diff --git a/cmd/ztest/ztest.c b/cmd/ztest/ztest.c index e1b1ae3ffff7..46dd903f8449 100644 --- a/cmd/ztest/ztest.c +++ b/cmd/ztest/ztest.c @@ -1,7250 +1,7255 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2016 by Delphix. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2013 Steven Hartland. All rights reserved. * Copyright (c) 2014 Integros [integros.com] * Copyright 2017 Joyent, Inc. */ /* * The objective of this program is to provide a DMU/ZAP/SPA stress test * that runs entirely in userland, is easy to use, and easy to extend. * * The overall design of the ztest program is as follows: * * (1) For each major functional area (e.g. adding vdevs to a pool, * creating and destroying datasets, reading and writing objects, etc) * we have a simple routine to test that functionality. These * individual routines do not have to do anything "stressful". * * (2) We turn these simple functionality tests into a stress test by * running them all in parallel, with as many threads as desired, * and spread across as many datasets, objects, and vdevs as desired. * * (3) While all this is happening, we inject faults into the pool to * verify that self-healing data really works. * * (4) Every time we open a dataset, we change its checksum and compression * functions. Thus even individual objects vary from block to block * in which checksum they use and whether they're compressed. * * (5) To verify that we never lose on-disk consistency after a crash, * we run the entire test in a child of the main process. * At random times, the child self-immolates with a SIGKILL. * This is the software equivalent of pulling the power cord. * The parent then runs the test again, using the existing * storage pool, as many times as desired. If backwards compatibility * testing is enabled ztest will sometimes run the "older" version * of ztest after a SIGKILL. * * (6) To verify that we don't have future leaks or temporal incursions, * many of the functional tests record the transaction group number * as part of their data. When reading old data, they verify that * the transaction group number is less than the current, open txg. * If you add a new test, please do this if applicable. * * (7) Threads are created with a reduced stack size, for sanity checking. * Therefore, it's important not to allocate huge buffers on the stack. * * When run with no arguments, ztest runs for about five minutes and * produces no output if successful. To get a little bit of information, * specify -V. To get more information, specify -VV, and so on. * * To turn this into an overnight stress test, use -T to specify run time. * * You can ask more more vdevs [-v], datasets [-d], or threads [-t] * to increase the pool capacity, fanout, and overall stress level. * * Use the -k option to set the desired frequency of kills. * * When ztest invokes itself it passes all relevant information through a * temporary file which is mmap-ed in the child process. This allows shared * memory to survive the exec syscall. The ztest_shared_hdr_t struct is always * stored at offset 0 of this file and contains information on the size and * number of shared structures in the file. The information stored in this file * must remain backwards compatible with older versions of ztest so that * ztest can invoke them during backwards compatibility testing (-B). */ #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 #include #include #include #ifdef __GLIBC__ #include /* for backtrace() */ #endif static int ztest_fd_data = -1; static int ztest_fd_rand = -1; typedef struct ztest_shared_hdr { uint64_t zh_hdr_size; uint64_t zh_opts_size; uint64_t zh_size; uint64_t zh_stats_size; uint64_t zh_stats_count; uint64_t zh_ds_size; uint64_t zh_ds_count; } ztest_shared_hdr_t; static ztest_shared_hdr_t *ztest_shared_hdr; typedef struct ztest_shared_opts { char zo_pool[ZFS_MAX_DATASET_NAME_LEN]; char zo_dir[ZFS_MAX_DATASET_NAME_LEN]; char zo_alt_ztest[MAXNAMELEN]; char zo_alt_libpath[MAXNAMELEN]; uint64_t zo_vdevs; uint64_t zo_vdevtime; size_t zo_vdev_size; int zo_ashift; int zo_mirrors; int zo_raidz; int zo_raidz_parity; int zo_datasets; int zo_threads; uint64_t zo_passtime; uint64_t zo_killrate; int zo_verbose; int zo_init; uint64_t zo_time; uint64_t zo_maxloops; uint64_t zo_metaslab_gang_bang; int zo_mmp_test; } ztest_shared_opts_t; static const ztest_shared_opts_t ztest_opts_defaults = { .zo_pool = { 'z', 't', 'e', 's', 't', '\0' }, .zo_dir = { '/', 't', 'm', 'p', '\0' }, .zo_alt_ztest = { '\0' }, .zo_alt_libpath = { '\0' }, .zo_vdevs = 5, .zo_ashift = SPA_MINBLOCKSHIFT, .zo_mirrors = 2, .zo_raidz = 4, .zo_raidz_parity = 1, .zo_vdev_size = SPA_MINDEVSIZE * 4, /* 256m default size */ .zo_datasets = 7, .zo_threads = 23, .zo_passtime = 60, /* 60 seconds */ .zo_killrate = 70, /* 70% kill rate */ .zo_verbose = 0, .zo_mmp_test = 0, .zo_init = 1, .zo_time = 300, /* 5 minutes */ .zo_maxloops = 50, /* max loops during spa_freeze() */ .zo_metaslab_gang_bang = 32 << 10 }; extern uint64_t metaslab_gang_bang; extern uint64_t metaslab_df_alloc_threshold; extern int metaslab_preload_limit; extern boolean_t zfs_compressed_arc_enabled; extern int zfs_abd_scatter_enabled; static ztest_shared_opts_t *ztest_shared_opts; static ztest_shared_opts_t ztest_opts; static char *ztest_wkeydata = "abcdefghijklmnopqrstuvwxyz012345"; typedef struct ztest_shared_ds { uint64_t zd_seq; } ztest_shared_ds_t; static ztest_shared_ds_t *ztest_shared_ds; #define ZTEST_GET_SHARED_DS(d) (&ztest_shared_ds[d]) #define BT_MAGIC 0x123456789abcdefULL #define MAXFAULTS() \ (MAX(zs->zs_mirrors, 1) * (ztest_opts.zo_raidz_parity + 1) - 1) enum ztest_io_type { ZTEST_IO_WRITE_TAG, ZTEST_IO_WRITE_PATTERN, ZTEST_IO_WRITE_ZEROES, ZTEST_IO_TRUNCATE, ZTEST_IO_SETATTR, ZTEST_IO_REWRITE, ZTEST_IO_TYPES }; typedef struct ztest_block_tag { uint64_t bt_magic; uint64_t bt_objset; uint64_t bt_object; uint64_t bt_dnodesize; uint64_t bt_offset; uint64_t bt_gen; uint64_t bt_txg; uint64_t bt_crtxg; } ztest_block_tag_t; typedef struct bufwad { uint64_t bw_index; uint64_t bw_txg; uint64_t bw_data; } bufwad_t; typedef struct rll { void *rll_writer; int rll_readers; kmutex_t rll_lock; kcondvar_t rll_cv; } rll_t; typedef struct zll { list_t z_list; kmutex_t z_lock; } zll_t; #define ZTEST_RANGE_LOCKS 64 #define ZTEST_OBJECT_LOCKS 64 /* * Object descriptor. Used as a template for object lookup/create/remove. */ typedef struct ztest_od { uint64_t od_dir; uint64_t od_object; dmu_object_type_t od_type; dmu_object_type_t od_crtype; uint64_t od_blocksize; uint64_t od_crblocksize; uint64_t od_crdnodesize; uint64_t od_gen; uint64_t od_crgen; char od_name[ZFS_MAX_DATASET_NAME_LEN]; } ztest_od_t; /* * Per-dataset state. */ typedef struct ztest_ds { ztest_shared_ds_t *zd_shared; objset_t *zd_os; rwlock_t zd_zilog_lock; zilog_t *zd_zilog; ztest_od_t *zd_od; /* debugging aid */ char zd_name[ZFS_MAX_DATASET_NAME_LEN]; kmutex_t zd_dirobj_lock; rll_t zd_object_lock[ZTEST_OBJECT_LOCKS]; zll_t zd_range_lock[ZTEST_RANGE_LOCKS]; } ztest_ds_t; /* * Per-iteration