lookup', 'insensitive_none_delete', 'mixed_none_lookup', 'mixed_none_delete'] tags = ['functional', 'casenorm'] [tests/functional/channel_program/lua_core] tests = ['tst.args_to_lua', 'tst.divide_by_zero', 'tst.exists', 'tst.encryption', 'tst.integer_illegal', 'tst.integer_overflow', 'tst.language_functions_neg', 'tst.language_functions_pos', 'tst.large_prog', 'tst.libraries', 'tst.memory_limit', 'tst.nested_neg', 'tst.nested_pos', 'tst.nvlist_to_lua', 'tst.recursive_neg', 'tst.recursive_pos', 'tst.return_large', 'tst.return_nvlist_neg', 'tst.return_nvlist_pos', 'tst.return_recursive_table', 'tst.stack_gsub', 'tst.timeout'] tags = ['functional', 'channel_program', 'lua_core'] [tests/functional/channel_program/synctask_core] tests = ['tst.destroy_fs', 'tst.destroy_snap', 'tst.get_count_and_limit', 'tst.get_index_props', 'tst.get_mountpoint', 'tst.get_neg', 'tst.get_number_props', 'tst.get_string_props', 'tst.get_type', 'tst.get_userquota', 'tst.get_written', 'tst.inherit', 'tst.list_bookmarks', 'tst.list_children', 'tst.list_clones', 'tst.list_holds', 'tst.list_snapshots', 'tst.list_system_props', 'tst.list_user_props', 'tst.parse_args_neg','tst.promote_conflict', 'tst.promote_multiple', 'tst.promote_simple', 'tst.rollback_mult', 'tst.rollback_one', 'tst.set_props', 'tst.snapshot_destroy', 'tst.snapshot_neg', 'tst.snapshot_recursive', 'tst.snapshot_simple', 'tst.bookmark.create', 'tst.bookmark.copy', 'tst.clone'] tags = ['functional', 'channel_program', 'synctask_core'] [tests/functional/cli_root/zdb] tests = ['zdb_003_pos', 'zdb_004_pos', 'zdb_005_pos'] pre = post = tags = ['functional', 'cli_root', 'zdb'] [tests/functional/cli_root/zfs] tests = ['zfs_001_neg', 'zfs_002_pos'] tags = ['functional', 'cli_root', 'zfs'] [tests/functional/cli_root/zfs_bookmark] tests = ['zfs_bookmark_cliargs'] tags = ['functional', 'cli_root', 'zfs_bookmark'] [tests/functional/cli_root/zfs_change-key] tests = ['zfs_change-key', 'zfs_change-key_child', 'zfs_change-key_format', 'zfs_change-key_inherit', 'zfs_change-key_load', 'zfs_change-key_location', 'zfs_change-key_pbkdf2iters', 'zfs_change-key_clones'] tags = ['functional', 'cli_root', 'zfs_change-key'] [tests/functional/cli_root/zfs_clone] tests = ['zfs_clone_001_neg', 'zfs_clone_002_pos', 'zfs_clone_003_pos', 'zfs_clone_004_pos', 'zfs_clone_005_pos', 'zfs_clone_006_pos', 'zfs_clone_007_pos', 'zfs_clone_008_neg', 'zfs_clone_009_neg', 'zfs_clone_encrypted'] tags = ['functional', 'cli_root', 'zfs_clone'] [tests/functional/cli_root/zfs_create] tests = ['zfs_create_001_pos', 'zfs_create_002_pos', 'zfs_create_003_pos', 'zfs_create_004_pos', 'zfs_create_005_pos', 'zfs_create_006_pos', 'zfs_create_007_pos', 'zfs_create_011_pos', 'zfs_create_012_pos', 'zfs_create_013_pos', 'zfs_create_014_pos', 'zfs_create_encrypted', 'zfs_create_dryrun', 'zfs_create_verbose'] tags = ['functional', 'cli_root', 'zfs_create'] [tests/functional/cli_root/zfs_destroy] tests = ['zfs_destroy_002_pos', 'zfs_destroy_003_pos', 'zfs_destroy_004_pos', 'zfs_destroy_006_neg', 'zfs_destroy_007_neg', 'zfs_destroy_008_pos', 'zfs_destroy_009_pos', 'zfs_destroy_010_pos', 'zfs_destroy_011_pos', 'zfs_destroy_012_pos', 'zfs_destroy_013_neg', 'zfs_destroy_014_pos', 'zfs_destroy_dev_removal', 'zfs_destroy_dev_removal_condense'] tags = ['functional', 'cli_root', 'zfs_destroy'] [tests/functional/cli_root/zfs_diff] tests = ['zfs_diff_cliargs', 'zfs_diff_encrypted'] tags = ['functional', 'cli_root', 'zfs_diff'] [tests/functional/cli_root/zfs_get] tests = ['zfs_get_003_pos', 'zfs_get_006_neg', 'zfs_get_007_neg', 'zfs_get_010_neg'] tags = ['functional', 'cli_root', 'zfs_get'] [tests/functional/cli_root/zfs_inherit] tests = ['zfs_inherit_001_neg', 'zfs_inherit_003_pos', 'zfs_inherit_mountpoint'] tags = ['functional', 'cli_root', 'zfs_inherit'] [tests/functional/cli_root/zfs_load-key] tests = ['zfs_load-key', 'zfs_load-key_all', 'zfs_load-key_file', 'zfs_load-key_https', 'zfs_load-key_location', 'zfs_load-key_noop', 'zfs_load-key_recursive'] tags = ['functional', 'cli_root', 'zfs_load-key'] [tests/functional/cli_root/zfs_mount] tests = ['zfs_mount_001_pos', 'zfs_mount_002_pos', 'zfs_mount_003_pos', 'zfs_mount_004_pos', 'zfs_mount_005_pos', 'zfs_mount_007_pos', 'zfs_mount_009_neg', 'zfs_mount_010_neg', 'zfs_mount_011_neg', 'zfs_mount_012_pos', 'zfs_mount_encrypted', 'zfs_mount_remount', 'zfs_mount_all_fail', 'zfs_mount_all_mountpoints', 'zfs_mount_test_race', 'zfs_mount_recursive'] tags = ['functional', 'cli_root', 'zfs_mount'] [tests/functional/cli_root/zfs_program] tests = ['zfs_program_json'] tags = ['functional', 'cli_root', 'zfs_program'] [tests/functional/cli_root/zfs_promote] tests = ['zfs_promote_001_pos', 'zfs_promote_002_pos', 'zfs_promote_003_pos', 'zfs_promote_004_pos', 'zfs_promote_005_pos', 'zfs_promote_006_neg', 'zfs_promote_007_neg', 'zfs_promote_008_pos', 'zfs_promote_encryptionroot'] tags = ['functional', 'cli_root', 'zfs_promote'] [tests/functional/cli_root/zfs_receive] tests = ['zfs_receive_001_pos', 'zfs_receive_002_pos', 'zfs_receive_003_pos', 'zfs_receive_004_neg', 'zfs_receive_005_neg', 'zfs_receive_006_pos', 'zfs_receive_007_neg', 'zfs_receive_008_pos', 'zfs_receive_009_neg', 'zfs_receive_010_pos', 'zfs_receive_011_pos', 'zfs_receive_012_pos', 'zfs_receive_013_pos', 'zfs_receive_014_pos', 'zfs_receive_015_pos', 'zfs_receive_016_pos', 'zfs_receive_from_encrypted', 'zfs_receive_to_encrypted', 'zfs_receive_raw', 'zfs_receive_raw_incremental', 'zfs_receive_-e', 'zfs_receive_raw_-d', 'zfs_receive_from_zstd', 'zfs_receive_new_props'] tags = ['functional', 'cli_root', 'zfs_receive'] [tests/functional/cli_root/zfs_rename] tests = ['zfs_rename_003_pos', 'zfs_rename_004_neg', 'zfs_rename_005_neg', 'zfs_rename_006_pos', 'zfs_rename_007_pos', 'zfs_rename_008_pos', 'zfs_rename_009_neg', 'zfs_rename_010_neg', 'zfs_rename_011_pos', 'zfs_rename_012_neg', 'zfs_rename_013_pos', 'zfs_rename_encrypted_child', 'zfs_rename_to_encrypted', 'zfs_rename_mountpoint', 'zfs_rename_nounmount'] tags = ['functional', 'cli_root', 'zfs_rename'] [tests/functional/cli_root/zfs_reservation] tests = ['zfs_reservation_001_pos', 'zfs_reservation_002_pos'] tags = ['functional', 'cli_root', 'zfs_reservation'] [tests/functional/cli_root/zfs_rewrite] -tests = ['zfs_rewrite'] +tests = ['zfs_rewrite', 'zfs_rewrite_physical'] tags = ['functional', 'cli_root', 'zfs_rewrite'] [tests/functional/cli_root/zfs_rollback] tests = ['zfs_rollback_003_neg', 'zfs_rollback_004_neg'] tags = ['functional', 'cli_root', 'zfs_rollback'] [tests/functional/cli_root/zfs_send] tests = ['zfs_send_001_pos', 'zfs_send_002_pos', 'zfs_send_003_pos', 'zfs_send_004_neg', 'zfs_send_005_pos', 'zfs_send_encrypted', 'zfs_send_raw'] tags = ['functional', 'cli_root', 'zfs_send'] [tests/functional/cli_root/zfs_set] tests = ['cache_001_pos', 'cache_002_neg', 'canmount_001_pos', 'canmount_002_pos', 'canmount_003_pos', 'canmount_004_pos', 'checksum_001_pos', 'compression_001_pos', 'mountpoint_001_pos', 'mountpoint_002_pos', 'user_property_002_pos', 'share_mount_001_neg', 'snapdir_001_pos', 'onoffs_001_pos', 'user_property_001_pos', 'user_property_003_neg', 'readonly_001_pos', 'user_property_004_pos', 'version_001_neg', 'zfs_set_003_neg', 'property_alias_001_pos', 'zfs_set_keylocation', 'zfs_set_feature_activation', 'zfs_set_nomount'] tags = ['functional', 'cli_root', 'zfs_set'] [tests/functional/cli_root/zfs_snapshot] tests = ['zfs_snapshot_001_neg', 'zfs_snapshot_002_neg', 'zfs_snapshot_003_neg', 'zfs_snapshot_006_pos', 'zfs_snapshot_007_neg'] tags = ['functional', 'cli_root', 'zfs_snapshot'] [tests/functional/cli_root/zfs_unload-key] tests = ['zfs_unload-key', 'zfs_unload-key_all', 'zfs_unload-key_recursive'] tags = ['functional', 'cli_root', 'zfs_unload-key'] [tests/functional/cli_root/zfs_unmount] tests = ['zfs_unmount_001_pos', 'zfs_unmount_002_pos', 'zfs_unmount_003_pos', 'zfs_unmount_004_pos', 'zfs_unmount_007_neg', 'zfs_unmount_008_neg', 'zfs_unmount_009_pos', 'zfs_unmount_unload_keys'] tags = ['functional', 'cli_root', 'zfs_unmount'] [tests/functional/cli_root/zfs_upgrade] tests = ['zfs_upgrade_001_pos', 'zfs_upgrade_002_pos', 'zfs_upgrade_006_neg', 'zfs_upgrade_007_neg'] tags = ['functional', 'cli_root', 'zfs_upgrade'] [tests/functional/cli_root/zfs_wait] tests = ['zfs_wait_deleteq', 'zfs_wait_getsubopt'] tags = ['functional', 'cli_root', 'zfs_wait'] [tests/functional/cli_root/zpool] tests = ['zpool_001_neg', 'zpool_003_pos', 'zpool_colors'] tags = ['functional', 'cli_root', 'zpool'] [tests/functional/cli_root/zpool_add] tests = ['zpool_add_002_pos', 'zpool_add_003_pos', 'zpool_add_004_pos', 'zpool_add_006_pos', 'zpool_add_007_neg', 'zpool_add_008_neg', 'zpool_add_009_neg'] tags = ['functional', 'cli_root', 'zpool_add'] [tests/functional/cli_root/zpool_attach] tests = ['zpool_attach_001_neg'] tags = ['functional', 'cli_root', 'zpool_attach'] [tests/functional/cli_root/zpool_clear] tests = ['zpool_clear_002_neg'] tags = ['functional', 'cli_root', 'zpool_clear'] [tests/functional/cli_root/zpool_create] tests = ['zpool_create_001_pos', 'zpool_create_002_pos', 'zpool_create_003_pos', 'zpool_create_004_pos', 'zpool_create_007_neg', 'zpool_create_008_pos', 'zpool_create_010_neg', 'zpool_create_011_neg', 'zpool_create_012_neg', 'zpool_create_014_neg', 'zpool_create_015_neg', 'zpool_create_017_neg', 'zpool_create_018_pos', 'zpool_create_019_pos', 'zpool_create_020_pos', 'zpool_create_021_pos', 'zpool_create_022_pos', 'zpool_create_encrypted', 'zpool_create_features_001_pos', 'zpool_create_features_002_pos', 'zpool_create_features_003_pos', 'zpool_create_features_004_neg', 'zpool_create_features_005_pos'] tags = ['functional', 'cli_root', 'zpool_create'] [tests/functional/cli_root/zpool_destroy] tests = ['zpool_destroy_001_pos', 'zpool_destroy_002_pos', 'zpool_destroy_003_neg'] pre = post = tags = ['functional', 'cli_root', 'zpool_destroy'] [tests/functional/cli_root/zpool_detach] tests = ['zpool_detach_001_neg'] tags = ['functional', 'cli_root', 'zpool_detach'] [tests/functional/cli_root/zpool_events] tests = ['zpool_events_clear', 'zpool_events_follow', 'zpool_events_poolname'] tags = ['functional', 'cli_root', 'zpool_events'] [tests/functional/cli_root/zpool_export] tests = ['zpool_export_001_pos', 'zpool_export_002_pos', 'zpool_export_003_neg'] tags = ['functional', 'cli_root', 'zpool_export'] [tests/functional/cli_root/zpool_get] tests = ['zpool_get_001_pos', 'zpool_get_002_pos', 'zpool_get_003_pos', 'zpool_get_004_neg', 'zpool_get_005_pos'] tags = ['functional', 'cli_root', 'zpool_get'] [tests/functional/cli_root/zpool_history] tests = ['zpool_history_001_neg', 'zpool_history_002_pos'] tags = ['functional', 'cli_root', 'zpool_history'] [tests/functional/cli_root/zpool_import] tests = ['zpool_import_003_pos', 'zpool_import_010_pos', 'zpool_import_011_neg', 'zpool_import_014_pos', 'zpool_import_features_001_pos', 'zpool_import_all_001_pos', 'zpool_import_encrypted'] tags = ['functional', 'cli_root', 'zpool_import'] [tests/functional/cli_root/zpool_labelclear] tests = ['zpool_labelclear_active', 'zpool_labelclear_exported', 'zpool_labelclear_removed', 'zpool_labelclear_valid'] pre = post = tags = ['functional', 'cli_root', 'zpool_labelclear'] [tests/functional/cli_root/zpool_initialize] tests = ['zpool_initialize_online_offline'] pre = tags = ['functional', 'cli_root', 'zpool_initialize'] [tests/functional/cli_root/zpool_offline] tests = ['zpool_offline_001_pos', 'zpool_offline_002_neg'] tags = ['functional', 'cli_root', 'zpool_offline'] [tests/functional/cli_root/zpool_online] tests = ['zpool_online_001_pos', 'zpool_online_002_neg'] tags = ['functional', 'cli_root', 'zpool_online'] [tests/functional/cli_root/zpool_remove] tests = ['zpool_remove_001_neg', 'zpool_remove_002_pos', 'zpool_remove_003_pos'] tags = ['functional', 'cli_root', 'zpool_remove'] [tests/functional/cli_root/zpool_replace] tests = ['zpool_replace_001_neg'] tags = ['functional', 'cli_root', 'zpool_replace'] [tests/functional/cli_root/zpool_resilver] tests = ['zpool_resilver_bad_args'] tags = ['functional', 'cli_root', 'zpool_resilver'] [tests/functional/cli_root/zpool_scrub] tests = ['zpool_scrub_001_neg', 'zpool_scrub_003_pos', 'zpool_scrub_encrypted_unloaded', 'zpool_scrub_print_repairing', 'zpool_scrub_offline_device', 'zpool_scrub_multiple_copies'] tags = ['functional', 'cli_root', 'zpool_scrub'] [tests/functional/cli_root/zpool_set] tests = ['zpool_set_001_pos', 'zpool_set_002_neg', 'zpool_set_003_neg', 'zpool_set_ashift', 'zpool_set_features'] tags = ['functional', 'cli_root', 'zpool_set'] [tests/functional/cli_root/zpool_split] tests = ['zpool_split_cliargs', 'zpool_split_devices', 'zpool_split_props', 'zpool_split_vdevs', 'zpool_split_indirect'] tags = ['functional', 'cli_root', 'zpool_split'] [tests/functional/cli_root/zpool_status] tests = ['zpool_status_001_pos', 'zpool_status_002_pos'] tags = ['functional', 'cli_root', 'zpool_status'] [tests/functional/cli_root/zpool_sync] tests = ['zpool_sync_002_neg'] tags = ['functional', 'cli_root', 'zpool_sync'] [tests/functional/cli_root/zpool_trim] tests = ['zpool_trim_attach_detach_add_remove', 'zpool_trim_neg', 'zpool_trim_offline_export_import_online', 'zpool_trim_online_offline', 'zpool_trim_rate_neg', 'zpool_trim_secure', 'zpool_trim_split', 'zpool_trim_start_and_cancel_neg', 'zpool_trim_start_and_cancel_pos'] tags = ['functional', 'zpool_trim'] [tests/functional/cli_root/zpool_upgrade] tests = ['zpool_upgrade_001_pos', 'zpool_upgrade_003_pos', 'zpool_upgrade_005_neg', 'zpool_upgrade_006_neg', 'zpool_upgrade_009_neg'] tags = ['functional', 'cli_root', 'zpool_upgrade'] [tests/functional/cli_root/zpool_wait] tests = ['zpool_wait_no_activity', 'zpool_wait_usage'] tags = ['functional', 'cli_root', 'zpool_wait'] [tests/functional/cli_root/zpool_wait/scan] tests = ['zpool_wait_scrub_flag'] tags = ['functional', 'cli_root', 'zpool_wait'] [tests/functional/cli_user/misc] tests = ['zdb_001_neg', 'zfs_001_neg', 'zfs_allow_001_neg', 'zfs_clone_001_neg', 'zfs_create_001_neg', 'zfs_destroy_001_neg', 'zfs_get_001_neg', 'zfs_inherit_001_neg', 'zfs_mount_001_neg', 'zfs_promote_001_neg', 'zfs_receive_001_neg', 'zfs_rename_001_neg', 'zfs_rollback_001_neg', 'zfs_send_001_neg', 'zfs_set_001_neg', 'zfs_snapshot_001_neg', 'zfs_unallow_001_neg', 'zfs_unmount_001_neg', 'zfs_upgrade_001_neg', 'zpool_001_neg', 'zpool_add_001_neg', 'zpool_attach_001_neg', 'zpool_clear_001_neg', 'zpool_create_001_neg', 'zpool_destroy_001_neg', 'zpool_detach_001_neg', 'zpool_export_001_neg', 'zpool_get_001_neg', 'zpool_history_001_neg', 'zpool_offline_001_neg', 'zpool_online_001_neg', 'zpool_remove_001_neg', 'zpool_scrub_001_neg', 'zpool_set_001_neg', 'zpool_status_001_neg', 'zpool_upgrade_001_neg', 'arcstat_001_pos', 'arc_summary_001_pos', 'arc_summary_002_neg', 'zpool_wait_privilege', 'zilstat_001_pos'] user = tags = ['functional', 'cli_user', 'misc'] [tests/functional/cli_user/zpool_iostat] tests = ['zpool_iostat_001_neg', 'zpool_iostat_002_pos', 'zpool_iostat_003_neg', 'zpool_iostat_004_pos', 'zpool_iostat_-c_disable', 'zpool_iostat_-c_homedir', 'zpool_iostat_-c_searchpath'] user = tags = ['functional', 'cli_user', 'zpool_iostat'] [tests/functional/cli_user/zpool_list] tests = ['zpool_list_001_pos', 'zpool_list_002_neg'] user = tags = ['functional', 'cli_user', 'zpool_list'] [tests/functional/compression] tests = ['compress_003_pos','compress_zstd_bswap'] tags = ['functional', 'compression'] [tests/functional/exec] tests = ['exec_001_pos', 'exec_002_neg'] tags = ['functional', 'exec'] [tests/functional/features/large_dnode] tests = ['large_dnode_003_pos', 'large_dnode_004_neg', 'large_dnode_005_pos', 'large_dnode_007_neg'] tags = ['functional', 'features', 'large_dnode'] [tests/functional/gang_blocks] tests = ['gang_blocks_001_pos'] tags = ['functional', 'gang_blocks'] [tests/functional/grow] pre = post = tests = ['grow_pool_001_pos', 'grow_replicas_001_pos'] tags = ['functional', 'grow'] [tests/functional/history] tests = ['history_004_pos', 'history_005_neg', 'history_007_pos', 'history_009_pos'] tags = ['functional', 'history'] [tests/functional/hkdf] pre = post = tests = ['hkdf_test'] tags = ['functional', 'hkdf'] [tests/functional/inuse] tests = ['inuse_004_pos', 'inuse_005_pos'] post = tags = ['functional', 'inuse'] [tests/functional/large_files] tests = ['large_files_001_pos', 'large_files_002_pos'] tags = ['functional', 'large_files'] [tests/functional/libzfs] tests = ['many_fds', 'libzfs_input'] tags = ['functional', 'libzfs'] [tests/functional/limits] tests = ['filesystem_count', 'snapshot_count'] tags = ['functional', 'limits'] [tests/functional/link_count] tests = ['link_count_root_inode'] tags = ['functional', 'link_count'] [tests/functional/log_spacemap] tests = ['log_spacemap_import_logs'] pre = post = tags = ['functional', 'log_spacemap'] [tests/functional/migration] tests = ['migration_001_pos', 'migration_002_pos', 'migration_003_pos', 'migration_004_pos', 'migration_005_pos', 'migration_006_pos', 'migration_007_pos', 'migration_008_pos', 'migration_009_pos', 'migration_010_pos', 'migration_011_pos', 'migration_012_pos'] tags = ['functional', 'migration'] [tests/functional/mmap] tests = ['mmap_read_001_pos'] tags = ['functional', 'mmap'] [tests/functional/nestedfs] tests = ['nestedfs_001_pos'] tags = ['functional', 