state. */ typedef void ztest_func_t(ztest_ds_t *zd, uint64_t id); typedef struct ztest_info { ztest_func_t *zi_func; /* test function */ uint64_t zi_iters; /* iterations per execution */ uint64_t *zi_interval; /* execute every seconds */ const char *zi_funcname; /* name of test function */ } ztest_info_t; typedef struct ztest_shared_callstate { uint64_t zc_count; /* per-pass count */ uint64_t zc_time; /* per-pass time */ uint64_t zc_next; /* next time to call this function */ } ztest_shared_callstate_t; static ztest_shared_callstate_t *ztest_shared_callstate; #define ZTEST_GET_SHARED_CALLSTATE(c) (&ztest_shared_callstate[c]) ztest_func_t ztest_dmu_read_write; ztest_func_t ztest_dmu_write_parallel; ztest_func_t ztest_dmu_object_alloc_free; ztest_func_t ztest_dmu_commit_callbacks; ztest_func_t ztest_zap; ztest_func_t ztest_zap_parallel; ztest_func_t ztest_zil_commit; ztest_func_t ztest_zil_remount; ztest_func_t ztest_dmu_read_write_zcopy; ztest_func_t ztest_dmu_objset_create_destroy; ztest_func_t ztest_dmu_prealloc; ztest_func_t ztest_fzap; ztest_func_t ztest_dmu_snapshot_create_destroy; ztest_func_t ztest_dsl_prop_get_set; ztest_func_t ztest_spa_prop_get_set; ztest_func_t ztest_spa_create_destroy; ztest_func_t ztest_fault_inject; ztest_func_t ztest_ddt_repair; ztest_func_t ztest_dmu_snapshot_hold; ztest_func_t ztest_mmp_enable_disable; ztest_func_t ztest_spa_rename; ztest_func_t ztest_scrub; ztest_func_t ztest_dsl_dataset_promote_busy; ztest_func_t ztest_vdev_attach_detach; ztest_func_t ztest_vdev_LUN_growth; ztest_func_t ztest_vdev_add_remove; ztest_func_t ztest_vdev_aux_add_remove; ztest_func_t ztest_split_pool; ztest_func_t ztest_reguid; ztest_func_t ztest_spa_upgrade; ztest_func_t ztest_fletcher; ztest_func_t ztest_fletcher_incr; ztest_func_t ztest_verify_dnode_bt; uint64_t zopt_always = 0ULL * NANOSEC; /* all the time */ uint64_t zopt_incessant = 1ULL * NANOSEC / 10; /* every 1/10 second */ uint64_t zopt_often = 1ULL * NANOSEC; /* every second */ uint64_t zopt_sometimes = 10ULL * NANOSEC; /* every 10 seconds */ uint64_t zopt_rarely = 60ULL * NANOSEC; /* every 60 seconds */ #define ZTI_INIT(func, iters, interval) \ { .zi_func = (func), \ .zi_iters = (iters), \ .zi_interval = (interval), \ .zi_funcname = # func } ztest_info_t ztest_info[] = { ZTI_INIT(ztest_dmu_read_write, 1, &zopt_always), ZTI_INIT(ztest_dmu_write_parallel, 10, &zopt_always), ZTI_INIT(ztest_dmu_object_alloc_free, 1, &zopt_always), ZTI_INIT(ztest_dmu_commit_callbacks, 1, &zopt_always), ZTI_INIT(ztest_zap, 30, &zopt_always), ZTI_INIT(ztest_zap_parallel, 100, &zopt_always), ZTI_INIT(ztest_split_pool, 1, &zopt_always), ZTI_INIT(ztest_zil_commit, 1, &zopt_incessant), ZTI_INIT(ztest_zil_remount, 1, &zopt_sometimes), ZTI_INIT(ztest_dmu_read_write_zcopy, 1, &zopt_often), ZTI_INIT(ztest_dmu_objset_create_destroy, 1, &zopt_often), ZTI_INIT(ztest_dsl_prop_get_set, 1, &zopt_often), ZTI_INIT(ztest_spa_prop_get_set, 1, &zopt_sometimes), #if 0 ZTI_INIT(ztest_dmu_prealloc, 1, &zopt_sometimes), #endif ZTI_INIT(ztest_fzap, 1, &zopt_sometimes), ZTI_INIT(ztest_dmu_snapshot_create_destroy, 1, &zopt_sometimes), ZTI_INIT(ztest_spa_create_destroy, 1, &zopt_sometimes), ZTI_INIT(ztest_fault_inject, 1, &zopt_sometimes), ZTI_INIT(ztest_ddt_repair, 1, &zopt_sometimes), ZTI_INIT(ztest_dmu_snapshot_hold, 1, &zopt_sometimes), ZTI_INIT(ztest_mmp_enable_disable, 1, &zopt_sometimes), ZTI_INIT(ztest_reguid, 1, &zopt_rarely), ZTI_INIT(ztest_spa_rename, 1, &zopt_rarely), ZTI_INIT(ztest_scrub, 