'nestedfs'] [tests/functional/nopwrite] tests = ['nopwrite_sync', 'nopwrite_volume'] tags = ['functional', 'nopwrite'] [tests/functional/pool_checkpoint] tests = ['checkpoint_conf_change', 'checkpoint_discard_many', 'checkpoint_removal', 'checkpoint_sm_scale', 'checkpoint_twice'] tags = ['functional', 'pool_checkpoint'] timeout = 1800 [tests/functional/poolversion] tests = ['poolversion_001_pos', 'poolversion_002_pos'] tags = ['functional', 'poolversion'] [tests/functional/redacted_send] tests = ['redacted_compressed', 'redacted_contents', 'redacted_deleted', 'redacted_disabled_feature', 'redacted_incrementals', 'redacted_largeblocks', 'redacted_mixed_recsize', 'redacted_negative', 'redacted_origin', 'redacted_props', 'redacted_resume', 'redacted_size'] tags = ['functional', 'redacted_send'] [tests/functional/raidz] tests = ['raidz_001_neg'] tags = ['functional', 'raidz'] [tests/functional/refquota] tests = ['refquota_001_pos', 'refquota_002_pos', 'refquota_003_pos', 'refquota_004_pos', 'refquota_005_pos', 'refquota_006_neg', 'refquota_007_neg'] tags = ['functional', 'refquota'] [tests/functional/refreserv] tests = ['refreserv_001_pos', 'refreserv_002_pos', 'refreserv_003_pos', 'refreserv_005_pos', 'refreserv_multi_raidz'] tags = ['functional', 'refreserv'] [tests/functional/removal] pre = tests = ['removal_all_vdev', 'removal_sanity', 'removal_with_dedup', 'removal_with_ganging', 'removal_with_faulted'] tags = ['functional', 'removal'] [tests/functional/replacement] tests = ['rebuild_raidz'] tags = ['functional', 'replacement'] [tests/functional/reservation] tests = ['reservation_001_pos', 'reservation_002_pos', 'reservation_003_pos', 'reservation_004_pos', 'reservation_005_pos', 'reservation_006_pos', 'reservation_007_pos', 'reservation_008_pos', 'reservation_009_pos', 'reservation_010_pos', 'reservation_011_pos', 'reservation_012_pos', 'reservation_014_pos', 'reservation_015_pos', 'reservation_016_pos', 'reservation_017_pos', 'reservation_018_pos', 'reservation_019_pos', 'reservation_020_pos', 'reservation_021_neg', 'reservation_022_pos'] tags = ['functional', 'reservation'] [tests/functional/rsend] tests = ['recv_dedup', 'recv_dedup_encrypted_zvol', 'rsend_001_pos', 'rsend_002_pos', 'rsend_003_pos', 'rsend_004_pos', 'rsend_005_pos', 'rsend_006_pos', 'rsend_009_pos', 'rsend_010_pos', 'rsend_011_pos', 'rsend_014_pos', 'rsend_016_neg', 'send-c_verify_contents', 'send-c_volume', 'send-c_zstreamdump', 'send-c_recv_dedup', 'send-L_toggle', 'send_encrypted_hierarchy', 'send_encrypted_props', 'send_encrypted_freeobjects', 'send_encrypted_truncated_files', 'send_freeobjects', 'send_holds', 'send_mixed_raw', 'send-wR_encrypted_zvol', 'send_partial_dataset', 'send_invalid'] tags = ['functional', 'rsend'] [tests/functional/scrub_mirror] tests = ['scrub_mirror_001_pos', 'scrub_mirror_002_pos'] tags = ['functional', 'scrub_mirror'] [tests/functional/slog] tests = ['slog_008_neg', 'slog_009_neg', 'slog_010_neg'] tags = ['functional', 'slog'] [tests/functional/snapshot] tests = ['clone_001_pos', 'rollback_001_pos', 'rollback_002_pos', 'snapshot_001_pos', 'snapshot_002_pos', 'snapshot_003_pos', 'snapshot_004_pos', 'snapshot_005_pos', 'snapshot_006_pos', 'snapshot_007_pos', 'snapshot_008_pos', 'snapshot_009_pos', 'snapshot_010_pos', 'snapshot_011_pos', 'snapshot_012_pos', 'snapshot_013_pos', 'snapshot_014_pos', 'snapshot_017_pos', 'snapshot_018_pos'] tags = ['functional', 'snapshot'] [tests/functional/snapused] tests = ['snapused_002_pos', 'snapused_004_pos', 'snapused_005_pos'] tags = ['functional', 'snapused'] [tests/functional/sparse] tests = ['sparse_001_pos'] tags = ['functional', 'sparse'] [tests/functional/suid] tests = ['suid_write_to_suid', 'suid_write_to_sgid', 'suid_write_to_suid_sgid', 'suid_write_to_none'] tags = ['functional', 'suid'] [tests/functional/append] tests = ['threadsappend_001_pos'] tags = ['functional', 'threadsappend'] [tests/functional/truncate] tests = ['truncate_001_pos', 'truncate_002_pos'] tags = ['functional', 'truncate'] [tests/functional/upgrade] tests = ['upgrade_userobj_001_pos', 'upgrade_readonly_pool'] tags = ['functional', 'upgrade'] [tests/functional/vdev_disk:Linux] pre = post = tests = ['page_alignment'] tags = ['functional', 'vdev_disk'] [tests/functional/vdev_zaps] tests = ['vdev_zaps_001_pos', 'vdev_zaps_003_pos', 'vdev_zaps_004_pos', 'vdev_zaps_005_pos', 'vdev_zaps_006_pos'] tags = ['functional', 'vdev_zaps'] [tests/functional/xattr] tests = ['xattr_001_pos', 'xattr_002_neg', 'xattr_003_neg', 'xattr_004_pos', 'xattr_005_pos', 'xattr_006_pos', 'xattr_007_neg', 'xattr_011_pos', 'xattr_013_pos', 'xattr_compat'] tags = ['functional', 'xattr'] [tests/functional/zvol/zvol_ENOSPC] tests = ['zvol_ENOSPC_001_pos'] tags = ['functional', 'zvol', 'zvol_ENOSPC'] [tests/functional/zvol/zvol_cli] tests = ['zvol_cli_001_pos', 'zvol_cli_002_pos', 'zvol_cli_003_neg'] tags = ['functional', 'zvol', 'zvol_cli'] [tests/functional/zvol/zvol_swap] tests = ['zvol_swap_001_pos', 'zvol_swap_002_pos'] tags = ['functional', 'zvol', 'zvol_swap'] [tests/functional/zpool_influxdb] tests = ['zpool_influxdb'] tags = ['functional', 'zpool_influxdb'] [tests/functional/pyzfs] tests = ['pyzfs_unittest'] pre = post = tags = ['functional', 'pyzfs'] diff --git a/tests/zfs-tests/tests/Makefile.am b/tests/zfs-tests/tests/Makefile.am index 5ab28b2d6c35..c2542287c1d7 100644 --- a/tests/zfs-tests/tests/Makefile.am +++ b/tests/zfs-tests/tests/Makefile.am @@ -1,2251 +1,2252 @@ CLEANFILES = dist_noinst_DATA = include $(top_srcdir)/config/Substfiles.am datadir_zfs_tests_testsdir = $(datadir)/$(PACKAGE)/zfs-tests/tests nobase_dist_datadir_zfs_tests_tests_DATA = \ perf/nfs-sample.cfg \ perf/perf.shlib \ \ perf/fio/mkfiles.fio \ perf/fio/random_reads.fio \ perf/fio/random_readwrite.fio \ perf/fio/random_readwrite_fixed.fio \ perf/fio/random_writes.fio \ perf/fio/sequential_reads.fio \ perf/fio/sequential_readwrite.fio \ perf/fio/sequential_writes.fio nobase_dist_datadir_zfs_tests_tests_SCRIPTS = \ perf/regression/random_reads.ksh \ perf/regression/random_readwrite.ksh \ perf/regression/random_readwrite_fixed.ksh \ perf/regression/random_writes.ksh \ perf/regression/random_writes_zil.ksh \ perf/regression/sequential_reads_arc_cached_clone.ksh \ perf/regression/sequential_reads_arc_cached.ksh \ perf/regression/sequential_reads_dbuf_cached.ksh \ perf/regression/sequential_reads.ksh \ perf/regression/sequential_writes.ksh \ perf/regression/setup.ksh \ \ perf/scripts/prefetch_io.sh # These lists can be regenerated by running make regen-tests at the root, or, on a *clean* source: # find functional/ ! -type d ! -name .gitignore ! -name .dirstamp ! -name '*.Po' ! -executable -name '*.in' | sort | sed 's/\.in$//;s/^/\t/;$!s/$/ \\/' # find functional/ ! -type d ! -name .gitignore ! -name .dirstamp ! -name '*.Po' -executable -name '*.in' | sort | sed 's/\.in$//;s/^/\t/;$!s/$/ \\/' # find functional/ ! -type d ! -name .gitignore ! -name .dirstamp ! -name '*.Po' ! -name '*.in' ! -name '*.c' | grep -Fe /simd -e /tmpfile | sort | sed 's/^/\t/;$!s/$/ \\/' # find functional/ ! -type d ! -name .gitignore ! -name .dirstamp ! -name '*.Po' ! -executable ! -name '*.in' ! -name '*.c' | grep -vFe /simd -e /tmpfile | sort | sed 's/^/\t/;$!s/$/ \\/' # find functional/ ! -type d ! -name .gitignore ! -name .dirstamp ! -name '*.Po' -executable ! -name '*.in' ! -name '*.c' | grep -vFe /simd -e /tmpfile | sort | sed 's/^/\t/;$!s/$/ \\/' # # simd and tmpfile are Linux-only and not installed elsewhere # # C programs are specced in ../Makefile.am above as part of the main Makefile find_common := find functional/ ! -type d ! -name .gitignore ! -name .dirstamp ! -name '*.Po' regen: @$(MAKE) -C $(top_builddir) clean @$(MAKE) clean $(SED) $(ac_inplace) '/^# -- >8 --/q' Makefile.am echo >> Makefile.am echo 'nobase_nodist_datadir_zfs_tests_tests_DATA = \' >> Makefile.am $(find_common) ! -executable -name '*.in' | sort | sed 's/\.in$$//;s/^/\t/;$$!s/$$/ \\/' >> Makefile.am echo 'nobase_nodist_datadir_zfs_tests_tests_SCRIPTS = \' >> Makefile.am $(find_common) -executable -name '*.in' | sort | sed 's/\.in$$//;s/^/\t/;$$!s/$$/ \\/' >> Makefile.am echo >> Makefile.am echo 'SUBSTFILES += $$(nobase_nodist_datadir_zfs_tests_tests_DATA) $$(nobase_nodist_datadir_zfs_tests_tests_SCRIPTS)' >> Makefile.am echo >> Makefile.am echo 'if BUILD_LINUX' >> Makefile.am echo 'nobase_dist_datadir_zfs_tests_tests_SCRIPTS += \' >> Makefile.am $(find_common) ! -name '*.in' ! -name '*.c' | grep -Fe /simd -e /tmpfile | sort | sed 's/^/\t/;$$!s/$$/ \\/' >> Makefile.am echo 'endif' >> Makefile.am echo >> Makefile.am echo 'nobase_dist_datadir_zfs_tests_tests_DATA += \' >> Makefile.am $(find_common) ! -executable ! -name '*.in' ! -name '*.c' | grep -vFe /simd -e /tmpfile | sort | sed 's/^/\t/;$$!s/$$/ \\/' >> Makefile.am echo >> Makefile.am echo 'nobase_dist_datadir_zfs_tests_tests_SCRIPTS += \' >> Makefile.am $(find_common) -executable ! -name '*.in' ! -name '*.c' | grep -vFe /simd -e /tmpfile | sort | sed 's/^/\t/;$$!s/$$/ \\/' >> Makefile.am # -- >8 -- nobase_nodist_datadir_zfs_tests_tests_DATA = \ functional/pam/utilities.kshlib nobase_nodist_datadir_zfs_tests_tests_SCRIPTS = \ functional/pyzfs/pyzfs_unittest.ksh SUBSTFILES += $(nobase_nodist_datadir_zfs_tests_tests_DATA) $(nobase_nodist_datadir_zfs_tests_tests_SCRIPTS) if BUILD_LINUX nobase_dist_datadir_zfs_tests_tests_SCRIPTS += \ functional/simd/simd_supported.ksh \ functional/tmpfile/cleanup.ksh \ functional/tmpfile/setup.ksh \ functional/luks/luks_sanity.ksh endif nobase_dist_datadir_zfs_tests_tests_DATA += \ functional/acl/acl.cfg \ functional/acl/acl_common.kshlib \ functional/alloc_class/alloc_class.cfg \ functional/alloc_class/alloc_class.kshlib \ functional/atime/atime.cfg \ functional/atime/atime_common.kshlib \ functional/bclone/bclone.cfg \ functional/bclone/bclone_common.kshlib \ functional/bclone/bclone_corner_cases.kshlib \ functional/block_cloning/block_cloning.kshlib \ functional/cache/cache.cfg \ functional/cache/cache.kshlib \ functional/cachefile/cachefile.cfg \ functional/cachefile/cachefile.kshlib \ functional/casenorm/casenorm.cfg \ functional/casenorm/casenorm.kshlib \ functional/channel_program/channel_common.kshlib \ functional/channel_program/lua_core/tst.args_to_lua.out \ functional/channel_program/lua_core/tst.args_to_lua.zcp \ functional/channel_program/lua_core/tst.divide_by_zero.err \ functional/channel_program/lua_core/tst.divide_by_zero.zcp \ functional/channel_program/lua_core/tst.encryption.zcp \ functional/channel_program/lua_core/tst.exists.zcp \ functional/channel_program/lua_core/tst.large_prog.out \ functional/channel_program/lua_core/tst.large_prog.zcp \ functional/channel_program/lua_core/tst.lib_base.lua \ functional/channel_program/lua_core/tst.lib_coroutine.lua \ functional/channel_program/lua_core/tst.lib_strings.lua \ functional/channel_program/lua_core/tst.lib_table.lua \ functional/channel_program/lua_core/tst.nested_neg.zcp \ functional/channel_program/lua_core/tst.nested_pos.zcp \ functional/channel_program/lua_core/tst.recursive.zcp \ functional/channel_program/lua_core/tst.return_large.zcp \ functional/channel_program/lua_core/tst.return_recursive_table.zcp \ functional/channel_program/lua_core/tst.stack_gsub.err \ functional/channel_program/lua_core/tst.stack_gsub.zcp \ functional/channel_program/lua_core/tst.timeout.zcp \ functional/channel_program/synctask_core/tst.bookmark.copy.zcp \ functional/channel_program/synctask_core/tst.bookmark.create.zcp \ functional/channel_program/synctask_core/tst.clone.zcp \ functional/channel_program/synctask_core/tst.get_index_props.out \ functional/channel_program/synctask_core/tst.get_index_props.zcp \ functional/channel_program/synctask_core/tst.get_number_props.out \ functional/channel_program/synctask_core/tst.get_number_props.zcp \ functional/channel_program/synctask_core/tst.get_string_props.out \ functional/channel_program/synctask_core/tst.get_string_props.zcp \ functional/channel_program/synctask_core/tst.promote_conflict.zcp \ functional/channel_program/synctask_core/tst.set_props.zcp \ functional/channel_program/synctask_core/tst.snapshot_destroy.zcp \ functional/channel_program/synctask_core/tst.snapshot_neg.zcp \ functional/channel_program/synctask_core/tst.snapshot_recursive.zcp \ functional/channel_program/synctask_core/tst.snapshot_rename.zcp \ functional/channel_program/synctask_core/tst.snapshot_simple.zcp \ functional/checksum/default.cfg \ functional/clean_mirror/clean_mirror_common.kshlib \ functional/clean_mirror/default.cfg \ functional/crypto/aes_ccm_test.json \ functional/crypto/aes_ccm_test.txt \ functional/crypto/aes_gcm_test.json \ functional/crypto/aes_gcm_test.txt \ functional/cli_root/cli_common.kshlib \ functional/cli_root/zfs_copies/zfs_copies.cfg \ functional/cli_root/zfs_copies/zfs_copies.kshlib \ functional/cli_root/zfs_create/properties.kshlib \ functional/cli_root/zfs_create/zfs_create.cfg \ functional/cli_root/zfs_create/zfs_create_common.kshlib \ functional/cli_root/zfs_destroy/zfs_destroy.cfg \ functional/cli_root/zfs_destroy/zfs_destroy_common.kshlib \ functional/cli_root/zfs_get/zfs_get_common.kshlib \ functional/cli_root/zfs_get/zfs_get_list_d.kshlib \ functional/cli_root/zfs_jail/jail.conf \ functional/cli_root/zfs_load-key/HEXKEY \ functional/cli_root/zfs_load-key/PASSPHRASE \ functional/cli_root/zfs_load-key/RAWKEY \ functional/cli_root/zfs_load-key/zfs_load-key.cfg \ functional/cli_root/zfs_load-key/zfs_load-key_common.kshlib \ functional/cli_root/zfs_mount/zfs_mount.cfg \ functional/cli_root/zfs_mount/zfs_mount.kshlib \ functional/cli_root/zfs_promote/zfs_promote.cfg \ functional/cli_root/zfs_receive/zstd_test_data.txt \ functional/cli_root/zfs_rename/zfs_rename.cfg \ functional/cli_root/zfs_rename/zfs_rename.kshlib \ functional/cli_root/zfs_rollback/zfs_rollback.cfg \ functional/cli_root/zfs_rollback/zfs_rollback_common.kshlib \ functional/cli_root/zfs_send/zfs_send.cfg \ functional/cli_root/zfs_set/zfs_set_common.kshlib \ functional/cli_root/zfs_share/zfs_share.cfg \ functional/cli_root/zfs_snapshot/zfs_snapshot.cfg \ functional/cli_root/zfs_unmount/zfs_unmount.cfg \ functional/cli_root/zfs_unmount/zfs_unmount.kshlib \ functional/cli_root/zfs_upgrade/zfs_upgrade.kshlib \ functional/cli_root/zfs_wait/zfs_wait.kshlib \ functional/cli_root/zpool_add/zpool_add.cfg \ functional/cli_root/zpool_add/zpool_add.kshlib \ functional/cli_root/zpool_clear/zpool_clear.cfg \ functional/cli_root/zpool_create/draidcfg.gz \ functional/cli_root/zpool_create/zpool_create.cfg \ functional/cli_root/zpool_create/zpool_create.shlib \ functional/cli_root/zpool_destroy/zpool_destroy.cfg \ functional/cli_root/zpool_events/zpool_events.cfg \ functional/cli_root/zpool_events/zpool_events.kshlib \ functional/cli_root/zpool_expand/zpool_expand.cfg \ functional/cli_root/zpool_export/zpool_export.cfg \ functional/cli_root/zpool_export/zpool_export.kshlib \ functional/cli_root/zpool_get/vdev_get.cfg \ functional/cli_root/zpool_get/zpool_get.cfg \ functional/cli_root/zpool_get/zpool_get_parsable.cfg \ functional/cli_root/zpool_import/blockfiles/cryptv0.dat.bz2 \ functional/cli_root/zpool_import/blockfiles/missing_ivset.dat.bz2 \ functional/cli_root/zpool_import/blockfiles/unclean_export.dat.bz2 \ functional/cli_root/zpool_import/zpool_import.cfg \ functional/cli_root/zpool_import/zpool_import.kshlib \ functional/cli_root/zpool_initialize/zpool_initialize.kshlib \ functional/cli_root/zpool_labelclear/labelclear.cfg \ functional/cli_root/zpool_remove/zpool_remove.cfg \ functional/cli_root/zpool_reopen/zpool_reopen.cfg \ functional/cli_root/zpool_reopen/zpool_reopen.shlib \ functional/cli_root/zpool_resilver/zpool_resilver.cfg \ functional/cli_root/zpool_scrub/zpool_scrub.cfg \ functional/cli_root/zpool_split/zpool_split.cfg \ functional/cli_root/zpool_trim/zpool_trim.kshlib \ functional/cli_root/zpool_upgrade/blockfiles/zfs-broken-mirror1.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-broken-mirror2.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v10.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v11.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v12.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v13.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v14.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v15.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v1.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v1mirror1.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v1mirror2.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v1mirror3.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v1raidz1.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v1raidz2.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v1raidz3.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v1stripe1.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v1stripe2.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v1stripe3.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v2.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v2mirror1.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v2mirror2.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v2mirror3.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v2raidz1.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v2raidz2.