1, &zopt_rarely), ZTI_INIT(ztest_spa_upgrade, 1, &zopt_rarely), ZTI_INIT(ztest_dsl_dataset_promote_busy, 1, &zopt_rarely), ZTI_INIT(ztest_vdev_attach_detach, 1, &zopt_sometimes), ZTI_INIT(ztest_vdev_LUN_growth, 1, &zopt_rarely), ZTI_INIT(ztest_vdev_add_remove, 1, &ztest_opts.zo_vdevtime), ZTI_INIT(ztest_vdev_aux_add_remove, 1, &ztest_opts.zo_vdevtime), ZTI_INIT(ztest_fletcher, 1, &zopt_rarely), ZTI_INIT(ztest_fletcher_incr, 1, &zopt_rarely), ZTI_INIT(ztest_verify_dnode_bt, 1, &zopt_sometimes), }; #define ZTEST_FUNCS (sizeof (ztest_info) / sizeof (ztest_info_t)) /* * The following struct is used to hold a list of uncalled commit callbacks. * The callbacks are ordered by txg number. */ typedef struct ztest_cb_list { kmutex_t zcl_callbacks_lock; list_t zcl_callbacks; } ztest_cb_list_t; /* * Stuff we need to share writably between parent and child. */ typedef struct ztest_shared { boolean_t zs_do_init; hrtime_t zs_proc_start; hrtime_t zs_proc_stop; hrtime_t zs_thread_start; hrtime_t zs_thread_stop; hrtime_t zs_thread_kill; uint64_t zs_enospc_count; uint64_t zs_vdev_next_leaf; uint64_t zs_vdev_aux; uint64_t zs_alloc; uint64_t zs_space; uint64_t zs_splits; uint64_t zs_mirrors; uint64_t zs_metaslab_sz; uint64_t zs_metaslab_df_alloc_threshold; uint64_t zs_guid; } ztest_shared_t; #define ID_PARALLEL -1ULL static char ztest_dev_template[] = "%s/%s.%llua"; static char ztest_aux_template[] = "%s/%s.%s.%llu"; ztest_shared_t *ztest_shared; static spa_t *ztest_spa = NULL; static ztest_ds_t *ztest_ds; static kmutex_t ztest_vdev_lock; /* * The ztest_name_lock protects the pool and dataset namespace used by * the individual tests. To modify the namespace, consumers must grab * this lock as writer. Grabbing the lock as reader will ensure that the * namespace does not change while the lock is held. */ static rwlock_t ztest_name_lock; static boolean_t ztest_dump_core = B_TRUE; static boolean_t ztest_exiting; /* Global commit callback list */ static ztest_cb_list_t zcl; /* Commit cb delay */ static uint64_t zc_min_txg_delay = UINT64_MAX; static int zc_cb_counter = 0; /* * Minimum number of commit callbacks that need to be registered for us to check * whether the minimum txg delay is acceptable. */ #define ZTEST_COMMIT_CB_MIN_REG 100 /* * If a number of txgs equal to this threshold have been created after a commit * callback has been registered but not called, then we assume there is an * implementation bug. */ #define ZTEST_COMMIT_CB_THRESH (TXG_CONCURRENT_STATES + 1000) extern uint64_t metaslab_gang_bang; extern uint64_t metaslab_df_alloc_threshold; enum ztest_object { ZTEST_META_DNODE = 0, ZTEST_DIROBJ, ZTEST_OBJECTS }; static void usage(boolean_t) __NORETURN; /* * These libumem hooks provide a reasonable set of defaults for the allocator's * debugging facilities. */ const char * _umem_debug_init(void) { return ("default,verbose"); /* $UMEM_DEBUG setting */ } const char * _umem_logging_init(void) { return ("fail,contents"); /* $UMEM_LOGGING setting */ } #define BACKTRACE_SZ 100 static void sig_handler(int signo) { struct sigaction action; #ifdef __GLIBC__ /* backtrace() is a GNU extension */ int nptrs; void *buffer[BACKTRACE_SZ]; nptrs = backtrace(buffer, BACKTRACE_SZ); backtrace_symbols_fd(buffer, nptrs, STDERR_FILENO); #endif /* * Restore default action and re-raise signal so SIGSEGV and * SIGABRT can trigger a core dump. */ action.sa_handler = SIG_DFL; sigemptyset(&action.sa_mask); action.sa_flags = 0; (void) sigaction(signo, &action, NULL); raise(signo); } #define FATAL_MSG_SZ 1024 char *fatal_msg; static void fatal(int do_perror, char *message, ...) { va_list args; int save_errno = errno; char *buf; (void) fflush(stdout); buf = umem_alloc(FATAL_MSG_SZ, UMEM_NOFAIL); va_start(args, message); (void) sprintf(buf, "ztest: "); /* LINTED */ (void) vsprintf(buf + strlen(buf), message, args); va_end(args); if (do_perror) { (void) snprintf(buf + strlen(buf), FATAL_MSG_SZ - strlen(buf), ": %s", strerror(save_errno)); } (void) fprintf(stderr, "%s\n", buf); fatal_msg = buf; /* to ease debugging */ if (ztest_dump_core) abort(); exit(3); } static int str2shift(const char *buf) { const char *ends = "BKMGTPEZ"; int i; if (buf[0] == '\0') return (0); for (i = 0; i < strlen(ends); i++) { if (toupper(buf[0]) == ends[i]) break; } if (i == strlen(ends)) { (void) fprintf(stderr, "ztest: invalid bytes suffix: %s\n", buf); usage(B_FALSE); } if (buf[1] == '\0' || (toupper(buf[1]) == 'B' && buf[2] == '\0')) { return (10*i); } (void) fprintf(stderr, "ztest: invalid bytes suffix: %s\n", buf); usage(B_FALSE); /* NOTREACHED */ } static uint64_t nicenumtoull(const char *buf) { char *end; uint64_t val; val = strtoull(buf, &end, 0); if (end == buf) { (void) fprintf(stderr, "ztest: bad numeric value: %s\n", buf); usage(B_FALSE); } else if (end[0] == '.') { double fval = strtod(buf, &end); fval *= pow(2, str2shift(end)); if (fval > UINT64_MAX) { (void) fprintf(stderr, "ztest: value too large: %s\n", buf); usage(B_FALSE); } val = (uint64_t)fval; } else { int shift = str2shift(end); if (shift >= 64 || (val << shift) >> shift != val) { (void) fprintf(stderr, "ztest: value too large: %s\n", buf); usage(B_FALSE); } val <<= shift; } return (val); } static void usage(boolean_t requested) { const ztest_shared_opts_t *zo = &ztest_opts_defaults; char nice_vdev_size[NN_NUMBUF_SZ]; char nice_gang_bang[NN_NUMBUF_SZ]; FILE *fp = requested ? stdout : stderr; nicenum(zo->zo_vdev_size, nice_vdev_size, sizeof (nice_vdev_size)); nicenum(zo->zo_metaslab_gang_bang, nice_gang_bang, sizeof (nice_gang_bang)); (void) fprintf(fp, "Usage: %s\n" "\t[-v vdevs (default: %llu)]\n" "\t[-s size_of_each_vdev (default: %s)]\n" "\t[-a alignment_shift (default: %d)] use 0 for random\n" "\t[-m mirror_copies (default: %d)]\n" "\t[-r raidz_disks (default: %d)]\n" "\t[-R raidz_parity (default: %d)]\n" "\t[-d datasets (default: %d)]\n" "\t[-t threads (default: %d)]\n" "\t[-g gang_block_threshold (default: %s)]\n" "\t[-i init_count (default: %d)] initialize pool i times\n" "\t[-k kill_percentage (default: %llu%%)]\n" "\t[-p pool_name (default: %s)]\n" "\t[-f dir (default: %s)] file directory for vdev files\n" "\t[-M] Multi-host simulate pool imported on remote host\n" "\t[-V] verbose (use multiple times for ever more blather)\n" "\t[-E] use existing pool instead of creating new one\n" "\t[-T time (default: %llu sec)] total run time\n" "\t[-F freezeloops (default: %llu)] max loops in spa_freeze()\n" "\t[-P passtime (default: %llu sec)] time per pass\n" "\t[-B alt_ztest (default: )] alternate ztest path\n" "\t[-o variable=value] ... set global variable to an unsigned\n" "\t 32-bit integer value\n" "\t[-h] (print help)\n" "", zo->zo_pool, (u_longlong_t)zo->zo_vdevs, /* -v */ nice_vdev_size, /* -s */ zo->zo_ashift, /* -a */ zo->zo_mirrors, /* -m */ zo->zo_raidz, /* -r */ zo->zo_raidz_parity, /* -R */ zo->zo_datasets, /* -d */ zo->zo_threads, /* -t */ nice_gang_bang, /* -g */ zo->zo_init, /* -i */ (u_longlong_t)zo->zo_killrate, /* -k */ zo->zo_pool, /* -p */ zo->zo_dir, /* -f */ (u_longlong_t)zo->zo_time, /* -T */ (u_longlong_t)zo->zo_maxloops, /* -F */ (u_longlong_t)zo->zo_passtime); exit(requested ? 