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v2raidz3.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v2stripe1.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v2stripe2.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v2stripe3.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3hotspare1.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3hotspare2.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3hotspare3.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3mirror1.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3mirror2.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3mirror3.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3raidz1.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3raidz21.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3raidz22.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3raidz23.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3raidz2.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3raidz3.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3stripe1.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3stripe2.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v3stripe3.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v4.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v5.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v6.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v7.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v8.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v999.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-v9.dat.bz2 \ functional/cli_root/zpool_upgrade/blockfiles/zfs-pool-vBROKEN.dat.bz2 \ functional/cli_root/zpool_upgrade/zpool_upgrade.cfg \ functional/cli_root/zpool_upgrade/zpool_upgrade.kshlib \ functional/cli_root/zpool_wait/zpool_wait.kshlib \ functional/cli_root/zhack/library.kshlib \ functional/cli_user/misc/misc.cfg \ functional/cli_user/zfs_list/zfs_list.cfg \ functional/cli_user/zfs_list/zfs_list.kshlib \ functional/compression/compress.cfg \ functional/compression/testpool_zstd.tar.gz \ functional/deadman/deadman.cfg \ functional/delegate/delegate.cfg \ functional/delegate/delegate_common.kshlib \ functional/devices/devices.cfg \ functional/devices/devices_common.kshlib \ functional/direct/dio.cfg \ functional/direct/dio.kshlib \ functional/events/events.cfg \ functional/events/events_common.kshlib \ functional/fault/fault.cfg \ functional/gang_blocks/gang_blocks.kshlib \ functional/grow/grow.cfg \ functional/history/history.cfg \ functional/history/history_common.kshlib \ functional/history/i386.migratedpool.DAT.Z \ functional/history/i386.orig_history.txt \ functional/history/sparc.migratedpool.DAT.Z \ functional/history/sparc.orig_history.txt \ functional/history/zfs-pool-v4.dat.Z \ functional/inheritance/config001.cfg \ functional/inheritance/config002.cfg \ functional/inheritance/config003.cfg \ functional/inheritance/config004.cfg \ functional/inheritance/config005.cfg \ functional/inheritance/config006.cfg \ functional/inheritance/config007.cfg \ functional/inheritance/config008.cfg \ functional/inheritance/config009.cfg \ functional/inheritance/config010.cfg \ functional/inheritance/config011.cfg \ functional/inheritance/config012.cfg \ functional/inheritance/config013.cfg \ functional/inheritance/config014.cfg \ functional/inheritance/config015.cfg \ functional/inheritance/config016.cfg \ functional/inheritance/config017.cfg \ functional/inheritance/config018.cfg \ functional/inheritance/config019.cfg \ functional/inheritance/config020.cfg \ functional/inheritance/config021.cfg \ functional/inheritance/config022.cfg \ functional/inheritance/config023.cfg \ functional/inheritance/config024.cfg \ functional/inheritance/inherit.kshlib \ functional/inheritance/README.config \ functional/inheritance/README.state \ functional/inheritance/state001.cfg \ functional/inheritance/state002.cfg \ functional/inheritance/state003.cfg \ functional/inheritance/state004.cfg \ functional/inheritance/state005.cfg \ functional/inheritance/state006.cfg \ functional/inheritance/state007.cfg \ functional/inheritance/state008.cfg \ functional/inheritance/state009.cfg \ functional/inheritance/state010.cfg \ functional/inheritance/state011.cfg \ functional/inheritance/state012.cfg \ functional/inheritance/state013.cfg \ functional/inheritance/state014.cfg \ functional/inheritance/state015.cfg \ functional/inheritance/state016.cfg \ functional/inheritance/state017.cfg \ functional/inheritance/state018.cfg \ functional/inheritance/state019.cfg \ functional/inheritance/state020.cfg \ functional/inheritance/state021.cfg \ functional/inheritance/state022.cfg \ functional/inheritance/state023.cfg \ functional/inheritance/state024.cfg \ functional/inuse/inuse.cfg \ functional/io/io.cfg \ functional/l2arc/l2arc.cfg \ functional/largest_pool/largest_pool.cfg \ functional/migration/migration.cfg \ functional/migration/migration.kshlib \ functional/mmap/mmap.cfg \ functional/mmp/mmp.cfg \ functional/mmp/mmp.kshlib \ functional/mv_files/mv_files.cfg \ functional/mv_files/mv_files_common.kshlib \ functional/nopwrite/nopwrite.shlib \ functional/no_space/enospc.cfg \ functional/online_offline/online_offline.cfg \ functional/pool_checkpoint/pool_checkpoint.kshlib \ functional/projectquota/projectquota.cfg \ functional/projectquota/projectquota_common.kshlib \ functional/quota/quota.cfg \ functional/quota/quota.kshlib \ functional/redacted_send/redacted.cfg \ functional/redacted_send/redacted.kshlib \ functional/redundancy/redundancy.cfg \ functional/redundancy/redundancy.kshlib \ functional/refreserv/refreserv.cfg \ functional/removal/removal.kshlib \ functional/replacement/replacement.cfg \ functional/reservation/reservation.cfg \ functional/reservation/reservation.shlib \ functional/rsend/dedup_encrypted_zvol.bz2 \ functional/rsend/dedup_encrypted_zvol.zsend.bz2 \ functional/rsend/dedup.zsend.bz2 \ functional/rsend/fs.tar.gz \ functional/rsend/rsend.cfg \ functional/rsend/rsend.kshlib \ functional/scrub_mirror/default.cfg \ functional/scrub_mirror/scrub_mirror_common.kshlib \ functional/slog/slog.cfg \ functional/slog/slog.kshlib \ functional/snapshot/snapshot.cfg \ functional/snapused/snapused.kshlib \ functional/sparse/sparse.cfg \ functional/trim/trim.cfg \ functional/trim/trim.kshlib \ functional/truncate/truncate.cfg \ functional/upgrade/upgrade_common.kshlib \ functional/user_namespace/user_namespace.cfg \ functional/user_namespace/user_namespace_common.kshlib \ functional/userquota/13709_reproducer.bz2 \ functional/userquota/userquota.cfg \ functional/userquota/userquota_common.kshlib \ functional/vdev_zaps/vdev_zaps.kshlib \ functional/xattr/xattr.cfg \ functional/xattr/xattr_common.kshlib \ functional/zvol/zvol.cfg \ functional/zvol/zvol_cli/zvol_cli.cfg \ functional/zvol/zvol_common.shlib \ functional/zvol/zvol_ENOSPC/zvol_ENOSPC.cfg \ functional/zvol/zvol_misc/zvol_misc_common.kshlib \ functional/zvol/zvol_swap/zvol_swap.cfg \ functional/idmap_mount/idmap_mount.cfg \ functional/idmap_mount/idmap_mount_common.kshlib nobase_dist_datadir_zfs_tests_tests_SCRIPTS += \ functional/acl/off/cleanup.ksh \ functional/acl/off/dosmode.ksh \ functional/acl/off/posixmode.ksh \ functional/acl/off/setup.ksh \ functional/acl/posix/cleanup.ksh \ functional/acl/posix/posix_001_pos.ksh \ functional/acl/posix/posix_002_pos.ksh \ functional/acl/posix/posix_003_pos.ksh \ functional/acl/posix/posix_004_pos.ksh \ functional/acl/posix-sa/cleanup.ksh \ functional/acl/posix-sa/posix_001_pos.ksh \ functional/acl/posix-sa/posix_002_pos.ksh \ functional/acl/posix-sa/posix_003_pos.ksh \ functional/acl/posix-sa/posix_004_pos.ksh \ functional/acl/posix-sa/setup.ksh \ functional/acl/posix/setup.ksh \ functional/alloc_class/alloc_class_001_pos.ksh \ functional/alloc_class/alloc_class_002_neg.ksh \ functional/alloc_class/alloc_class_003_pos.ksh \ functional/alloc_class/alloc_class_004_pos.ksh \ functional/alloc_class/alloc_class_005_pos.ksh \ functional/alloc_class/alloc_class_006_pos.ksh \ functional/alloc_class/alloc_class_007_pos.ksh \ functional/alloc_class/alloc_class_008_pos.ksh \ functional/alloc_class/alloc_class_009_pos.ksh \ functional/alloc_class/alloc_class_010_pos.ksh \ functional/alloc_class/alloc_class_011_neg.ksh \ functional/alloc_class/alloc_class_012_pos.ksh \ functional/alloc_class/alloc_class_013_pos.ksh \ functional/alloc_class/alloc_class_016_pos.ksh \ functional/alloc_class/cleanup.ksh \ functional/alloc_class/setup.ksh \ functional/append/file_append.ksh \ functional/append/threadsappend_001_pos.ksh \ functional/append/cleanup.ksh \ functional/append/setup.ksh \ functional/arc/arcstats_runtime_tuning.ksh \ functional/arc/cleanup.ksh \ functional/arc/dbufstats_001_pos.ksh \ functional/arc/dbufstats_002_pos.ksh \ functional/arc/dbufstats_003_pos.ksh \ functional/arc/setup.ksh \ functional/atime/atime_001_pos.ksh \ functional/atime/atime_002_neg.ksh \ functional/atime/atime_003_pos.ksh \ functional/atime/cleanup.ksh \ functional/atime/root_atime_off.ksh \ functional/atime/root_atime_on.ksh \ functional/atime/root_relatime_on.ksh \ functional/atime/setup.ksh \ functional/bclone/bclone_crossfs_corner_cases.ksh \ functional/bclone/bclone_crossfs_corner_cases_limited.ksh \ functional/bclone/bclone_crossfs_data.ksh \ functional/bclone/bclone_crossfs_embedded.ksh \ functional/bclone/bclone_crossfs_hole.ksh \ functional/bclone/bclone_diffprops_all.ksh \ functional/bclone/bclone_diffprops_checksum.ksh \ functional/bclone/bclone_diffprops_compress.ksh \ functional/bclone/bclone_diffprops_copies.ksh \ functional/bclone/bclone_diffprops_recordsize.ksh \ functional/bclone/bclone_prop_sync.ksh \ functional/bclone/bclone_samefs_corner_cases.ksh \ functional/bclone/bclone_samefs_corner_cases_limited.ksh \ functional/bclone/bclone_samefs_data.ksh \ functional/bclone/bclone_samefs_embedded.ksh \ functional/bclone/bclone_samefs_hole.ksh \ functional/bclone/cleanup.ksh \ functional/bclone/setup.ksh \ functional/block_cloning/cleanup.ksh \ functional/block_cloning/setup.ksh \ functional/block_cloning/block_cloning_clone_mmap_cached.ksh \ functional/block_cloning/block_cloning_clone_mmap_write.ksh \ functional/block_cloning/block_cloning_copyfilerange_cross_dataset.ksh \ functional/block_cloning/block_cloning_copyfilerange_fallback.ksh \ functional/block_cloning/block_cloning_copyfilerange_fallback_same_txg.ksh \ functional/block_cloning/block_cloning_copyfilerange.ksh \ functional/block_cloning/block_cloning_copyfilerange_partial.ksh \ functional/block_cloning/block_cloning_disabled_copyfilerange.ksh \ functional/block_cloning/block_cloning_disabled_ficlone.ksh \ functional/block_cloning/block_cloning_disabled_ficlonerange.ksh \ functional/block_cloning/block_cloning_ficlone.ksh \ functional/block_cloning/block_cloning_ficlonerange.ksh \ functional/block_cloning/block_cloning_ficlonerange_partial.ksh \ functional/block_cloning/block_cloning_cross_enc_dataset.ksh \ functional/block_cloning/block_cloning_replay.ksh \ functional/block_cloning/block_cloning_replay_encrypted.ksh \ functional/block_cloning/block_cloning_lwb_buffer_overflow.ksh \ functional/block_cloning/block_cloning_rlimit_fsize.ksh \ functional/block_cloning/block_cloning_large_offset.ksh \ functional/bootfs/bootfs_001_pos.ksh \ functional/bootfs/bootfs_002_neg.ksh \ functional/bootfs/bootfs_003_pos.ksh \ functional/bootfs/bootfs_004_neg.ksh \ functional/bootfs/bootfs_005_neg.ksh \ functional/bootfs/bootfs_006_pos.ksh \ functional/bootfs/bootfs_007_pos.ksh \ functional/bootfs/bootfs_008_pos.ksh \ functional/bootfs/cleanup.ksh \ functional/bootfs/setup.ksh \ functional/btree/btree_negative.ksh \ functional/btree/btree_positive.ksh \ functional/cache/cache_001_pos.ksh \ functional/cache/cache_002_pos.ksh \ functional/cache/cache_003_pos.ksh \ functional/cache/cache_004_neg.ksh \ functional/cache/cache_005_neg.ksh \ functional/cache/cache_006_pos.ksh \ functional/cache/cache_007_neg.ksh \ functional/cache/cache_008_neg.ksh \ functional/cache/cache_009_pos.ksh \ functional/cache/cache_010_pos.ksh \ functional/cache/cache_011_pos.ksh \ functional/cache/cache_012_pos.ksh \ functional/cache/cleanup.ksh \ functional/cachefile/cachefile_001_pos.ksh \ functional/cachefile/cachefile_002_pos.ksh \ functional/cachefile/cachefile_003_pos.ksh \ functional/cachefile/cachefile_004_pos.ksh \ functional/cachefile/cleanup.ksh \ functional/cachefile/setup.ksh \ functional/cache/setup.ksh \ functional/casenorm/case_all_values.ksh \ functional/casenorm/cleanup.ksh \ functional/casenorm/insensitive_formd_delete.ksh \ functional/casenorm/insensitive_formd_lookup.ksh \ functional/casenorm/insensitive_none_delete.ksh \ functional/casenorm/insensitive_none_lookup.ksh \ functional/casenorm/mixed_create_failure.ksh \ functional/casenorm/mixed_formd_delete.ksh \ functional/casenorm/mixed_formd_lookup_ci.ksh \ functional/casenorm/mixed_formd_lookup.ksh \ functional/casenorm/mixed_none_delete.ksh \ functional/casenorm/mixed_none_lookup_ci.ksh \ functional/casenorm/mixed_none_lookup.ksh \ functional/casenorm/norm_all_values.ksh \ functional/casenorm/sensitive_formd_delete.ksh \ functional/casenorm/sensitive_formd_lookup.ksh \ functional/casenorm/sensitive_none_delete.ksh \ functional/casenorm/sensitive_none_lookup.ksh \ functional/casenorm/setup.ksh \ functional/channel_program/lua_core/cleanup.ksh \ functional/channel_program/lua_core/setup.ksh \ functional/channel_program/lua_core/tst.args_to_lua.ksh \ functional/channel_program/lua_core/tst.divide_by_zero.ksh \ functional/channel_program/lua_core/tst.encryption.ksh \ functional/channel_program/lua_core/tst.exists.ksh \ functional/channel_program/lua_core/tst.integer_illegal.ksh \ functional/channel_program/lua_core/tst.integer_overflow.ksh \ functional/channel_program/lua_core/tst.language_functions_neg.ksh \ functional/channel_program/lua_core/tst.language_functions_pos.ksh \ functional/channel_program/lua_core/tst.large_prog.ksh \ functional/channel_program/lua_core/tst.libraries.ksh \ functional/channel_program/lua_core/tst.memory_limit.ksh \ functional/channel_program/lua_core/tst.nested_neg.ksh \ functional/channel_program/lua_core/tst.nested_pos.ksh \ functional/channel_program/lua_core/tst.nvlist_to_lua.ksh \ functional/channel_program/lua_core/tst.recursive_neg.ksh \ functional/channel_program/lua_core/tst.recursive_pos.ksh \ functional/channel_program/lua_core/tst.return_large.ksh \ functional/channel_program/lua_core/tst.return_nvlist_neg.ksh \ functional/channel_program/lua_core/tst.return_nvlist_pos.ksh \ functional/channel_program/lua_core/tst.return_recursive_table.ksh \ functional/channel_program/lua_core/tst.stack_gsub.ksh \ functional/channel_program/lua_core/tst.timeout.ksh \ functional/channel_program/synctask_core/cleanup.ksh \ functional/channel_program/synctask_core/setup.ksh \ functional/channel_program/synctask_core/tst.bookmark.copy.ksh \ functional/channel_program/synctask_core/tst.bookmark.create.ksh \ functional/channel_program/synctask_core/tst.clone.ksh \ functional/channel_program/synctask_core/tst.destroy_fs.ksh \ functional/channel_program/synctask_core/tst.destroy_snap.ksh \ functional/channel_program/synctask_core/tst.get_count_and_limit.ksh \ functional/channel_program/synctask_core/tst.get_index_props.ksh \ functional/channel_program/synctask_core/tst.get_mountpoint.ksh \ functional/channel_program/synctask_core/tst.get_neg.ksh \ functional/channel_program/synctask_core/tst.get_number_props.ksh \ functional/channel_program/synctask_core/tst.get_string_props.ksh \ functional/channel_program/synctask_core/tst.get_type.ksh \ functional/channel_program/synctask_core/tst.get_userquota.ksh \ functional/channel_program/synctask_core/tst.get_written.ksh \ functional/channel_program/synctask_core/tst.inherit.ksh \ functional/channel_program/synctask_core/tst.list_bookmarks.ksh \ functional/channel_program/synctask_core/tst.list_children.ksh \ functional/channel_program/synctask_core/tst.list_clones.ksh \ functional/channel_program/synctask_core/tst.list_holds.ksh \ functional/channel_program/synctask_core/tst.list_snapshots.ksh \ functional/channel_program/synctask_core/tst.list_system_props.ksh \ functional/channel_program/synctask_core/tst.list_user_props.ksh \ functional/channel_program/synctask_core/tst.parse_args_neg.ksh \ functional/channel_program/synctask_core/tst.promote_conflict.ksh \ functional/channel_program/synctask_core/tst.promote_multiple.ksh \ functional/channel_program/synctask_core/tst.promote_simple.ksh \ functional/channel_program/synctask_core/tst.rollback_mult.ksh \ functional/channel_program/synctask_core/tst.rollback_one.ksh \ functional/channel_program/synctask_core/tst.set_props.ksh \ functional/channel_program/synctask_core/tst.snapshot_destroy.ksh \ functional/channel_program/synctask_core/tst.snapshot_neg.ksh \ functional/channel_program/synctask_core/tst.snapshot_recursive.ksh \ functional/channel_program/synctask_core/tst.snapshot_rename.ksh \ functional/channel_program/synctask_core/tst.snapshot_simple.ksh \ functional/channel_program/synctask_core/tst.terminate_by_signal.ksh \ functional/chattr/chattr_001_pos.ksh \ functional/chattr/chattr_002_neg.ksh \ functional/chattr/cleanup.ksh \ functional/chattr/setup.ksh \ functional/checksum/cleanup.ksh \ functional/checksum/filetest_001_pos.ksh \ functional/checksum/filetest_002_pos.ksh \ functional/checksum/run_blake3_test.ksh \ functional/checksum/run_edonr_test.ksh \ functional/checksum/run_sha2_test.ksh \ functional/checksum/run_skein_test.ksh \ functional/checksum/setup.ksh \ functional/clean_mirror/clean_mirror_001_pos.ksh \ functional/clean_mirror/clean_mirror_002_pos.ksh \ functional/clean_mirror/clean_mirror_003_pos.ksh \ functional/clean_mirror/clean_mirror_004_pos.ksh \ functional/clean_mirror/cleanup.ksh \ functional/clean_mirror/setup.ksh \ functional/cli_root/json/cleanup.ksh \ functional/cli_root/json/setup.ksh \ functional/cli_root/json/json_sanity.ksh \ functional/cli_root/zinject/zinject_args.ksh \ functional/cli_root/zinject/zinject_counts.ksh \ functional/cli_root/zinject/zinject_probe.ksh \ functional/cli_root/zdb/zdb_002_pos.ksh \ functional/cli_root/zdb/zdb_003_pos.ksh \ functional/cli_root/zdb/zdb_004_pos.ksh \ functional/cli_root/zdb/zdb_005_pos.ksh \ functional/cli_root/zdb/zdb_006_pos.ksh \ functional/cli_root/zdb/zdb_args_neg.ksh \ functional/cli_root/zdb/zdb_args_pos.ksh \ functional/cli_root/zdb/zdb_backup.ksh \ functional/cli_root/zdb/zdb_block_size_histogram.ksh \ functional/cli_root/zdb/zdb_checksum.ksh \ functional/cli_root/zdb/zdb_decompress.ksh \ functional/cli_root/zdb/zdb_decompress_zstd.ksh \ functional/cli_root/zdb/zdb_display_block.ksh \ functional/cli_root/zdb/zdb_encrypted.ksh \ functional/cli_root/zdb/zdb_label_checksum.ksh \ functional/cli_root/zdb/zdb_object_range_neg.ksh \ functional/cli_root/zdb/zdb_object_range_pos.ksh \ functional/cli_root/zdb/zdb_objset_id.ksh \ functional/cli_root/zdb/zdb_recover_2.ksh \ functional/cli_root/zdb/zdb_recover.ksh \ functional/cli_root/zdb/zdb_tunables.ksh \ functional/cli_root/zfs_bookmark/cleanup.ksh \ functional/cli_root/zfs_bookmark/setup.ksh \ functional/cli_root/zfs_bookmark/zfs_bookmark_cliargs.ksh \ functional/cli_root/zfs_change-key/cleanup.ksh \ functional/cli_root/zfs_change-key/setup.ksh \ functional/cli_root/zfs_change-key/zfs_change-key_child.ksh \ functional/cli_root/zfs_change-key/zfs_change-key_clones.ksh \ functional/cli_root/zfs_change-key/zfs_change-key_format.ksh \ functional/cli_root/zfs_change-key/zfs_change-key_inherit.ksh \ functional/cli_root/zfs_change-key/zfs_change-key.ksh \ functional/cli_root/zfs_change-key/zfs_change-key_load.ksh \ functional/cli_root/zfs_change-key/zfs_change-key_location.ksh \ functional/cli_root/zfs_change-key/zfs_change-key_pbkdf2iters.ksh \ functional/cli_root/zfs/cleanup.ksh \ functional/cli_root/zfs_clone/cleanup.ksh \ functional/cli_root/zfs_clone/setup.ksh \ functional/cli_root/zfs_clone/zfs_clone_001_neg.ksh \ functional/cli_root/zfs_clone/zfs_clone_002_pos.ksh \ functional/cli_root/zfs_clone/zfs_clone_003_pos.ksh \ functional/cli_root/zfs_clone/zfs_clone_004_pos.ksh \ functional/cli_root/zfs_clone/zfs_clone_005_pos.ksh \ functional/cli_root/zfs_clone/zfs_clone_006_pos.ksh \ functional/cli_root/zfs_clone/zfs_clone_007_pos.ksh \ functional/cli_root/zfs_clone/zfs_clone_008_neg.ksh \ functional/cli_root/zfs_clone/zfs_clone_009_neg.ksh \ functional/cli_root/zfs_clone/zfs_clone_010_pos.ksh \ functional/cli_root/zfs_clone/zfs_clone_deeply_nested.ksh \ functional/cli_root/zfs_clone/zfs_clone_encrypted.ksh \ functional/cli_root/zfs_clone/zfs_clone_rm_nested.ksh \ functional/cli_root/zfs_copies/cleanup.ksh \ functional/cli_root/zfs_copies/setup.ksh \ functional/cli_root/zfs_copies/zfs_copies_001_pos.ksh \ functional/cli_root/zfs_copies/zfs_copies_002_pos.ksh \ functional/cli_root/zfs_copies/zfs_copies_003_pos.ksh \ functional/cli_root/zfs_copies/zfs_copies_004_neg.ksh \ functional/cli_root/zfs_copies/zfs_copies_005_neg.ksh \ functional/cli_root/zfs_copies/zfs_copies_006_pos.ksh \ functional/cli_root/zfs_create/cleanup.ksh \ functional/cli_root/zfs_create/setup.ksh \ functional/cli_root/zfs_create/zfs_create_001_pos.ksh \ functional/cli_root/zfs_create/zfs_create_002_pos.ksh \ functional/cli_root/zfs_create/zfs_create_003_pos.ksh \ functional/cli_root/zfs_create/zfs_create_004_pos.ksh \ functional/cli_root/zfs_create/zfs_create_005_pos.ksh \ functional/cli_root/zfs_create/zfs_create_006_pos.ksh \ functional/cli_root/zfs_create/zfs_create_007_pos.ksh \ functional/cli_root/zfs_create/zfs_create_008_neg.ksh \ functional/cli_root/zfs_create/zfs_create_009_neg.ksh \ functional/cli_root/zfs_create/zfs_create_010_neg.ksh \ functional/cli_root/zfs_create/zfs_create_011_pos.ksh \ functional/cli_root/zfs_create/zfs_create_012_pos.ksh \ functional/cli_root/zfs_create/zfs_create_013_pos.ksh \ functional/cli_root/zfs_create/zfs_create_014_pos.ksh \ functional/cli_root/zfs_create/zfs_create_crypt_combos.ksh \ functional/cli_root/zfs_create/zfs_create_dryrun.ksh \ functional/cli_root/zfs_create/zfs_create_encrypted.ksh \ functional/cli_root/zfs_create/zfs_create_nomount.ksh \ functional/cli_root/zfs_create/zfs_create_verbose.ksh \ functional/cli_root/zfs_destroy/cleanup.ksh \ functional/cli_root/zfs_destroy/setup.ksh \ functional/cli_root/zfs_destroy/zfs_clone_livelist_condense_and_disable.ksh \ functional/cli_root/zfs_destroy/zfs_clone_livelist_condense_races.ksh \ functional/cli_root/zfs_destroy/zfs_clone_livelist_dedup.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_001_pos.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_002_pos.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_003_pos.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_004_pos.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_005_neg.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_006_neg.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_007_neg.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_008_pos.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_009_pos.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_010_pos.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_011_pos.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_012_pos.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_013_neg.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_014_pos.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_015_pos.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_016_pos.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_clone_livelist.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_dev_removal_condense.ksh \ functional/cli_root/zfs_destroy/zfs_destroy_dev_removal.ksh \ functional/cli_root/zfs_diff/cleanup.ksh \ functional/cli_root/zfs_diff/setup.ksh \ functional/cli_root/zfs_diff/zfs_diff_changes.ksh \ functional/cli_root/zfs_diff/zfs_diff_cliargs.ksh \ functional/cli_root/zfs_diff/zfs_diff_encrypted.ksh \ functional/cli_root/zfs_diff/zfs_diff_mangle.ksh \ functional/cli_root/zfs_diff/zfs_diff_timestamp.ksh \ functional/cli_root/zfs_diff/zfs_diff_types.ksh \ functional/cli_root/zfs_get/cleanup.ksh \ functional/cli_root/zfs_get/setup.ksh \ functional/cli_root/zfs_get/zfs_get_001_pos.ksh \ functional/cli_root/zfs_get/zfs_get_002_pos.ksh \ functional/cli_root/zfs_get/zfs_get_003_pos.ksh \ functional/cli_root/zfs_get/zfs_get_004_pos.ksh \ functional/cli_root/zfs_get/zfs_get_005_neg.ksh \ functional/cli_root/zfs_get/zfs_get_006_neg.ksh \ functional/cli_root/zfs_get/zfs_get_007_neg.ksh \ functional/cli_root/zfs_get/zfs_get_008_pos.ksh \ functional/cli_root/zfs_get/zfs_get_009_pos.ksh \ functional/cli_root/zfs_get/zfs_get_010_neg.ksh \ functional/cli_root/zfs_ids_to_path/cleanup.ksh \ functional/cli_root/zfs_ids_to_path/setup.ksh \ functional/cli_root/zfs_ids_to_path/zfs_ids_to_path_001_pos.ksh \ functional/cli_root/zfs_inherit/cleanup.ksh \ functional/cli_root/zfs_inherit/setup.ksh \ functional/cli_root/zfs_inherit/zfs_inherit_001_neg.ksh \ functional/cli_root/zfs_inherit/zfs_inherit_002_neg.ksh \ functional/cli_root/zfs_inherit/zfs_inherit_003_pos.ksh \ functional/cli_root/zfs_inherit/zfs_inherit_mountpoint.ksh \ functional/cli_root/zfs_jail/cleanup.ksh \ functional/cli_root/zfs_jail/setup.ksh \ functional/cli_root/zfs_jail/zfs_jail_001_pos.ksh \ functional/cli_root/zfs_load-key/cleanup.ksh \ functional/cli_root/zfs_load-key/setup.ksh \ functional/cli_root/zfs_load-key/zfs_load-key_all.ksh \ functional/cli_root/zfs_load-key/zfs_load-key_file.ksh \ functional/cli_root/zfs_load-key/zfs_load-key_https.ksh \ functional/cli_root/zfs_load-key/zfs_load-key.ksh \ functional/cli_root/zfs_load-key/zfs_load-key_location.ksh \ functional/cli_root/zfs_load-key/zfs_load-key_noop.ksh \ functional/cli_root/zfs_load-key/zfs_load-key_recursive.ksh \ functional/cli_root/zfs_mount/cleanup.ksh \ functional/cli_root/zfs_mount/setup.ksh \ functional/cli_root/zfs_mount/zfs_mount_001_pos.ksh \ functional/cli_root/zfs_mount/zfs_mount_002_pos.ksh \ functional/cli_root/zfs_mount/zfs_mount_003_pos.ksh \ functional/cli_root/zfs_mount/zfs_mount_004_pos.ksh \ functional/cli_root/zfs_mount/zfs_mount_005_pos.ksh \ functional/cli_root/zfs_mount/zfs_mount_006_pos.ksh \ functional/cli_root/zfs_mount/zfs_mount_007_pos.ksh \ functional/cli_root/zfs_mount/zfs_mount_008_pos.ksh \ functional/cli_root/zfs_mount/zfs_mount_009_neg.ksh \ functional/cli_root/zfs_mount/zfs_mount_010_neg.ksh \ functional/cli_root/zfs_mount/zfs_mount_011_neg.ksh \ functional/cli_root/zfs_mount/zfs_mount_012_pos.ksh \ functional/cli_root/zfs_mount/zfs_mount_013_pos.ksh \ functional/cli_root/zfs_mount/zfs_mount_014_neg.ksh \ functional/cli_root/zfs_mount/zfs_mount_all_001_pos.ksh \ functional/cli_root/zfs_mount/zfs_mount_all_fail.ksh \ functional/cli_root/zfs_mount/zfs_mount_all_mountpoints.ksh \ functional/cli_root/zfs_mount/zfs_mount_encrypted.ksh \ functional/cli_root/zfs_mount/zfs_mount_recursive.ksh \ functional/cli_root/zfs_mount/zfs_mount_remount.ksh \ functional/cli_root/zfs_mount/zfs_mount_test_race.ksh \ functional/cli_root/zfs_mount/zfs_multi_mount.ksh \ functional/cli_root/zfs_program/cleanup.ksh \ functional/cli_root/zfs_program/setup.ksh \ functional/cli_root/zfs_program/zfs_program_json.ksh \ functional/cli_root/zfs_promote/cleanup.ksh \ functional/cli_root/zfs_promote/setup.ksh \ functional/cli_root/zfs_promote/zfs_promote_001_pos.ksh \ functional/cli_root/zfs_promote/zfs_promote_002_pos.ksh \ functional/cli_root/zfs_promote/zfs_promote_003_pos.ksh \ functional/cli_root/zfs_promote/zfs_promote_004_pos.ksh \ functional/cli_root/zfs_promote/zfs_promote_005_pos.ksh \ functional/cli_root/zfs_promote/zfs_promote_006_neg.ksh \ functional/cli_root/zfs_promote/zfs_promote_007_neg.ksh \ functional/cli_root/zfs_promote/zfs_promote_008_pos.ksh \ functional/cli_root/zfs_promote/zfs_promote_encryptionroot.ksh \ functional/cli_root/zfs_property/cleanup.ksh \ functional/cli_root/zfs_property/setup.ksh \ functional/cli_root/zfs_property/zfs_written_property_001_pos.ksh \ functional/cli_root/zfs_receive/cleanup.ksh \ functional/cli_root/zfs_receive/receive-o-x_props_aliases.ksh \ functional/cli_root/zfs_receive/receive-o-x_props_override.ksh \ functional/cli_root/zfs_receive/setup.ksh \ functional/cli_root/zfs_receive/zfs_receive_001_pos.ksh \ functional/cli_root/zfs_receive/zfs_receive_002_pos.ksh \ functional/cli_root/zfs_receive/zfs_receive_003_pos.ksh \ functional/cli_root/zfs_receive/zfs_receive_004_neg.ksh \ functional/cli_root/zfs_receive/zfs_receive_005_neg.ksh \ functional/cli_root/zfs_receive/zfs_receive_006_pos.ksh \ functional/cli_root/zfs_receive/zfs_receive_007_neg.ksh \ functional/cli_root/zfs_receive/zfs_receive_008_pos.ksh \ functional/cli_root/zfs_receive/zfs_receive_009_neg.ksh \ functional/cli_root/zfs_receive/zfs_receive_010_pos.ksh \ functional/cli_root/zfs_receive/zfs_receive_011_pos.ksh \ functional/cli_root/zfs_receive/zfs_receive_012_pos.ksh \ functional/cli_root/zfs_receive/zfs_receive_013_pos.ksh \ functional/cli_root/zfs_receive/zfs_receive_014_pos.ksh \ functional/cli_root/zfs_receive/zfs_receive_015_pos.ksh \ functional/cli_root/zfs_receive/zfs_receive_016_pos.ksh \ functional/cli_root/zfs_receive/zfs_receive_-e.ksh \ functional/cli_root/zfs_receive/zfs_receive_from_encrypted.ksh \ functional/cli_root/zfs_receive/zfs_receive_from_zstd.ksh \ functional/cli_root/zfs_receive/zfs_receive_new_props.ksh \ functional/cli_root/zfs_receive/zfs_receive_raw_-d.ksh \ functional/cli_root/zfs_receive/zfs_receive_raw_incremental.ksh \ functional/cli_root/zfs_receive/zfs_receive_raw.ksh \ functional/cli_root/zfs_receive/zfs_receive_to_encrypted.ksh \ functional/cli_root/zfs_receive/zfs_receive_-wR-encrypted-mix.ksh \ functional/cli_root/zfs_receive/zfs_receive_corrective.ksh \ functional/cli_root/zfs_receive/zfs_receive_compressed_corrective.ksh \ functional/cli_root/zfs_receive/zfs_receive_large_block_corrective.ksh \ functional/cli_root/zfs_rename/cleanup.ksh \ functional/cli_root/zfs_rename/setup.ksh \ functional/cli_root/zfs_rename/zfs_rename_001_pos.ksh \ functional/cli_root/zfs_rename/zfs_rename_002_pos.ksh \ functional/cli_root/zfs_rename/zfs_rename_003_pos.ksh \ functional/cli_root/zfs_rename/zfs_rename_004_neg.ksh \ functional/cli_root/zfs_rename/zfs_rename_005_neg.ksh \ functional/cli_root/zfs_rename/zfs_rename_006_pos.ksh \ functional/cli_root/zfs_rename/zfs_rename_007_pos.ksh \ functional/cli_root/zfs_rename/zfs_rename_008_pos.ksh \ functional/cli_root/zfs_rename/zfs_rename_009_neg.ksh \ functional/cli_root/zfs_rename/zfs_rename_010_neg.ksh \ functional/cli_root/zfs_rename/zfs_rename_011_pos.ksh \ functional/cli_root/zfs_rename/zfs_rename_012_neg.ksh \ functional/cli_root/zfs_rename/zfs_rename_013_pos.ksh \ functional/cli_root/zfs_rename/zfs_rename_014_neg.ksh \ functional/cli_root/zfs_rename/zfs_rename_encrypted_child.ksh \ functional/cli_root/zfs_rename/zfs_rename_mountpoint.ksh \ functional/cli_root/zfs_rename/zfs_rename_nounmount.ksh \ functional/cli_root/zfs_rename/zfs_rename_to_encrypted.ksh \ functional/cli_root/zfs_reservation/cleanup.ksh \ functional/cli_root/zfs_reservation/setup.ksh \ functional/cli_root/zfs_reservation/zfs_reservation_001_pos.ksh \ functional/cli_root/zfs_reservation/zfs_reservation_002_pos.ksh \ functional/cli_root/zfs_rewrite/cleanup.ksh \ functional/cli_root/zfs_rewrite/setup.ksh \ functional/cli_root/zfs_rewrite/zfs_rewrite.ksh \ + functional/cli_root/zfs_rewrite/zfs_rewrite_physical.ksh \ functional/cli_root/zfs_rollback/cleanup.ksh \ functional/cli_root/zfs_rollback/setup.ksh \ functional/cli_root/zfs_rollback/zfs_rollback_001_pos.ksh \ functional/cli_root/zfs_rollback/zfs_rollback_002_pos.ksh \ functional/cli_root/zfs_rollback/zfs_rollback_003_neg.ksh \ functional/cli_root/zfs_rollback/zfs_rollback_004_neg.ksh \ functional/cli_root/zfs_send/cleanup.ksh \ functional/cli_root/zfs_send/setup.ksh \ functional/cli_root/zfs_send/zfs_send_001_pos.ksh \ functional/cli_root/zfs_send/zfs_send_002_pos.ksh \ functional/cli_root/zfs_send/zfs_send_003_pos.ksh \ functional/cli_root/zfs_send/zfs_send_004_neg.ksh \ functional/cli_root/zfs_send/zfs_send_005_pos.ksh \ functional/cli_root/zfs_send/zfs_send_006_pos.ksh \ functional/cli_root/zfs_send/zfs_send_007_pos.ksh \ functional/cli_root/zfs_send/zfs_send-b.ksh \ functional/cli_root/zfs_send/zfs_send_encrypted.ksh \ functional/cli_root/zfs_send/zfs_send_encrypted_unloaded.ksh \ functional/cli_root/zfs_send/zfs_send_raw.ksh \ functional/cli_root/zfs_send/zfs_send_skip_missing.ksh \ functional/cli_root/zfs_send/zfs_send_sparse.ksh \ functional/cli_root/zfs_set/cache_001_pos.ksh \ functional/cli_root/zfs_set/cache_002_neg.ksh \ functional/cli_root/zfs_set/canmount_001_pos.ksh \ functional/cli_root/zfs_set/canmount_002_pos.ksh \ functional/cli_root/zfs_set/canmount_003_pos.ksh \ functional/cli_root/zfs_set/canmount_004_pos.ksh \ functional/cli_root/zfs_set/checksum_001_pos.ksh \ functional/cli_root/zfs_set/cleanup.ksh \ functional/cli_root/zfs_set/compression_001_pos.ksh \ functional/cli_root/zfs_set/mountpoint_001_pos.ksh \ functional/cli_root/zfs_set/mountpoint_002_pos.ksh \ functional/cli_root/zfs_set/mountpoint_003_pos.ksh \ functional/cli_root/zfs_set/onoffs_001_pos.ksh \ functional/cli_root/zfs_set/property_alias_001_pos.ksh \ functional/cli_root/zfs_set/readonly_001_pos.ksh \ functional/cli_root/zfs_set/reservation_001_neg.ksh \ functional/cli_root/zfs_set/ro_props_001_pos.ksh \ functional/cli_root/zfs_set/setup.ksh \ functional/cli_root/zfs_set/share_mount_001_neg.ksh \ functional/cli_root/zfs_set/snapdir_001_pos.ksh \ functional/cli_root/zfs/setup.ksh \ functional/cli_root/zfs_set/user_property_001_pos.ksh \ functional/cli_root/zfs_set/user_property_002_pos.ksh \ functional/cli_root/zfs_set/user_property_003_neg.ksh \ functional/cli_root/zfs_set/user_property_004_pos.ksh \ functional/cli_root/zfs_set/version_001_neg.ksh \ functional/cli_root/zfs_set/zfs_set_001_neg.ksh \ functional/cli_root/zfs_set/zfs_set_002_neg.ksh \ functional/cli_root/zfs_set/zfs_set_003_neg.ksh \ functional/cli_root/zfs_set/zfs_set_feature_activation.ksh \ functional/cli_root/zfs_set/zfs_set_keylocation.ksh \ functional/cli_root/zfs_set/zfs_set_nomount.ksh \ functional/cli_root/zfs_share/cleanup.ksh \ functional/cli_root/zfs_share/setup.ksh \ functional/cli_root/zfs_share/zfs_share_001_pos.ksh \ functional/cli_root/zfs_share/zfs_share_002_pos.ksh \ functional/cli_root/zfs_share/zfs_share_003_pos.ksh \ functional/cli_root/zfs_share/zfs_share_004_pos.ksh \ functional/cli_root/zfs_share/zfs_share_005_pos.ksh \ functional/cli_root/zfs_share/zfs_share_006_pos.ksh \ functional/cli_root/zfs_share/zfs_share_007_neg.ksh \ functional/cli_root/zfs_share/zfs_share_008_neg.ksh \ functional/cli_root/zfs_share/zfs_share_009_neg.ksh \ functional/cli_root/zfs_share/zfs_share_010_neg.ksh \ functional/cli_root/zfs_share/zfs_share_011_pos.ksh \ functional/cli_root/zfs_share/zfs_share_012_pos.ksh \ functional/cli_root/zfs_share/zfs_share_013_pos.ksh \ functional/cli_root/zfs_share/zfs_share_concurrent_shares.ksh \ functional/cli_root/zfs_share/zfs_share_after_mount.ksh \ functional/cli_root/zfs_snapshot/cleanup.ksh \ functional/cli_root/zfs_snapshot/setup.ksh \ functional/cli_root/zfs_snapshot/zfs_snapshot_001_neg.ksh \ functional/cli_root/zfs_snapshot/zfs_snapshot_002_neg.ksh \ functional/cli_root/zfs_snapshot/zfs_snapshot_003_neg.ksh \ functional/cli_root/zfs_snapshot/zfs_snapshot_004_neg.ksh \ functional/cli_root/zfs_snapshot/zfs_snapshot_005_neg.ksh \ functional/cli_root/zfs_snapshot/zfs_snapshot_006_pos.ksh \ functional/cli_root/zfs_snapshot/zfs_snapshot_007_neg.ksh \ functional/cli_root/zfs_snapshot/zfs_snapshot_008_neg.ksh \ functional/cli_root/zfs_snapshot/zfs_snapshot_009_pos.ksh \ functional/cli_root/zfs_sysfs/cleanup.ksh \ functional/cli_root/zfs_sysfs/setup.ksh \ functional/cli_root/zfs_sysfs/zfeature_set_unsupported.ksh \ functional/cli_root/zfs_sysfs/zfs_get_unsupported.ksh \ functional/cli_root/zfs_sysfs/zfs_set_unsupported.ksh \ functional/cli_root/zfs_sysfs/zfs_sysfs_live.ksh \ functional/cli_root/zfs_sysfs/zpool_get_unsupported.ksh \ functional/cli_root/zfs_sysfs/zpool_set_unsupported.ksh \ functional/cli_root/zfs_unload-key/cleanup.ksh \ functional/cli_root/zfs_unload-key/setup.ksh \ functional/cli_root/zfs_unload-key/zfs_unload-key_all.ksh \ functional/cli_root/zfs_unload-key/zfs_unload-key.ksh \ functional/cli_root/zfs_unload-key/zfs_unload-key_recursive.ksh \ functional/cli_root/zfs_unmount/cleanup.ksh \ functional/cli_root/zfs_unmount/setup.ksh \ functional/cli_root/zfs_unmount/zfs_unmount_001_pos.ksh \ functional/cli_root/zfs_unmount/zfs_unmount_002_pos.ksh \ functional/cli_root/zfs_unmount/zfs_unmount_003_pos.ksh \ functional/cli_root/zfs_unmount/zfs_unmount_004_pos.ksh \ functional/cli_root/zfs_unmount/zfs_unmount_005_pos.ksh \ functional/cli_root/zfs_unmount/zfs_unmount_006_pos.ksh \ functional/cli_root/zfs_unmount/zfs_unmount_007_neg.ksh \ functional/cli_root/zfs_unmount/zfs_unmount_008_neg.ksh \ functional/cli_root/zfs_unmount/zfs_unmount_009_pos.ksh \ functional/cli_root/zfs_unmount/zfs_unmount_all_001_pos.ksh \ functional/cli_root/zfs_unmount/zfs_unmount_nested.ksh \ functional/cli_root/zfs_unmount/zfs_unmount_unload_keys.ksh \ functional/cli_root/zfs_unshare/cleanup.ksh \ functional/cli_root/zfs_unshare/setup.ksh \ functional/cli_root/zfs_unshare/zfs_unshare_001_pos.ksh \ functional/cli_root/zfs_unshare/zfs_unshare_002_pos.ksh \ functional/cli_root/zfs_unshare/zfs_unshare_003_pos.ksh \ functional/cli_root/zfs_unshare/zfs_unshare_004_neg.ksh \ functional/cli_root/zfs_unshare/zfs_unshare_005_neg.ksh \ functional/cli_root/zfs_unshare/zfs_unshare_006_pos.ksh \ functional/cli_root/zfs_unshare/zfs_unshare_007_pos.ksh \ functional/cli_root/zfs_unshare/zfs_unshare_008_pos.ksh \ functional/cli_root/zfs_upgrade/cleanup.ksh \ functional/cli_root/zfs_upgrade/setup.ksh \ functional/cli_root/zfs_upgrade/zfs_upgrade_001_pos.ksh \ functional/cli_root/zfs_upgrade/zfs_upgrade_002_pos.ksh \ functional/cli_root/zfs_upgrade/zfs_upgrade_003_pos.ksh \ functional/cli_root/zfs_upgrade/zfs_upgrade_004_pos.ksh \ functional/cli_root/zfs_upgrade/zfs_upgrade_005_pos.ksh \ functional/cli_root/zfs_upgrade/zfs_upgrade_006_neg.ksh \ functional/cli_root/zfs_upgrade/zfs_upgrade_007_neg.ksh \ functional/cli_root/zfs_wait/cleanup.ksh \ functional/cli_root/zfs_wait/setup.ksh \ functional/cli_root/zfs_wait/zfs_wait_deleteq.ksh \ functional/cli_root/zfs_wait/zfs_wait_getsubopt.ksh \ functional/cli_root/zfs/zfs_001_neg.ksh \ functional/cli_root/zfs/zfs_002_pos.ksh \ functional/cli_root/zfs/zfs_003_neg.ksh \ functional/cli_root/zhack/zhack_label_repair_001.ksh \ functional/cli_root/zhack/zhack_label_repair_002.ksh \ functional/cli_root/zhack/zhack_label_repair_003.ksh \ functional/cli_root/zhack/zhack_label_repair_004.ksh \ functional/cli_root/zpool_add/add_nested_replacing_spare.ksh \ functional/cli_root/zpool_add/add-o_ashift.ksh \ functional/cli_root/zpool_add/add_prop_ashift.ksh \ functional/cli_root/zpool_add/cleanup.ksh \ functional/cli_root/zpool_add/setup.ksh \ functional/cli_root/zpool_add/zpool_add_001_pos.ksh \ functional/cli_root/zpool_add/zpool_add_002_pos.ksh \ functional/cli_root/zpool_add/zpool_add_003_pos.ksh \ functional/cli_root/zpool_add/zpool_add_004_pos.ksh \ functional/cli_root/zpool_add/zpool_add_005_pos.ksh \ functional/cli_root/zpool_add/zpool_add_006_pos.ksh \ functional/cli_root/zpool_add/zpool_add_007_neg.ksh \ functional/cli_root/zpool_add/zpool_add_008_neg.ksh \ functional/cli_root/zpool_add/zpool_add_009_neg.ksh \ functional/cli_root/zpool_add/zpool_add_010_pos.ksh \ functional/cli_root/zpool_add/zpool_add_dryrun_output.ksh \ functional/cli_root/zpool_attach/attach-o_ashift.ksh \ functional/cli_root/zpool_attach/cleanup.ksh \ functional/cli_root/zpool_attach/setup.ksh \ functional/cli_root/zpool_attach/zpool_attach_001_neg.ksh \ functional/cli_root/zpool/cleanup.ksh \ functional/cli_root/zpool_clear/cleanup.ksh \ functional/cli_root/zpool_clear/setup.ksh \ functional/cli_root/zpool_clear/zpool_clear_001_pos.ksh \ functional/cli_root/zpool_clear/zpool_clear_002_neg.ksh \ functional/cli_root/zpool_clear/zpool_clear_003_neg.ksh \ functional/cli_root/zpool_clear/zpool_clear_readonly.ksh \ functional/cli_root/zpool_create/cleanup.ksh \ functional/cli_root/zpool_create/create-o_ashift.ksh \ functional/cli_root/zpool_create/setup.ksh \ functional/cli_root/zpool_create/zpool_create_001_pos.ksh \ functional/cli_root/zpool_create/zpool_create_002_pos.ksh \ functional/cli_root/zpool_create/zpool_create_003_pos.ksh \ functional/cli_root/zpool_create/zpool_create_004_pos.ksh \ functional/cli_root/zpool_create/zpool_create_005_pos.ksh \ functional/cli_root/zpool_create/zpool_create_006_pos.ksh \ functional/cli_root/zpool_create/zpool_create_007_neg.ksh \ functional/cli_root/zpool_create/zpool_create_008_pos.ksh \ functional/cli_root/zpool_create/zpool_create_009_neg.ksh \ functional/cli_root/zpool_create/zpool_create_010_neg.ksh \ functional/cli_root/zpool_create/zpool_create_011_neg.ksh \ functional/cli_root/zpool_create/zpool_create_012_neg.ksh \ functional/cli_root/zpool_create/zpool_create_014_neg.ksh \ functional/cli_root/zpool_create/zpool_create_015_neg.ksh \ functional/cli_root/zpool_create/zpool_create_016_pos.ksh \ functional/cli_root/zpool_create/zpool_create_017_neg.ksh \ functional/cli_root/zpool_create/zpool_create_018_pos.ksh \ functional/cli_root/zpool_create/zpool_create_019_pos.ksh \ functional/cli_root/zpool_create/zpool_create_020_pos.ksh \ functional/cli_root/zpool_create/zpool_create_021_pos.ksh \ functional/cli_root/zpool_create/zpool_create_022_pos.ksh \ functional/cli_root/zpool_create/zpool_create_023_neg.ksh \ functional/cli_root/zpool_create/zpool_create_024_pos.ksh \ functional/cli_root/zpool_create/zpool_create_crypt_combos.ksh \ functional/cli_root/zpool_create/zpool_create_draid_001_pos.ksh \ functional/cli_root/zpool_create/zpool_create_draid_002_pos.ksh \ functional/cli_root/zpool_create/zpool_create_draid_003_pos.ksh \ functional/cli_root/zpool_create/zpool_create_draid_004_pos.ksh \ functional/cli_root/zpool_create/zpool_create_dryrun_output.ksh \ functional/cli_root/zpool_create/zpool_create_encrypted.ksh \ functional/cli_root/zpool_create/zpool_create_features_001_pos.ksh \ functional/cli_root/zpool_create/zpool_create_features_002_pos.ksh \ functional/cli_root/zpool_create/zpool_create_features_003_pos.ksh \ functional/cli_root/zpool_create/zpool_create_features_004_neg.ksh \ functional/cli_root/zpool_create/zpool_create_features_005_pos.ksh \ functional/cli_root/zpool_create/zpool_create_features_006_pos.ksh \ functional/cli_root/zpool_create/zpool_create_features_007_pos.ksh \ functional/cli_root/zpool_create/zpool_create_features_008_pos.ksh \ functional/cli_root/zpool_create/zpool_create_features_009_pos.ksh \ functional/cli_root/zpool_create/zpool_create_tempname.ksh \ functional/cli_root/zpool_destroy/zpool_destroy_001_pos.ksh \ functional/cli_root/zpool_destroy/zpool_destroy_002_pos.ksh \ functional/cli_root/zpool_destroy/zpool_destroy_003_neg.ksh \ functional/cli_root/zpool_detach/cleanup.ksh \ functional/cli_root/zpool_detach/setup.ksh \ functional/cli_root/zpool_detach/zpool_detach_001_neg.ksh \ functional/cli_root/zpool_events/cleanup.ksh \ functional/cli_root/zpool_events/setup.ksh \ functional/cli_root/zpool_events/zpool_events_clear.ksh \ functional/cli_root/zpool_events/zpool_events_clear_retained.ksh \ functional/cli_root/zpool_events/zpool_events_cliargs.ksh \ functional/cli_root/zpool_events/zpool_events_duplicates.ksh \ functional/cli_root/zpool_events/zpool_events_errors.ksh \ functional/cli_root/zpool_events/zpool_events_follow.ksh \ functional/cli_root/zpool_events/zpool_events_poolname.ksh \ functional/cli_root/zpool_events/zpool_events_scrub_txg_continue_from_last.ksh \ functional/cli_root/zpool_expand/cleanup.ksh \ functional/cli_root/zpool_expand/setup.ksh \ functional/cli_root/zpool_expand/zpool_expand_001_pos.ksh \ functional/cli_root/zpool_expand/zpool_expand_002_pos.ksh \ functional/cli_root/zpool_expand/zpool_expand_003_neg.ksh \ functional/cli_root/zpool_expand/zpool_expand_004_pos.ksh \ functional/cli_root/zpool_expand/zpool_expand_005_pos.ksh \ functional/cli_root/zpool_export/cleanup.ksh \ functional/cli_root/zpool_export/setup.ksh \ functional/cli_root/zpool_export/zpool_export_001_pos.ksh \ functional/cli_root/zpool_export/zpool_export_002_pos.ksh \ functional/cli_root/zpool_export/zpool_export_003_neg.ksh \ functional/cli_root/zpool_export/zpool_export_004_pos.ksh \ functional/cli_root/zpool_export/zpool_export_parallel_admin.ksh \ functional/cli_root/zpool_export/zpool_export_parallel_pos.ksh \ functional/cli_root/zpool_get/cleanup.ksh \ functional/cli_root/zpool_get/setup.ksh \ functional/cli_root/zpool_get/vdev_get_001_pos.ksh \ functional/cli_root/zpool_get/vdev_get_all.ksh \ functional/cli_root/zpool_get/zpool_get_001_pos.ksh \ functional/cli_root/zpool_get/zpool_get_002_pos.ksh \ functional/cli_root/zpool_get/zpool_get_003_pos.ksh \ functional/cli_root/zpool_get/zpool_get_004_neg.ksh \ functional/cli_root/zpool_get/zpool_get_005_pos.ksh \ functional/cli_root/zpool_history/cleanup.ksh \ functional/cli_root/zpool_history/setup.ksh \ functional/cli_root/zpool_history/zpool_history_001_neg.ksh \ functional/cli_root/zpool_history/zpool_history_002_pos.ksh \ functional/cli_root/zpool_import/cleanup.ksh \ functional/cli_root/zpool_import/import_cachefile_device_added.ksh \ functional/cli_root/zpool_import/import_cachefile_device_removed.ksh \ functional/cli_root/zpool_import/import_cachefile_device_replaced.ksh \ functional/cli_root/zpool_import/import_cachefile_mirror_attached.ksh \ functional/cli_root/zpool_import/import_cachefile_mirror_detached.ksh \ functional/cli_root/zpool_import/import_cachefile_paths_changed.ksh \ functional/cli_root/zpool_import/import_cachefile_shared_device.ksh \ functional/cli_root/zpool_import/import_devices_missing.ksh \ functional/cli_root/zpool_import/import_log_missing.ksh \ functional/cli_root/zpool_import/import_paths_changed.ksh \ functional/cli_root/zpool_import/import_rewind_config_changed.ksh \ functional/cli_root/zpool_import/import_rewind_device_replaced.ksh \ functional/cli_root/zpool_import/setup.ksh \ functional/cli_root/zpool_import/zpool_import_001_pos.ksh \ functional/cli_root/zpool_import/zpool_import_002_pos.ksh \ functional/cli_root/zpool_import/zpool_import_003_pos.ksh \ functional/cli_root/zpool_import/zpool_import_004_pos.ksh \ functional/cli_root/zpool_import/zpool_import_005_pos.ksh \ functional/cli_root/zpool_import/zpool_import_006_pos.ksh \ functional/cli_root/zpool_import/zpool_import_007_pos.ksh \ functional/cli_root/zpool_import/zpool_import_008_pos.ksh \ functional/cli_root/zpool_import/zpool_import_009_neg.ksh \ functional/cli_root/zpool_import/zpool_import_010_pos.ksh \ functional/cli_root/zpool_import/zpool_import_011_neg.ksh \ functional/cli_root/zpool_import/zpool_import_012_pos.ksh \ functional/cli_root/zpool_import/zpool_import_013_neg.ksh \ functional/cli_root/zpool_import/zpool_import_014_pos.ksh \ functional/cli_root/zpool_import/zpool_import_015_pos.ksh \ functional/cli_root/zpool_import/zpool_import_016_pos.ksh \ functional/cli_root/zpool_import/zpool_import_017_pos.ksh \ functional/cli_root/zpool_import/zpool_import_all_001_pos.ksh \ functional/cli_root/zpool_import/zpool_import_encrypted.ksh \ functional/cli_root/zpool_import/zpool_import_encrypted_load.ksh \ functional/cli_root/zpool_import/zpool_import_errata3.ksh \ functional/cli_root/zpool_import/zpool_import_errata4.ksh \ functional/cli_root/zpool_import/zpool_import_features_001_pos.ksh \ functional/cli_root/zpool_import/zpool_import_features_002_neg.ksh \ functional/cli_root/zpool_import/zpool_import_features_003_pos.ksh \ functional/cli_root/zpool_import/zpool_import_hostid_changed.ksh \ functional/cli_root/zpool_import/zpool_import_hostid_changed_unclean_export.ksh \ functional/cli_root/zpool_import/zpool_import_hostid_changed_cachefile.ksh \ functional/cli_root/zpool_import/zpool_import_hostid_changed_cachefile_unclean_export.ksh \ functional/cli_root/zpool_import/zpool_import_missing_001_pos.ksh \ functional/cli_root/zpool_import/zpool_import_missing_002_pos.ksh \ functional/cli_root/zpool_import/zpool_import_missing_003_pos.ksh \ functional/cli_root/zpool_import/zpool_import_rename_001_pos.ksh \ functional/cli_root/zpool_import/zpool_import_status.ksh \ functional/cli_root/zpool_import/zpool_import_parallel_admin.ksh \ functional/cli_root/zpool_import/zpool_import_parallel_neg.ksh \ functional/cli_root/zpool_import/zpool_import_parallel_pos.ksh \ functional/cli_root/zpool_initialize/cleanup.ksh \ functional/cli_root/zpool_initialize/zpool_initialize_attach_detach_add_remove.ksh \ functional/cli_root/zpool_initialize/zpool_initialize_fault_export_import_online.ksh \ functional/cli_root/zpool_initialize/zpool_initialize_import_export.ksh \ functional/cli_root/zpool_initialize/zpool_initialize_multiple_pools.ksh \ functional/cli_root/zpool_initialize/zpool_initialize_offline_export_import_online.ksh \ functional/cli_root/zpool_initialize/zpool_initialize_online_offline.ksh \ functional/cli_root/zpool_initialize/zpool_initialize_split.ksh \ functional/cli_root/zpool_initialize/zpool_initialize_start_and_cancel_neg.ksh \ functional/cli_root/zpool_initialize/zpool_initialize_start_and_cancel_pos.ksh \ functional/cli_root/zpool_initialize/zpool_initialize_suspend_resume.ksh \ functional/cli_root/zpool_initialize/zpool_initialize_uninit.ksh \ functional/cli_root/zpool_initialize/zpool_initialize_unsupported_vdevs.ksh \ functional/cli_root/zpool_initialize/zpool_initialize_verify_checksums.ksh \ functional/cli_root/zpool_initialize/zpool_initialize_verify_initialized.ksh \ functional/cli_root/zpool_labelclear/zpool_labelclear_active.ksh \ functional/cli_root/zpool_labelclear/zpool_labelclear_exported.ksh \ functional/cli_root/zpool_labelclear/zpool_labelclear_removed.ksh \ functional/cli_root/zpool_labelclear/zpool_labelclear_valid.ksh \ functional/cli_root/zpool_offline/cleanup.ksh \ functional/cli_root/zpool_offline/setup.ksh \ functional/cli_root/zpool_offline/zpool_offline_001_pos.ksh \ functional/cli_root/zpool_offline/zpool_offline_002_neg.ksh \ functional/cli_root/zpool_offline/zpool_offline_003_pos.ksh \ functional/cli_root/zpool_online/cleanup.ksh \ functional/cli_root/zpool_online/setup.ksh \ functional/cli_root/zpool_online/zpool_online_001_pos.ksh \ functional/cli_root/zpool_online/zpool_online_002_neg.ksh \ functional/cli_root/zpool_prefetch/cleanup.ksh \ functional/cli_root/zpool_prefetch/setup.ksh \ functional/cli_root/zpool_prefetch/zpool_prefetch_001_pos.ksh \ functional/cli_root/zpool_reguid/cleanup.ksh \ functional/cli_root/zpool_reguid/setup.ksh \ functional/cli_root/zpool_reguid/zpool_reguid_001_pos.ksh \ functional/cli_root/zpool_reguid/zpool_reguid_002_neg.ksh \ functional/cli_root/zpool_remove/cleanup.ksh \ functional/cli_root/zpool_remove/setup.ksh \ functional/cli_root/zpool_remove/zpool_remove_001_neg.ksh \ functional/cli_root/zpool_remove/zpool_remove_002_pos.ksh \ functional/cli_root/zpool_remove/zpool_remove_003_pos.ksh \ functional/cli_root/zpool_reopen/cleanup.ksh \ functional/cli_root/zpool_reopen/setup.ksh \ functional/cli_root/zpool_reopen/zpool_reopen_001_pos.ksh \ functional/cli_root/zpool_reopen/zpool_reopen_002_pos.ksh \ functional/cli_root/zpool_reopen/zpool_reopen_003_pos.ksh \ functional/cli_root/zpool_reopen/zpool_reopen_004_pos.ksh \ functional/cli_root/zpool_reopen/zpool_reopen_005_pos.ksh \ functional/cli_root/zpool_reopen/zpool_reopen_006_neg.ksh \ functional/cli_root/zpool_reopen/zpool_reopen_007_pos.ksh \ functional/cli_root/zpool_replace/cleanup.ksh \ functional/cli_root/zpool_replace/replace-o_ashift.ksh \ functional/cli_root/zpool_replace/replace_prop_ashift.ksh \ functional/cli_root/zpool_replace/setup.ksh \ functional/cli_root/zpool_replace/zpool_replace_001_neg.ksh \ functional/cli_root/zpool_resilver/cleanup.ksh \ functional/cli_root/zpool_resilver/setup.ksh \ functional/cli_root/zpool_resilver/zpool_resilver_bad_args.ksh \ functional/cli_root/zpool_resilver/zpool_resilver_restart.ksh \ functional/cli_root/zpool_resilver/zpool_resilver_concurrent.ksh \ functional/cli_root/zpool_scrub/cleanup.ksh \ functional/cli_root/zpool_scrub/setup.ksh \ functional/cli_root/zpool_scrub/zpool_scrub_001_neg.ksh \ functional/cli_root/zpool_scrub/zpool_scrub_002_pos.ksh \ functional/cli_root/zpool_scrub/zpool_scrub_003_pos.ksh \ functional/cli_root/zpool_scrub/zpool_scrub_004_pos.ksh \ functional/cli_root/zpool_scrub/zpool_scrub_005_pos.ksh \ functional/cli_root/zpool_scrub/zpool_scrub_encrypted_unloaded.ksh \ functional/cli_root/zpool_scrub/zpool_scrub_multiple_copies.ksh \ functional/cli_root/zpool_scrub/zpool_scrub_multiple_pools.ksh \ functional/cli_root/zpool_scrub/zpool_scrub_offline_device.ksh \ functional/cli_root/zpool_scrub/zpool_scrub_print_repairing.ksh \ functional/cli_root/zpool_scrub/zpool_scrub_txg_continue_from_last.ksh \ functional/cli_root/zpool_scrub/zpool_scrub_date_range_001.ksh \ functional/cli_root/zpool_scrub/zpool_error_scrub_001_pos.ksh \ functional/cli_root/zpool_scrub/zpool_error_scrub_002_pos.ksh \ functional/cli_root/zpool_scrub/zpool_error_scrub_003_pos.ksh \ functional/cli_root/zpool_scrub/zpool_error_scrub_004_pos.ksh \ functional/cli_root/zpool_set/cleanup.ksh \ functional/cli_root/zpool_set/setup.ksh \ functional/cli_root/zpool/setup.ksh \ functional/cli_root/zpool_set/vdev_set_001_pos.ksh \ functional/cli_root/zpool_set/zpool_set_common.kshlib \ functional/cli_root/zpool_set/zpool_set_001_pos.ksh \ functional/cli_root/zpool_set/zpool_set_002_neg.ksh \ functional/cli_root/zpool_set/zpool_set_003_neg.ksh \ functional/cli_root/zpool_set/zpool_set_ashift.ksh \ functional/cli_root/zpool_set/user_property_001_pos.ksh \ functional/cli_root/zpool_set/user_property_002_neg.ksh \ functional/cli_root/zpool_set/zpool_set_features.ksh \ functional/cli_root/zpool_set/zpool_set_clear_userprop.ksh \ functional/cli_root/zpool_split/cleanup.ksh \ functional/cli_root/zpool_split/setup.ksh \ functional/cli_root/zpool_split/zpool_split_cliargs.ksh \ functional/cli_root/zpool_split/zpool_split_devices.ksh \ functional/cli_root/zpool_split/zpool_split_dryrun_output.ksh \ functional/cli_root/zpool_split/zpool_split_encryption.ksh \ functional/cli_root/zpool_split/zpool_split_indirect.ksh \ functional/cli_root/zpool_split/zpool_split_props.ksh \ functional/cli_root/zpool_split/zpool_split_resilver.ksh \ functional/cli_root/zpool_split/zpool_split_vdevs.ksh \ functional/cli_root/zpool_split/zpool_split_wholedisk.ksh \ functional/cli_root/zpool_status/cleanup.ksh \ functional/cli_root/zpool_status/setup.ksh \ functional/cli_root/zpool_status/zpool_status_001_pos.ksh \ functional/cli_root/zpool_status/zpool_status_002_pos.ksh \ functional/cli_root/zpool_status/zpool_status_003_pos.ksh \ functional/cli_root/zpool_status/zpool_status_004_pos.ksh \ functional/cli_root/zpool_status/zpool_status_005_pos.ksh \ functional/cli_root/zpool_status/zpool_status_006_pos.ksh \ functional/cli_root/zpool_status/zpool_status_007_pos.ksh \ functional/cli_root/zpool_status/zpool_status_008_pos.ksh \ functional/cli_root/zpool_status/zpool_status_features_001_pos.ksh \ functional/cli_root/zpool_sync/cleanup.ksh \ functional/cli_root/zpool_sync/setup.ksh \ functional/cli_root/zpool_sync/zpool_sync_001_pos.ksh \ functional/cli_root/zpool_sync/zpool_sync_002_neg.ksh \ functional/cli_root/zpool_trim/cleanup.ksh \ functional/cli_root/zpool_trim/setup.ksh \ functional/cli_root/zpool_trim/zpool_trim_attach_detach_add_remove.ksh \ functional/cli_root/zpool_trim/zpool_trim_fault_export_import_online.ksh \ functional/cli_root/zpool_trim/zpool_trim_import_export.ksh \ functional/cli_root/zpool_trim/zpool_trim_multiple.ksh \ functional/cli_root/zpool_trim/zpool_trim_multiple_pools.ksh \ functional/cli_root/zpool_trim/zpool_trim_neg.ksh \ functional/cli_root/zpool_trim/zpool_trim_offline_export_import_online.ksh \ functional/cli_root/zpool_trim/zpool_trim_online_offline.ksh \ functional/cli_root/zpool_trim/zpool_trim_partial.ksh \ functional/cli_root/zpool_trim/zpool_trim_rate.ksh \ functional/cli_root/zpool_trim/zpool_trim_rate_neg.ksh \ functional/cli_root/zpool_trim/zpool_trim_secure.ksh \ functional/cli_root/zpool_trim/zpool_trim_split.ksh \ functional/cli_root/zpool_trim/zpool_trim_start_and_cancel_neg.ksh \ functional/cli_root/zpool_trim/zpool_trim_start_and_cancel_pos.ksh \ functional/cli_root/zpool_trim/zpool_trim_suspend_resume.ksh \ functional/cli_root/zpool_trim/zpool_trim_unsupported_vdevs.ksh \ functional/cli_root/zpool_trim/zpool_trim_verify_checksums.ksh \ functional/cli_root/zpool_trim/zpool_trim_verify_trimmed.ksh \ functional/cli_root/zpool_upgrade/cleanup.ksh \ functional/cli_root/zpool_upgrade/setup.ksh \ functional/cli_root/zpool_upgrade/zpool_upgrade_001_pos.ksh \ functional/cli_root/zpool_upgrade/zpool_upgrade_002_pos.ksh \ functional/cli_root/zpool_upgrade/zpool_upgrade_003_pos.ksh \ functional/cli_root/zpool_upgrade/zpool_upgrade_004_pos.ksh \ functional/cli_root/zpool_upgrade/zpool_upgrade_005_neg.ksh \ functional/cli_root/zpool_upgrade/zpool_upgrade_006_neg.ksh \ functional/cli_root/zpool_upgrade/zpool_upgrade_007_pos.ksh \ functional/cli_root/zpool_upgrade/zpool_upgrade_008_pos.ksh \ functional/cli_root/zpool_upgrade/zpool_upgrade_009_neg.ksh \ functional/cli_root/zpool_upgrade/zpool_upgrade_features_001_pos.ksh \ functional/cli_root/zpool_wait/cleanup.ksh \ functional/cli_root/zpool_wait/scan/cleanup.ksh \ functional/cli_root/zpool_wait/scan/setup.ksh \ functional/cli_root/zpool_wait/scan/zpool_wait_rebuild.ksh \ functional/cli_root/zpool_wait/scan/zpool_wait_replace_cancel.ksh \ functional/cli_root/zpool_wait/scan/zpool_wait_replace.ksh \ functional/cli_root/zpool_wait/scan/zpool_wait_resilver.ksh \ functional/cli_root/zpool_wait/scan/zpool_wait_scrub_basic.ksh \ functional/cli_root/zpool_wait/scan/zpool_wait_scrub_cancel.ksh \ functional/cli_root/zpool_wait/scan/zpool_wait_scrub_flag.ksh \ functional/cli_root/zpool_wait/setup.ksh \ functional/cli_root/zpool_wait/zpool_wait_discard.ksh \ functional/cli_root/zpool_wait/zpool_wait_freeing.ksh \ functional/cli_root/zpool_wait/zpool_wait_initialize_basic.ksh \ functional/cli_root/zpool_wait/zpool_wait_initialize_cancel.ksh \ functional/cli_root/zpool_wait/zpool_wait_initialize_flag.ksh \ functional/cli_root/zpool_wait/zpool_wait_multiple.ksh \ functional/cli_root/zpool_wait/zpool_wait_no_activity.ksh \ functional/cli_root/zpool_wait/zpool_wait_remove_cancel.ksh \ functional/cli_root/zpool_wait/zpool_wait_remove.ksh \ functional/cli_root/zpool_wait/zpool_wait_trim_basic.ksh \ functional/cli_root/zpool_wait/zpool_wait_trim_cancel.ksh \ functional/cli_root/zpool_wait/zpool_wait_trim_flag.ksh \ functional/cli_root/zpool_wait/zpool_wait_usage.ksh \ functional/cli_root/zpool/zpool_001_neg.ksh \ functional/cli_root/zpool/zpool_002_pos.ksh \ functional/cli_root/zpool/zpool_003_pos.ksh \ functional/cli_root/zpool/zpool_colors.ksh \ functional/cli_user/misc/arcstat_001_pos.ksh \ functional/cli_user/misc/arc_summary_001_pos.ksh \ functional/cli_user/misc/arc_summary_002_neg.ksh \ functional/cli_user/misc/zilstat_001_pos.ksh \ functional/cli_user/misc/cleanup.ksh \ functional/cli_user/misc/setup.ksh \ functional/cli_user/misc/zdb_001_neg.ksh \ functional/cli_user/misc/zfs_001_neg.ksh \ functional/cli_user/misc/zfs_allow_001_neg.ksh \ functional/cli_user/misc/zfs_clone_001_neg.ksh \ functional/cli_user/misc/zfs_create_001_neg.ksh \ functional/cli_user/misc/zfs_destroy_001_neg.ksh \ functional/cli_user/misc/zfs_get_001_neg.ksh \ functional/cli_user/misc/zfs_inherit_001_neg.ksh \ functional/cli_user/misc/zfs_mount_001_neg.ksh \ functional/cli_user/misc/zfs_promote_001_neg.ksh \ functional/cli_user/misc/zfs_receive_001_neg.ksh \ functional/cli_user/misc/zfs_rename_001_neg.ksh \ functional/cli_user/misc/zfs_rollback_001_neg.ksh \ functional/cli_user/misc/zfs_send_001_neg.ksh \ functional/cli_user/misc/zfs_set_001_neg.ksh \ functional/cli_user/misc/zfs_share_001_neg.ksh \ functional/cli_user/misc/zfs_snapshot_001_neg.ksh \ functional/cli_user/misc/zfs_unallow_001_neg.ksh \ functional/cli_user/misc/zfs_unmount_001_neg.ksh \ functional/cli_user/misc/zfs_unshare_001_neg.ksh \ functional/cli_user/misc/zfs_upgrade_001_neg.ksh \ functional/cli_user/misc/zpool_001_neg.ksh \ functional/cli_user/misc/zpool_add_001_neg.ksh \ functional/cli_user/misc/zpool_attach_001_neg.ksh \ functional/cli_user/misc/zpool_clear_001_neg.ksh \ functional/cli_user/misc/zpool_create_001_neg.ksh \ functional/cli_user/misc/zpool_destroy_001_neg.ksh \ functional/cli_user/misc/zpool_detach_001_neg.ksh \ functional/cli_user/misc/zpool_export_001_neg.ksh \ functional/cli_user/misc/zpool_get_001_neg.ksh \ functional/cli_user/misc/zpool_history_001_neg.ksh \ functional/cli_user/misc/zpool_import_001_neg.ksh \ functional/cli_user/misc/zpool_import_002_neg.ksh \ functional/cli_user/misc/zpool_offline_001_neg.ksh \ functional/cli_user/misc/zpool_online_001_neg.ksh \ functional/cli_user/misc/zpool_remove_001_neg.ksh \ functional/cli_user/misc/zpool_replace_001_neg.ksh \ functional/cli_user/misc/zpool_scrub_001_neg.ksh \ functional/cli_user/misc/zpool_set_001_neg.ksh \ functional/cli_user/misc/zpool_status_001_neg.ksh \ functional/cli_user/misc/zpool_upgrade_001_neg.ksh \ functional/cli_user/misc/zpool_wait_privilege.ksh \ functional/cli_user/zfs_list/cleanup.ksh \ functional/cli_user/zfs_list/setup.ksh \ functional/cli_user/zfs_list/zfs_list_001_pos.ksh \ functional/cli_user/zfs_list/zfs_list_002_pos.ksh \ functional/cli_user/zfs_list/zfs_list_003_pos.ksh \ functional/cli_user/zfs_list/zfs_list_004_neg.ksh \ functional/cli_user/zfs_list/zfs_list_005_neg.ksh \ functional/cli_user/zfs_list/zfs_list_007_pos.ksh \ functional/cli_user/zfs_list/zfs_list_008_neg.ksh \ functional/cli_user/zpool_iostat/cleanup.ksh \ functional/cli_user/zpool_iostat/setup.ksh \ functional/cli_user/zpool_iostat/zpool_iostat_001_neg.ksh \ functional/cli_user/zpool_iostat/zpool_iostat_002_pos.ksh \ functional/cli_user/zpool_iostat/zpool_iostat_003_neg.ksh \ functional/cli_user/zpool_iostat/zpool_iostat_004_pos.ksh \ functional/cli_user/zpool_iostat/zpool_iostat_005_pos.ksh \ functional/cli_user/zpool_iostat/zpool_iostat_-c_disable.ksh \ functional/cli_user/zpool_iostat/zpool_iostat_-c_homedir.ksh \ functional/cli_user/zpool_iostat/zpool_iostat_-c_searchpath.ksh \ functional/cli_user/zpool_list/cleanup.ksh \ functional/cli_user/zpool_list/setup.ksh \ functional/cli_user/zpool_list/zpool_list_001_pos.ksh \ functional/cli_user/zpool_list/zpool_list_002_neg.ksh \ functional/cli_user/zpool_status/cleanup.ksh \ functional/cli_user/zpool_status/setup.ksh \ functional/cli_user/zpool_status/zpool_status_003_pos.ksh \ functional/cli_user/zpool_status/zpool_status_-c_disable.ksh \ functional/cli_user/zpool_status/zpool_status_-c_homedir.ksh \ functional/cli_user/zpool_status/zpool_status_-c_searchpath.ksh \ functional/compression/cleanup.ksh \ functional/compression/compress_001_pos.ksh \ functional/compression/compress_002_pos.ksh \ functional/compression/compress_003_pos.ksh \ functional/compression/compress_004_pos.ksh \ functional/compression/compress_zstd_bswap.ksh \ functional/compression/l2arc_compressed_arc_disabled.ksh \ functional/compression/l2arc_compressed_arc.ksh \ functional/compression/l2arc_encrypted.ksh \ functional/compression/l2arc_encrypted_no_compressed_arc.ksh \ functional/compression/setup.ksh \ functional/cp_files/cleanup.ksh \ functional/cp_files/cp_files_001_pos.ksh \ functional/cp_files/cp_files_002_pos.ksh \ functional/cp_files/cp_stress.ksh \ functional/cp_files/setup.ksh \ functional/crtime/cleanup.ksh \ functional/crtime/crtime_001_pos.ksh \ functional/crtime/setup.ksh \ functional/crypto/icp_aes_ccm.ksh \ functional/crypto/icp_aes_gcm.ksh \ functional/ctime/cleanup.ksh \ functional/ctime/ctime_001_pos.ksh \ functional/ctime/setup.ksh \ functional/deadman/deadman_ratelimit.ksh \ functional/deadman/deadman_sync.ksh \ functional/deadman/deadman_zio.ksh \ functional/dedup/cleanup.ksh \ functional/dedup/setup.ksh \ functional/dedup/dedup_fdt_create.ksh \ functional/dedup/dedup_fdt_import.ksh \ functional/dedup/dedup_fdt_pacing.ksh \ functional/dedup/dedup_legacy_create.ksh \ functional/dedup/dedup_legacy_import.ksh \ functional/dedup/dedup_legacy_fdt_upgrade.ksh \ functional/dedup/dedup_legacy_fdt_mixed.ksh \ functional/dedup/dedup_prune.ksh \ functional/dedup/dedup_quota.ksh \ functional/dedup/dedup_zap_shrink.ksh \ functional/delegate/cleanup.ksh \ functional/delegate/setup.ksh \ functional/delegate/zfs_allow_001_pos.ksh \ functional/delegate/zfs_allow_002_pos.ksh \ functional/delegate/zfs_allow_003_pos.ksh \ functional/delegate/zfs_allow_004_pos.ksh \ functional/delegate/zfs_allow_005_pos.ksh \ functional/delegate/zfs_allow_006_pos.ksh \ functional/delegate/zfs_allow_007_pos.ksh \ functional/delegate/zfs_allow_008_pos.ksh \ functional/delegate/zfs_allow_009_neg.ksh \ functional/delegate/zfs_allow_010_pos.ksh \ functional/delegate/zfs_allow_011_neg.ksh \ functional/delegate/zfs_allow_012_neg.ksh \ functional/delegate/zfs_unallow_001_pos.ksh \ functional/delegate/zfs_unallow_002_pos.ksh \ functional/delegate/zfs_unallow_003_pos.ksh \ functional/delegate/zfs_unallow_004_pos.ksh \ functional/delegate/zfs_unallow_005_pos.ksh \ functional/delegate/zfs_unallow_006_pos.ksh \ functional/delegate/zfs_unallow_007_neg.ksh \ functional/delegate/zfs_unallow_008_neg.ksh \ functional/devices/cleanup.ksh \ functional/devices/devices_001_pos.ksh \ functional/devices/devices_002_neg.ksh \ functional/devices/devices_003_pos.ksh \ functional/devices/setup.ksh \ functional/direct/dio_aligned_block.ksh \ functional/direct/dio_async_always.ksh \ functional/direct/dio_async_fio_ioengines.ksh \ functional/direct/dio_compression.ksh \ functional/direct/dio_dedup.ksh \ functional/direct/dio_encryption.ksh \ functional/direct/dio_grow_block.ksh \ functional/direct/dio_loopback_dev.ksh \ functional/direct/dio_max_recordsize.ksh \ functional/direct/dio_mixed.ksh \ functional/direct/dio_mmap.ksh \ functional/direct/dio_overwrites.ksh \ functional/direct/dio_property.ksh \ functional/direct/dio_random.ksh \ functional/direct/dio_read_verify.ksh \ functional/direct/dio_recordsize.ksh \ functional/direct/dio_unaligned_block.ksh \ functional/direct/dio_unaligned_filesize.ksh \ functional/direct/dio_write_verify.ksh \ functional/direct/dio_write_stable_pages.ksh \ functional/direct/setup.ksh \ functional/direct/cleanup.ksh \ functional/dos_attributes/cleanup.ksh \ functional/dos_attributes/read_dos_attrs_001.ksh \ functional/dos_attributes/setup.ksh \ functional/dos_attributes/write_dos_attrs_001.ksh \ functional/events/cleanup.ksh \ functional/events/events_001_pos.ksh \ functional/events/events_002_pos.ksh \ functional/events/setup.ksh \ functional/events/zed_cksum_config.ksh \ functional/events/zed_cksum_reported.ksh \ functional/events/zed_diagnose_multiple.ksh \ functional/events/zed_fd_spill.ksh \ functional/events/zed_io_config.ksh \ functional/events/zed_rc_filter.ksh \ functional/events/zed_slow_io.ksh \ functional/events/zed_slow_io_many_vdevs.ksh \ functional/exec/cleanup.ksh \ functional/exec/exec_001_pos.ksh \ functional/exec/exec_002_neg.ksh \ functional/exec/setup.ksh \ functional/fadvise/cleanup.ksh \ functional/fadvise/fadvise_willneed.ksh \ functional/fadvise/setup.ksh \ functional/failmode/cleanup.ksh \ functional/failmode/failmode_dmu_tx_wait.ksh \ functional/failmode/failmode_dmu_tx_continue.ksh \ functional/failmode/setup.ksh \ functional/fallocate/cleanup.ksh \ functional/fallocate/fallocate_prealloc.ksh \ functional/fallocate/fallocate_punch-hole.ksh \ functional/fallocate/fallocate_zero-range.ksh \ functional/fallocate/setup.ksh \ functional/fault/auto_offline_001_pos.ksh \ functional/fault/auto_online_001_pos.ksh \ functional/fault/auto_online_002_pos.ksh \ functional/fault/auto_replace_001_pos.ksh \ functional/fault/auto_replace_002_pos.ksh \ functional/fault/auto_spare_001_pos.ksh \ functional/fault/auto_spare_002_pos.ksh \ functional/fault/auto_spare_ashift.ksh \ functional/fault/auto_spare_double.ksh \ functional/fault/auto_spare_multiple.ksh \ functional/fault/auto_spare_shared.ksh \ functional/fault/cleanup.ksh \ functional/fault/decompress_fault.ksh \ functional/fault/decrypt_fault.ksh \ functional/fault/fault_limits.ksh \ functional/fault/scrub_after_resilver.ksh \ functional/fault/suspend_on_probe_errors.ksh \ functional/fault/suspend_resume_single.ksh \ functional/fault/setup.ksh \ functional/fault/zpool_status_-s.ksh \ functional/features/async_destroy/async_destroy_001_pos.ksh \ functional/features/async_destroy/cleanup.ksh \ functional/features/async_destroy/setup.ksh \ functional/features/large_dnode/cleanup.ksh \ functional/features/large_dnode/large_dnode_001_pos.ksh \ functional/features/large_dnode/large_dnode_002_pos.ksh \ functional/features/large_dnode/large_dnode_003_pos.ksh \ functional/features/large_dnode/large_dnode_004_neg.ksh \ functional/features/large_dnode/large_dnode_005_pos.ksh \ functional/features/large_dnode/large_dnode_006_pos.ksh \ functional/features/large_dnode/large_dnode_007_neg.ksh \ functional/features/large_dnode/large_dnode_008_pos.ksh \ functional/features/large_dnode/large_dnode_009_pos.ksh \ functional/features/large_dnode/setup.ksh \ functional/gang_blocks/cleanup.ksh \ functional/gang_blocks/gang_blocks_001_pos.ksh \ functional/gang_blocks/gang_blocks_ddt_copies.ksh \ functional/gang_blocks/gang_blocks_redundant.ksh \ functional/gang_blocks/gang_blocks_dyn_header_neg.ksh \ functional/gang_blocks/gang_blocks_dyn_header_pos.ksh \ functional/gang_blocks/gang_blocks_dyn_multi.ksh \ functional/gang_blocks/setup.ksh \ functional/grow/grow_pool_001_pos.ksh \ functional/grow/grow_replicas_001_pos.ksh \ functional/history/cleanup.ksh \ functional/history/history_001_pos.ksh \ functional/history/history_002_pos.ksh \ functional/history/history_003_pos.ksh \ functional/history/history_004_pos.ksh \ functional/history/history_005_neg.ksh \ functional/history/history_006_neg.ksh \ functional/history/history_007_pos.ksh \ functional/history/history_008_pos.ksh \ functional/history/history_009_pos.ksh \ functional/history/history_010_pos.ksh \ functional/history/setup.ksh \ functional/inheritance/cleanup.ksh \ functional/inheritance/inherit_001_pos.ksh \ functional/inuse/inuse_001_pos.ksh \ functional/inuse/inuse_003_pos.ksh \ functional/inuse/inuse_004_pos.ksh \ functional/inuse/inuse_005_pos.ksh \ functional/inuse/inuse_006_pos.ksh \ functional/inuse/inuse_007_pos.ksh \ functional/inuse/inuse_008_pos.ksh \ functional/inuse/inuse_009_pos.ksh \ functional/inuse/setup.ksh \ functional/io/cleanup.ksh \ functional/io/io_uring.ksh \ functional/io/libaio.ksh \ functional/io/mmap.ksh \ functional/io/posixaio.ksh \ functional/io/psync.ksh \ functional/io/setup.ksh \ functional/io/sync.ksh \ functional/l2arc/cleanup.ksh \ functional/l2arc/l2arc_arcstats_pos.ksh \ functional/l2arc/l2arc_l2miss_pos.ksh \ functional/l2arc/l2arc_mfuonly_pos.ksh \ functional/l2arc/persist_l2arc_001_pos.ksh \ functional/l2arc/persist_l2arc_002_pos.ksh \ functional/l2arc/persist_l2arc_003_neg.ksh \ functional/l2arc/persist_l2arc_004_pos.ksh \ functional/l2arc/persist_l2arc_005_pos.ksh \ functional/l2arc/setup.ksh \ functional/large_files/cleanup.ksh \ functional/large_files/large_files_001_pos.ksh \ functional/large_files/large_files_002_pos.ksh \ functional/large_files/setup.ksh \ functional/largest_pool/largest_pool_001_pos.ksh \ functional/libzfs/cleanup.ksh \ functional/libzfs/libzfs_input.ksh \ functional/libzfs/setup.ksh \ functional/limits/cleanup.ksh \ functional/limits/filesystem_count.ksh \ functional/limits/filesystem_limit.ksh \ functional/limits/setup.ksh \ functional/limits/snapshot_count.ksh \ functional/limits/snapshot_limit.ksh \ functional/link_count/cleanup.ksh \ functional/link_count/link_count_001.ksh \ functional/link_count/link_count_root_inode.ksh \ functional/link_count/setup.ksh \ functional/longname/cleanup.ksh \ functional/longname/longname_001_pos.ksh \ functional/longname/longname_002_pos.ksh \ functional/longname/longname_003_pos.ksh \ functional/longname/setup.ksh \ functional/log_spacemap/log_spacemap_import_logs.ksh \ functional/migration/cleanup.ksh \ functional/migration/migration_001_pos.ksh \ functional/migration/migration_002_pos.ksh \ functional/migration/migration_003_pos.ksh \ functional/migration/migration_004_pos.ksh \ functional/migration/migration_005_pos.ksh \ functional/migration/migration_006_pos.ksh \ functional/migration/migration_007_pos.ksh \ functional/migration/migration_008_pos.ksh \ functional/migration/migration_009_pos.ksh \ functional/migration/migration_010_pos.ksh \ functional/migration/migration_011_pos.ksh \ functional/migration/migration_012_pos.ksh \ functional/migration/setup.ksh \ functional/mmap/cleanup.ksh \ functional/mmap/mmap_libaio_001_pos.ksh \ functional/mmap/mmap_mixed.ksh \ functional/mmap/mmap_read_001_pos.ksh \ functional/mmap/mmap_seek_001_pos.ksh \ functional/mmap/mmap_sync_001_pos.ksh \ functional/mmap/mmap_write_001_pos.ksh \ functional/mmap/mmap_ftruncate.ksh \ functional/mmap/setup.ksh \ functional/mmp/cleanup.ksh \ functional/mmp/mmp_active_import.ksh \ functional/mmp/mmp_exported_import.ksh \ functional/mmp/mmp_hostid.ksh \ functional/mmp/mmp_inactive_import.ksh \ functional/mmp/mmp_interval.ksh \ functional/mmp/mmp_on_off.ksh \ functional/mmp/mmp_on_thread.ksh \ functional/mmp/mmp_on_uberblocks.ksh \ functional/mmp/mmp_on_zdb.ksh \ functional/mmp/mmp_reset_interval.ksh \ functional/mmp/mmp_write_distribution.ksh \ functional/mmp/mmp_write_slow_disk.ksh \ functional/mmp/mmp_write_uberblocks.ksh \ functional/mmp/multihost_history.ksh \ functional/mmp/setup.ksh \ functional/mount/cleanup.ksh \ functional/mount/setup.ksh \ functional/mount/umount_001.ksh \ functional/mount/umountall_001.ksh \ functional/mount/umount_unlinked_drain.ksh \ functional/mv_files/cleanup.ksh \ functional/mv_files/mv_files_001_pos.ksh \ functional/mv_files/mv_files_002_pos.ksh \ functional/mv_files/random_creation.ksh \ functional/mv_files/setup.ksh \ functional/nestedfs/cleanup.ksh \ functional/nestedfs/nestedfs_001_pos.ksh \ functional/nestedfs/setup.ksh \ functional/nopwrite/cleanup.ksh \ functional/nopwrite/nopwrite_copies.ksh \ functional/nopwrite/nopwrite_mtime.ksh \ functional/nopwrite/nopwrite_negative.ksh \ functional/nopwrite/nopwrite_promoted_clone.ksh \ functional/nopwrite/nopwrite_recsize.ksh \ functional/nopwrite/nopwrite_sync.ksh \ functional/nopwrite/nopwrite_varying_compression.ksh \ functional/nopwrite/nopwrite_volume.ksh \ functional/nopwrite/setup.ksh \ functional/no_space/cleanup.ksh \ functional/no_space/enospc_001_pos.ksh \ functional/no_space/enospc_002_pos.ksh \ functional/no_space/enospc_003_pos.ksh \ functional/no_space/enospc_df.ksh \ functional/no_space/enospc_ganging.ksh \ functional/no_space/enospc_rm.ksh \ functional/no_space/setup.ksh \ functional/online_offline/cleanup.ksh \ functional/online_offline/online_offline_001_pos.ksh \ functional/online_offline/online_offline_002_neg.ksh \ functional/online_offline/online_offline_003_neg.ksh \ functional/online_offline/setup.ksh \ functional/pam/cleanup.ksh \ functional/pam/pam_basic.ksh \ functional/pam/pam_change_unmounted.ksh \ functional/pam/pam_mount_recursively.ksh \ functional/pam/pam_nounmount.ksh \ functional/pam/pam_recursive.ksh \ functional/pam/pam_short_password.ksh \ functional/pam/setup.ksh \ functional/pool_checkpoint/checkpoint_after_rewind.ksh \ functional/pool_checkpoint/checkpoint_big_rewind.ksh \ functional/pool_checkpoint/checkpoint_capacity.ksh \ functional/pool_checkpoint/checkpoint_conf_change.ksh \ functional/pool_checkpoint/checkpoint_discard_busy.ksh \ functional/pool_checkpoint/checkpoint_discard.ksh \ functional/pool_checkpoint/checkpoint_discard_many.ksh \ functional/pool_checkpoint/checkpoint_indirect.ksh \ functional/pool_checkpoint/checkpoint_invalid.ksh \ functional/pool_checkpoint/checkpoint_lun_expsz.ksh \ functional/pool_checkpoint/checkpoint_open.ksh \ functional/pool_checkpoint/checkpoint_removal.ksh \ functional/pool_checkpoint/checkpoint_rewind.ksh \ functional/pool_checkpoint/checkpoint_ro_rewind.ksh \ functional/pool_checkpoint/checkpoint_sm_scale.ksh \ functional/pool_checkpoint/checkpoint_twice.ksh \ functional/pool_checkpoint/checkpoint_vdev_add.ksh \ functional/pool_checkpoint/checkpoint_zdb.ksh \ functional/pool_checkpoint/checkpoint_zhack_feat.ksh \ functional/pool_checkpoint/cleanup.ksh \ functional/pool_checkpoint/setup.ksh \ functional/pool_names/pool_names_001_pos.ksh \ functional/pool_names/pool_names_002_neg.ksh \ functional/poolversion/cleanup.ksh \ functional/poolversion/poolversion_001_pos.ksh \ functional/poolversion/poolversion_002_pos.ksh \ functional/poolversion/setup.ksh \ functional/privilege/cleanup.ksh \ functional/privilege/privilege_001_pos.ksh \ functional/privilege/privilege_002_pos.ksh \ functional/privilege/setup.ksh \ functional/procfs/cleanup.ksh \ functional/procfs/pool_state.ksh \ functional/procfs/procfs_list_basic.ksh \ functional/procfs/procfs_list_concurrent_readers.ksh \ functional/procfs/procfs_list_stale_read.ksh \ functional/procfs/setup.ksh \ functional/projectquota/cleanup.ksh \ functional/projectquota/projectid_001_pos.ksh \ functional/projectquota/projectid_002_pos.ksh \ functional/projectquota/projectid_003_pos.ksh \ functional/projectquota/projectquota_001_pos.ksh \ functional/projectquota/projectquota_002_pos.ksh \ functional/projectquota/projectquota_003_pos.ksh \ functional/projectquota/projectquota_004_neg.ksh \ functional/projectquota/projectquota_005_pos.ksh \ functional/projectquota/projectquota_006_pos.ksh \ functional/projectquota/projectquota_007_pos.ksh \ functional/projectquota/projectquota_008_pos.ksh \ functional/projectquota/projectquota_009_pos.ksh \ functional/projectquota/defaultprojectquota_001_pos.ksh \ functional/projectquota/defaultprojectquota_002_pos.ksh \ functional/projectquota/defaultprojectquota_003_neg.ksh \ functional/projectquota/defaultprojectquota_004_pos.ksh \ functional/projectquota/defaultprojectquota_005_pos.ksh \ functional/projectquota/defaultprojectquota_006_pos.ksh \ functional/projectquota/defaultprojectquota_007_pos.ksh \ functional/projectquota/projectspace_001_pos.ksh \ functional/projectquota/projectspace_002_pos.ksh \ functional/projectquota/projectspace_003_pos.ksh \ functional/projectquota/projectspace_004_pos.ksh \ functional/projectquota/projectspace_005_pos.ksh \ functional/projectquota/projecttree_001_pos.ksh \ functional/projectquota/projecttree_002_pos.ksh \ functional/projectquota/projecttree_003_neg.ksh \ functional/projectquota/setup.ksh \ functional/quota/cleanup.ksh \ functional/quota/quota_001_pos.ksh \ functional/quota/quota_002_pos.ksh \ functional/quota/quota_003_pos.ksh \ functional/quota/quota_004_pos.ksh \ functional/quota/quota_005_pos.ksh \ functional/quota/quota_006_neg.ksh \ functional/quota/setup.ksh \ functional/raidz/cleanup.ksh \ functional/raidz/raidz_001_neg.ksh \ functional/raidz/raidz_002_pos.ksh \ functional/raidz/raidz_expand_001_pos.ksh \ functional/raidz/raidz_expand_002_pos.ksh \ functional/raidz/raidz_expand_003_neg.ksh \ functional/raidz/raidz_expand_003_pos.ksh \ functional/raidz/raidz_expand_004_pos.ksh \ functional/raidz/raidz_expand_005_pos.ksh \ functional/raidz/raidz_expand_006_neg.ksh \ functional/raidz/raidz_expand_007_neg.ksh \ functional/raidz/setup.ksh \ functional/redacted_send/cleanup.ksh \ functional/redacted_send/redacted_compressed.ksh \ functional/redacted_send/redacted_contents.ksh \ functional/redacted_send/redacted_deleted.ksh \ functional/redacted_send/redacted_disabled_feature.ksh \ functional/redacted_send/redacted_embedded.ksh \ functional/redacted_send/redacted_holes.ksh \ functional/redacted_send/redacted_incrementals.ksh \ functional/redacted_send/redacted_largeblocks.ksh \ functional/redacted_send/redacted_many_clones.ksh \ functional/redacted_send/redacted_mixed_recsize.ksh \ functional/redacted_send/redacted_mounts.ksh \ functional/redacted_send/redacted_negative.ksh \ functional/redacted_send/redacted_origin.ksh \ functional/redacted_send/redacted_panic.ksh \ functional/redacted_send/redacted_props.ksh \ functional/redacted_send/redacted_resume.ksh \ functional/redacted_send/redacted_size.ksh \ functional/redacted_send/redacted_volume.ksh \ functional/redacted_send/setup.ksh \ functional/redundancy/cleanup.ksh \ functional/redundancy/redundancy_draid1.ksh \ functional/redundancy/redundancy_draid2.ksh \ functional/redundancy/redundancy_draid3.ksh \ functional/redundancy/redundancy_draid_damaged1.ksh \ functional/redundancy/redundancy_draid_damaged2.ksh \ functional/redundancy/redundancy_draid.ksh \ functional/redundancy/redundancy_draid_spare1.ksh \ functional/redundancy/redundancy_draid_spare2.ksh \ functional/redundancy/redundancy_draid_spare3.ksh \ functional/redundancy/redundancy_mirror.ksh \ functional/redundancy/redundancy_raidz1.ksh \ functional/redundancy/redundancy_raidz2.ksh \ functional/redundancy/redundancy_raidz3.ksh \ functional/redundancy/redundancy_raidz.ksh \ functional/redundancy/redundancy_stripe.ksh \ functional/redundancy/setup.ksh \ functional/refquota/cleanup.ksh \ functional/refquota/refquota_001_pos.ksh \ functional/refquota/refquota_002_pos.ksh \ functional/refquota/refquota_003_pos.ksh \ functional/refquota/refquota_004_pos.ksh \ functional/refquota/refquota_005_pos.ksh \ functional/refquota/refquota_006_neg.ksh \ functional/refquota/refquota_007_neg.ksh \ functional/refquota/refquota_008_neg.ksh \ functional/refquota/setup.ksh \ functional/refreserv/cleanup.ksh \ functional/refreserv/refreserv_001_pos.ksh \ functional/refreserv/refreserv_002_pos.ksh \ functional/refreserv/refreserv_003_pos.ksh \ functional/refreserv/refreserv_004_pos.ksh \ functional/refreserv/refreserv_005_pos.ksh \ functional/refreserv/refreserv_multi_raidz.ksh \ functional/refreserv/refreserv_raidz.ksh \ functional/refreserv/setup.ksh \ functional/removal/cleanup.ksh \ functional/removal/removal_all_vdev.ksh \ functional/removal/removal_cancel.ksh \ functional/removal/removal_check_space.ksh \ functional/removal/removal_condense_export.ksh \ functional/removal/removal_multiple_indirection.ksh \ functional/removal/removal_nopwrite.ksh \ functional/removal/removal_remap_deadlists.ksh \ functional/removal/removal_reservation.ksh \ functional/removal/removal_resume_export.ksh \ functional/removal/removal_sanity.ksh \ functional/removal/removal_with_add.ksh \ functional/removal/removal_with_create_fs.ksh \ functional/removal/removal_with_dedup.ksh \ functional/removal/removal_with_errors.ksh \ functional/removal/removal_with_export.ksh \ functional/removal/removal_with_faulted.ksh \ functional/removal/removal_with_ganging.ksh \ functional/removal/removal_with_hole.ksh \ functional/removal/removal_with_indirect.ksh \ functional/removal/removal_with_remove.ksh \ functional/removal/removal_with_scrub.ksh \ functional/removal/removal_with_send.ksh \ functional/removal/removal_with_send_recv.ksh \ functional/removal/removal_with_snapshot.ksh \ functional/removal/removal_with_write.ksh \ functional/removal/removal_with_zdb.ksh \ functional/removal/remove_attach_mirror.ksh \ functional/removal/remove_expanded.ksh \ functional/removal/remove_indirect.ksh \ functional/removal/remove_mirror.ksh \ functional/removal/remove_mirror_sanity.ksh \ functional/removal/remove_raidz.ksh \ functional/rename_dirs/cleanup.ksh \ functional/rename_dirs/rename_dirs_001_pos.ksh \ functional/rename_dirs/setup.ksh \ functional/renameat2/cleanup.ksh \ functional/renameat2/setup.ksh \ functional/renameat2/renameat2_exchange.ksh \ functional/renameat2/renameat2_noreplace.ksh \ functional/renameat2/renameat2_whiteout.ksh \ functional/replacement/attach_import.ksh \ functional/replacement/attach_multiple.ksh \ functional/replacement/attach_rebuild.ksh \ functional/replacement/attach_resilver.ksh \ functional/replacement/cleanup.ksh \ functional/replacement/detach.ksh \ functional/replacement/rebuild_disabled_feature.ksh \ functional/replacement/rebuild_multiple.ksh \ functional/replacement/rebuild_raidz.ksh \ functional/replacement/replace_import.ksh \ functional/replacement/replace_rebuild.ksh \ functional/replacement/replace_resilver.ksh \ functional/replacement/resilver_restart_001.ksh \ functional/replacement/resilver_restart_002.ksh \ functional/replacement/scrub_cancel.ksh \ functional/replacement/setup.ksh \ functional/reservation/cleanup.ksh \ functional/reservation/reservation_001_pos.ksh \ functional/reservation/reservation_002_pos.ksh \ functional/reservation/reservation_003_pos.ksh \ functional/reservation/reservation_004_pos.ksh \ functional/reservation/reservation_005_pos.ksh \ functional/reservation/reservation_006_pos.ksh \ functional/reservation/reservation_007_pos.ksh \ functional/reservation/reservation_008_pos.ksh \ functional/reservation/reservation_009_pos.ksh \ functional/reservation/reservation_010_pos.ksh \ functional/reservation/reservation_011_pos.ksh \ functional/reservation/reservation_012_pos.ksh \ functional/reservation/reservation_013_pos.ksh \ functional/reservation/reservation_014_pos.ksh \ functional/reservation/reservation_015_pos.ksh \ functional/reservation/reservation_016_pos.ksh \ functional/reservation/reservation_017_pos.ksh \ functional/reservation/reservation_018_pos.ksh \ functional/reservation/reservation_019_pos.ksh \ functional/reservation/reservation_020_pos.ksh \ functional/reservation/reservation_021_neg.ksh \ functional/reservation/reservation_022_pos.ksh \ functional/reservation/setup.ksh \ functional/rootpool/cleanup.ksh \ functional/rootpool/rootpool_002_neg.ksh \ functional/rootpool/rootpool_003_neg.ksh \ functional/rootpool/rootpool_007_pos.ksh \ functional/rootpool/setup.ksh \ functional/rsend/cleanup.ksh \ functional/rsend/recv_dedup_encrypted_zvol.ksh \ functional/rsend/recv_dedup.ksh \ functional/rsend/rsend_001_pos.ksh \ functional/rsend/rsend_002_pos.ksh \ functional/rsend/rsend_003_pos.ksh \ functional/rsend/rsend_004_pos.ksh \ functional/rsend/rsend_005_pos.ksh \ functional/rsend/rsend_006_pos.ksh \ functional/rsend/rsend_007_pos.ksh \ functional/rsend/rsend_008_pos.ksh \ functional/rsend/rsend_009_pos.ksh \ functional/rsend/rsend_010_pos.ksh \ functional/rsend/rsend_011_pos.ksh \ functional/rsend/rsend_012_pos.ksh \ functional/rsend/rsend_013_pos.ksh \ functional/rsend/rsend_014_pos.ksh \ functional/rsend/rsend_016_neg.ksh \ functional/rsend/rsend_019_pos.ksh \ functional/rsend/rsend_020_pos.ksh \ functional/rsend/rsend_021_pos.ksh \ functional/rsend/rsend_022_pos.ksh \ functional/rsend/rsend_024_pos.ksh \ functional/rsend/rsend_025_pos.ksh \ functional/rsend/rsend_026_neg.ksh \ functional/rsend/rsend_027_pos.ksh \ functional/rsend/rsend_028_neg.ksh \ functional/rsend/rsend_029_neg.ksh \ functional/rsend/rsend_030_pos.ksh \ functional/rsend/rsend_031_pos.ksh \ functional/rsend/send-c_embedded_blocks.ksh \ functional/rsend/send-c_incremental.ksh \ functional/rsend/send-c_longname.ksh \ functional/rsend/send-c_lz4_disabled.ksh \ functional/rsend/send-c_mixed_compression.ksh \ functional/rsend/send-c_props.ksh \ functional/rsend/send-c_recv_dedup.ksh \ functional/rsend/send-c_recv_lz4_disabled.ksh \ functional/rsend/send-c_resume.ksh \ functional/rsend/send-c_stream_size_estimate.ksh \ functional/rsend/send-c_verify_contents.ksh \ functional/rsend/send-c_verify_ratio.ksh \ functional/rsend/send-c_volume.ksh \ functional/rsend/send-c_zstream_recompress.ksh \ functional/rsend/send-c_zstreamdump.ksh \ functional/rsend/send-cpL_varied_recsize.ksh \ functional/rsend/send_doall.ksh \ functional/rsend/send_encrypted_incremental.ksh \ functional/rsend/send_encrypted_files.ksh \ functional/rsend/send_encrypted_freeobjects.ksh \ functional/rsend/send_encrypted_hierarchy.ksh \ functional/rsend/send_encrypted_props.ksh \ functional/rsend/send_encrypted_truncated_files.ksh \ functional/rsend/send_freeobjects.ksh \ functional/rsend/send_holds.ksh \ functional/rsend/send_hole_birth.ksh \ functional/rsend/send_invalid.ksh \ functional/rsend/send_leak_keymaps.ksh \ functional/rsend/send-L_toggle.ksh \ functional/rsend/send_mixed_raw.ksh \ functional/rsend/send_partial_dataset.ksh \ functional/rsend/send_raw_ashift.ksh \ functional/rsend/send_raw_spill_block.ksh \ functional/rsend/send_raw_large_blocks.ksh \ functional/rsend/send_realloc_dnode_size.ksh \ functional/rsend/send_realloc_encrypted_files.ksh \ functional/rsend/send_realloc_files.ksh \ functional/rsend/send_spill_block.ksh \ functional/rsend/send-wR_encrypted_zvol.ksh \ functional/rsend/setup.ksh \ functional/scrub_mirror/cleanup.ksh \ functional/scrub_mirror/scrub_mirror_001_pos.ksh \ functional/scrub_mirror/scrub_mirror_002_pos.ksh \ functional/scrub_mirror/scrub_mirror_003_pos.ksh \ functional/scrub_mirror/scrub_mirror_004_pos.ksh \ functional/scrub_mirror/setup.ksh \ functional/slog/cleanup.ksh \ functional/slog/setup.ksh \ functional/slog/slog_001_pos.ksh \ functional/slog/slog_002_pos.ksh \ functional/slog/slog_003_pos.ksh \ functional/slog/slog_004_pos.ksh \ functional/slog/slog_005_pos.ksh \ functional/slog/slog_006_pos.ksh \ functional/slog/slog_007_pos.ksh \ functional/slog/slog_008_neg.ksh \ functional/slog/slog_009_neg.ksh \ functional/slog/slog_010_neg.ksh \ functional/slog/slog_011_neg.ksh \ functional/slog/slog_012_neg.ksh \ functional/slog/slog_013_pos.ksh \ functional/slog/slog_014_pos.ksh \ functional/slog/slog_015_neg.ksh \ functional/slog/slog_016_pos.ksh \ functional/slog/slog_replay_fs_001.ksh \ functional/slog/slog_replay_fs_002.ksh \ functional/slog/slog_replay_volume.ksh \ functional/snapshot/cleanup.ksh \ functional/snapshot/clone_001_pos.ksh \ functional/snapshot/rollback_001_pos.ksh \ functional/snapshot/rollback_002_pos.ksh \ functional/snapshot/rollback_003_pos.ksh \ functional/snapshot/setup.ksh \ functional/snapshot/snapshot_001_pos.ksh \ functional/snapshot/snapshot_002_pos.ksh \ functional/snapshot/snapshot_003_pos.ksh \ functional/snapshot/snapshot_004_pos.ksh \ functional/snapshot/snapshot_005_pos.ksh \ functional/snapshot/snapshot_006_pos.ksh \ functional/snapshot/snapshot_007_pos.ksh \ functional/snapshot/snapshot_008_pos.ksh \ functional/snapshot/snapshot_009_pos.ksh \ functional/snapshot/snapshot_010_pos.ksh \ functional/snapshot/snapshot_011_pos.ksh \ functional/snapshot/snapshot_012_pos.ksh \ functional/snapshot/snapshot_013_pos.ksh \ functional/snapshot/snapshot_014_pos.ksh \ functional/snapshot/snapshot_015_pos.ksh \ functional/snapshot/snapshot_016_pos.ksh \ functional/snapshot/snapshot_017_pos.ksh \ functional/snapshot/snapshot_018_pos.ksh \ functional/snapused/cleanup.ksh \ functional/snapused/setup.ksh \ functional/snapused/snapused_001_pos.ksh \ functional/snapused/snapused_002_pos.ksh \ functional/snapused/snapused_003_pos.ksh \ functional/snapused/snapused_004_pos.ksh \ functional/snapused/snapused_005_pos.ksh \ functional/sparse/cleanup.ksh \ functional/sparse/setup.ksh \ functional/sparse/sparse_001_pos.ksh \ functional/stat/cleanup.ksh \ functional/stat/setup.ksh \ functional/stat/stat_001_pos.ksh \ functional/stat/statx_dioalign.ksh \ functional/syncfs/syncfs_suspend.ksh \ functional/suid/cleanup.ksh \ functional/suid/setup.ksh \ functional/suid/suid_write_to_none.ksh \ functional/suid/suid_write_to_sgid.ksh \ functional/suid/suid_write_to_suid.ksh \ functional/suid/suid_write_to_suid_sgid.ksh \ functional/suid/suid_write_zil_replay.ksh \ functional/trim/autotrim_config.ksh \ functional/trim/autotrim_integrity.ksh \ functional/trim/autotrim_trim_integrity.ksh \ functional/trim/cleanup.ksh \ functional/trim/setup.ksh \ functional/trim/trim_config.ksh \ functional/trim/trim_integrity.ksh \ functional/trim/trim_l2arc.ksh \ functional/truncate/cleanup.ksh \ functional/truncate/setup.ksh \ functional/truncate/truncate_001_pos.ksh \ functional/truncate/truncate_002_pos.ksh \ functional/truncate/truncate_timestamps.ksh \ functional/upgrade/cleanup.ksh \ functional/upgrade/setup.ksh \ functional/upgrade/upgrade_projectquota_001_pos.ksh \ functional/upgrade/upgrade_projectquota_002_pos.ksh \ functional/upgrade/upgrade_readonly_pool.ksh \ functional/upgrade/upgrade_userobj_001_pos.ksh \ functional/user_namespace/cleanup.ksh \ functional/user_namespace/setup.ksh \ functional/user_namespace/user_namespace_001.ksh \ functional/user_namespace/user_namespace_002.ksh \ functional/user_namespace/user_namespace_003.ksh \ functional/user_namespace/user_namespace_004.ksh \ functional/userquota/cleanup.ksh \ functional/userquota/groupspace_001_pos.ksh \ functional/userquota/groupspace_002_pos.ksh \ functional/userquota/groupspace_003_pos.ksh \ functional/userquota/groupspace_004_pos.ksh \ functional/userquota/setup.ksh \ functional/userquota/defaultuserquota_001_pos.ksh \ functional/userquota/defaultuserquota_002_pos.ksh \ functional/userquota/defaultuserquota_003_pos.ksh \ functional/userquota/defaultuserquota_004_neg.ksh \ functional/userquota/defaultuserquota_005_pos.ksh \ functional/userquota/defaultuserquota_006_pos.ksh \ functional/userquota/defaultuserquota_007_pos.ksh \ functional/userquota/defaultuserquota_008_pos.ksh \ functional/userquota/defaultuserquota_009_pos.ksh \ functional/userquota/defaultuserquota_010_neg.ksh \ functional/userquota/defaultuserquota_011_neg.ksh \ functional/userquota/defaultuserquota_012_neg.ksh \ functional/userquota/defaultuserquota_013_neg.ksh \ functional/userquota/userquota_001_pos.ksh \ functional/userquota/userquota_002_pos.ksh \ functional/userquota/userquota_003_pos.ksh \ functional/userquota/userquota_004_pos.ksh \ functional/userquota/userquota_005_neg.ksh \ functional/userquota/userquota_006_pos.ksh \ functional/userquota/userquota_007_pos.ksh \ functional/userquota/userquota_008_pos.ksh \ functional/userquota/userquota_009_pos.ksh \ functional/userquota/userquota_010_pos.ksh \ functional/userquota/userquota_011_pos.ksh \ functional/userquota/userquota_012_neg.ksh \ functional/userquota/userquota_013_pos.ksh \ functional/userquota/userspace_001_pos.ksh \ functional/userquota/userspace_002_pos.ksh \ functional/userquota/userspace_003_pos.ksh \ functional/userquota/userspace_004_pos.ksh \ functional/userquota/userspace_encrypted.ksh \ functional/userquota/userspace_send_encrypted.ksh \ functional/userquota/userspace_encrypted_13709.ksh \ functional/vdev_zaps/cleanup.ksh \ functional/vdev_zaps/setup.ksh \ functional/vdev_zaps/vdev_zaps_001_pos.ksh \ functional/vdev_zaps/vdev_zaps_002_pos.ksh \ functional/vdev_zaps/vdev_zaps_003_pos.ksh \ functional/vdev_zaps/vdev_zaps_004_pos.ksh \ functional/vdev_zaps/vdev_zaps_005_pos.ksh \ functional/vdev_zaps/vdev_zaps_006_pos.ksh \ functional/vdev_zaps/vdev_zaps_007_pos.ksh \ functional/write_dirs/cleanup.ksh \ functional/write_dirs/setup.ksh \ functional/write_dirs/write_dirs_001_pos.ksh \ functional/write_dirs/write_dirs_002_pos.ksh \ functional/xattr/cleanup.ksh \ functional/xattr/setup.ksh \ functional/xattr/xattr_001_pos.ksh \ functional/xattr/xattr_002_neg.ksh \ functional/xattr/xattr_003_neg.ksh \ functional/xattr/xattr_004_pos.ksh \ functional/xattr/xattr_005_pos.ksh \ functional/xattr/xattr_006_pos.ksh \ functional/xattr/xattr_007_neg.ksh \ functional/xattr/xattr_008_pos.ksh \ functional/xattr/xattr_009_neg.ksh \ functional/xattr/xattr_010_neg.ksh \ functional/xattr/xattr_011_pos.ksh \ functional/xattr/xattr_012_pos.ksh \ functional/xattr/xattr_013_pos.ksh \ functional/xattr/xattr_compat.ksh \ functional/zap_shrink/cleanup.ksh \ functional/zap_shrink/zap_shrink_001_pos.ksh \ functional/zap_shrink/setup.ksh \ functional/zpool_influxdb/cleanup.ksh \ functional/zpool_influxdb/setup.ksh \ functional/zpool_influxdb/zpool_influxdb.ksh \ functional/zvol/zvol_cli/cleanup.ksh \ functional/zvol/zvol_cli/setup.ksh \ functional/zvol/zvol_cli/zvol_cli_001_pos.ksh \ functional/zvol/zvol_cli/zvol_cli_002_pos.ksh \ functional/zvol/zvol_cli/zvol_cli_003_neg.ksh \ functional/zvol/zvol_ENOSPC/cleanup.ksh \ functional/zvol/zvol_ENOSPC/setup.ksh \ functional/zvol/zvol_ENOSPC/zvol_ENOSPC_001_pos.ksh \ functional/zvol/zvol_misc/cleanup.ksh \ functional/zvol/zvol_misc/setup.ksh \ functional/zvol/zvol_misc/zvol_misc_001_neg.ksh \ functional/zvol/zvol_misc/zvol_misc_002_pos.ksh \ functional/zvol/zvol_misc/zvol_misc_003_neg.ksh \ functional/zvol/zvol_misc/zvol_misc_004_pos.ksh \ functional/zvol/zvol_misc/zvol_misc_005_neg.ksh \ functional/zvol/zvol_misc/zvol_misc_006_pos.ksh \ functional/zvol/zvol_misc/zvol_misc_fua.ksh \ functional/zvol/zvol_misc/zvol_misc_hierarchy.ksh \ functional/zvol/zvol_misc/zvol_misc_rename_inuse.ksh \ functional/zvol/zvol_misc/zvol_misc_snapdev.ksh \ functional/zvol/zvol_misc/zvol_misc_trim.ksh \ functional/zvol/zvol_misc/zvol_misc_volmode.ksh \ functional/zvol/zvol_misc/zvol_misc_zil.ksh \ functional/zvol/zvol_stress/cleanup.ksh \ functional/zvol/zvol_stress/setup.ksh \ functional/zvol/zvol_stress/zvol_stress.ksh \ functional/zvol/zvol_swap/cleanup.ksh \ functional/zvol/zvol_swap/setup.ksh \ functional/zvol/zvol_swap/zvol_swap_001_pos.ksh \ functional/zvol/zvol_swap/zvol_swap_002_pos.ksh \ functional/zvol/zvol_swap/zvol_swap_003_pos.ksh \ functional/zvol/zvol_swap/zvol_swap_004_pos.ksh \ functional/zvol/zvol_swap/zvol_swap_005_pos.ksh \ functional/zvol/zvol_swap/zvol_swap_006_pos.ksh \ functional/idmap_mount/cleanup.ksh \ functional/idmap_mount/setup.ksh \ functional/idmap_mount/idmap_mount_001.ksh \ functional/idmap_mount/idmap_mount_002.ksh \ functional/idmap_mount/idmap_mount_003.ksh \ functional/idmap_mount/idmap_mount_004.ksh \ functional/idmap_mount/idmap_mount_005.ksh diff --git a/tests/zfs-tests/tests/functional/cli_root/zfs_rewrite/zfs_rewrite_physical.ksh b/tests/zfs-tests/tests/functional/cli_root/zfs_rewrite/zfs_rewrite_physical.ksh new file mode 100755 index 000000000000..142e44f53515 --- /dev/null +++ b/tests/zfs-tests/tests/functional/cli_root/zfs_rewrite/zfs_rewrite_physical.ksh @@ -0,0 +1,100 @@ +#!/bin/ksh -p +# SPDX-License-Identifier: CDDL-1.0 +# +# 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 https://opensource.org/licenses/CDDL-1.0. +# 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) 2025, iXsystems, Inc. +# + +# DESCRIPTION: +# Verify zfs rewrite -P flag correctly preserves logical birth times. +# +# STRATEGY: +# 1. Create a test file and sync it. +# 2. Create a snapshot to capture the original birth time. +# 3. Test default rewrite behavior (updates logical birth time). +# 4. Test -P flag behavior (preserves logical birth time). +# 5. Verify incremental send behavior difference. + +. $STF_SUITE/include/libtest.shlib + +typeset tmp=$(mktemp) +typeset send_default=$(mktemp) +typeset send_physical=$(mktemp) + +function cleanup +{ + rm -rf $tmp $send_default $send_physical $TESTDIR/* + zfs destroy -R $TESTPOOL/$TESTFS@snap1 2>/dev/null || true + zfs destroy -R $TESTPOOL/$TESTFS@snap2 2>/dev/null || true + zfs destroy -R $TESTPOOL/$TESTFS@snap3 2>/dev/null || true +} + +log_assert "zfs rewrite -P flag correctly preserves logical birth times" + +log_onexit cleanup + +log_must zfs set recordsize=128k $TESTPOOL/$TESTFS + +# Create test file and initial snapshot +log_must dd if=/dev/urandom of=$TESTDIR/testfile bs=128k count=4 +log_must sync_pool $TESTPOOL +typeset orig_hash=$(xxh128digest $TESTDIR/testfile) +log_must zfs snapshot $TESTPOOL/$TESTFS@snap1 + +# Test default rewrite behavior (updates logical birth time) +log_must zfs rewrite $TESTDIR/testfile +log_must sync_pool $TESTPOOL +typeset default_hash=$(xxh128digest $TESTDIR/testfile) +log_must [ "$orig_hash" = "$default_hash" ] +log_must zfs snapshot $TESTPOOL/$TESTFS@snap2 + +# Test incremental send size - should be large with updated birth time +log_must eval "zfs send -i @snap1 $TESTPOOL/$TESTFS@snap2 > $send_default" +typeset default_size=$(wc -c < $send_default) +log_note "Default rewrite incremental send size: $default_size bytes" + +# Reset the file to original state +log_must zfs rollback -r $TESTPOOL/$TESTFS@snap1 + +# Test -P flag behavior (preserves logical birth time) +log_must zfs rewrite -P $TESTDIR/testfile +log_must sync_pool $TESTPOOL +typeset physical_hash=$(xxh128digest $TESTDIR/testfile) +log_must [ "$orig_hash" = "$physical_hash" ] +log_must zfs snapshot $TESTPOOL/$TESTFS@snap3 + +# Test incremental send size - should be minimal with preserved birth time +log_must eval "zfs send -i @snap1 $TESTPOOL/$TESTFS@snap3 > $send_physical" +typeset physical_size=$(wc -c < $send_physical) +log_note "Physical rewrite incremental send size: $physical_size bytes" + +# Verify that -P flag produces smaller incremental send +if [[ $physical_size -lt $default_size ]]; then + log_note "SUCCESS: -P flag produces smaller incremental send" \ + "($physical_size < $default_size)" +else + log_fail "FAIL: -P flag should produce smaller incremental send" \ + "($physical_size >= $default_size)" +fi + +log_pass diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_get/zpool_get.cfg b/tests/zfs-tests/tests/functional/cli_root/zpool_get/zpool_get.cfg index 3389dcf72f89..bdf5fdf85cff 100644 --- a/tests/zfs-tests/tests/functional/cli_root/zpool_get/zpool_get.cfg +++ b/tests/zfs-tests/tests/functional/cli_root/zpool_get/zpool_get.cfg @@ -1,120 +1,121 @@ # SPDX-License-Identifier: CDDL-1.0 # # 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 https://opensource.org/licenses/CDDL-1.0. # 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 2009 Sun Microsystems, Inc. All rights reserved. # Use is subject to license terms. # # # Copyright (c) 2013, 2014 by Delphix. All rights reserved. # Copyright 2016 Nexenta Systems, Inc. All rights reserved. # # Set the expected properties of zpool typeset -a properties=( "size" "capacity" "altroot" "health" "guid" "load_guid" "version" "bootfs" "delegation" "autoreplace" "cachefile" "checkpoint" "failmode" "listsnapshots" "autoexpand" "dedupratio" "dedup_table_quota" "dedup_table_size" "free" "allocated" "readonly" "comment" "expandsize" "freeing" "fragmentation" "leaked" "multihost" "autotrim" "compatibility" "bcloneused" "bclonesaved" "bcloneratio" "last_scrubbed_txg" "feature@async_destroy" "feature@empty_bpobj" "feature@lz4_compress" "feature@multi_vdev_crash_dump" "feature@spacemap_histogram" "feature@enabled_txg" "feature@hole_birth" "feature@extensible_dataset" "feature@embedded_data" "feature@bookmarks" "feature@filesystem_limits" "feature@large_blocks" "feature@sha512" "feature@skein" "feature@edonr" "feature@device_removal" "feature@obsolete_counts" "feature@zpool_checkpoint" "feature@spacemap_v2" "feature@redaction_bookmarks" "feature@redacted_datasets" "feature@bookmark_written" "feature@log_spacemap" "feature@device_rebuild" "feature@draid" "feature@redaction_list_spill" "feature@dynamic_gang_header" + "feature@physical_rewrite" ) if is_linux || is_freebsd; then properties+=( "ashift" "feature@large_dnode" "feature@userobj_accounting" "feature@encryption" "feature@project_quota" "feature@allocation_classes" "feature@resilver_defer" "feature@bookmark_v2" "feature@livelist" "feature@zstd_compress" "feature@zilsaxattr" "feature@head_errlog" "feature@blake3" "feature@block_cloning" "feature@vdev_zaps_v2" "feature@raidz_expansion" "feature@fast_dedup" "feature@longname" "feature@large_microzap" "feature@block_cloning_endian" ) fi