0 : 1); } static void process_options(int argc, char **argv) { char *path; ztest_shared_opts_t *zo = &ztest_opts; int opt; uint64_t value; char altdir[MAXNAMELEN] = { 0 }; bcopy(&ztest_opts_defaults, zo, sizeof (*zo)); while ((opt = getopt(argc, argv, "v:s:a:m:r:R:d:t:g:i:k:p:f:MVET:P:hF:B:o:")) != EOF) { value = 0; switch (opt) { case 'v': case 's': case 'a': case 'm': case 'r': case 'R': case 'd': case 't': case 'g': case 'i': case 'k': case 'T': case 'P': case 'F': value = nicenumtoull(optarg); } switch (opt) { case 'v': zo->zo_vdevs = value; break; case 's': zo->zo_vdev_size = MAX(SPA_MINDEVSIZE, value); break; case 'a': zo->zo_ashift = value; break; case 'm': zo->zo_mirrors = value; break; case 'r': zo->zo_raidz = MAX(1, value); break; case 'R': zo->zo_raidz_parity = MIN(MAX(value, 1), 3); break; case 'd': zo->zo_datasets = MAX(1, value); break; case 't': zo->zo_threads = MAX(1, value); break; case 'g': zo->zo_metaslab_gang_bang = MAX(SPA_MINBLOCKSIZE << 1, value); break; case 'i': zo->zo_init = value; break; case 'k': zo->zo_killrate = value; break; case 'p': (void) strlcpy(zo->zo_pool, optarg, sizeof (zo->zo_pool)); break; case 'f': path = realpath(optarg, NULL); if (path == NULL) { (void) fprintf(stderr, "error: %s: %s\n", optarg, strerror(errno)); usage(B_FALSE); } else { (void) strlcpy(zo->zo_dir, path, sizeof (zo->zo_dir)); free(path); } break; case 'M': zo->zo_mmp_test = 1; break; case 'V': zo->zo_verbose++; break; case 'E': zo->zo_init = 0; break; case 'T': zo->zo_time = value; break; case 'P': zo->zo_passtime = MAX(1, value); break; case 'F': zo->zo_maxloops = MAX(1, value); break; case 'B': (void) strlcpy(altdir, optarg, sizeof (altdir)); break; case 'o': if (set_global_var(optarg) != 0) usage(B_FALSE); break; case 'h': usage(B_TRUE); break; case '?': default: usage(B_FALSE); break; } } zo->zo_raidz_parity = MIN(zo->zo_raidz_parity, zo->zo_raidz - 1); zo->zo_vdevtime = (zo->zo_vdevs > 0 ? zo->zo_time * NANOSEC / zo->zo_vdevs : UINT64_MAX >> 2); if (strlen(altdir) > 0) { char *cmd; char *realaltdir; char *bin; char *ztest; char *isa; int isalen; cmd = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); realaltdir = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); VERIFY(NULL != realpath(getexecname(), cmd)); if (0 != access(altdir, F_OK)) { ztest_dump_core = B_FALSE; fatal(B_TRUE, "invalid alternate ztest path: %s", altdir); } VERIFY(NULL != realpath(altdir, realaltdir)); /* * 'cmd' should be of the form "/usr/bin//ztest". * We want to extract to determine if we should use * 32 or 64 bit binaries. */ bin = strstr(cmd, "/usr/bin/"); ztest = strstr(bin, "/ztest"); isa = bin + 9; isalen = ztest - isa; (void) snprintf(zo->zo_alt_ztest, sizeof (zo->zo_alt_ztest), "%s/usr/bin/%.*s/ztest", realaltdir, isalen, isa); (void) snprintf(zo->zo_alt_libpath, sizeof (zo->zo_alt_libpath), "%s/usr/lib/%.*s", realaltdir, isalen, isa); if (0 != access(zo->zo_alt_ztest, X_OK)) { ztest_dump_core = B_FALSE; fatal(B_TRUE, "invalid alternate ztest: %s", zo->zo_alt_ztest); } else if (0 != access(zo->zo_alt_libpath, X_OK)) { ztest_dump_core = B_FALSE; fatal(B_TRUE, "invalid alternate lib directory %s", zo->zo_alt_libpath); } umem_free(cmd, MAXPATHLEN); umem_free(realaltdir, MAXPATHLEN); } } static void ztest_kill(ztest_shared_t *zs) { zs->zs_alloc = metaslab_class_get_alloc(spa_normal_class(ztest_spa)); zs->zs_space = metaslab_class_get_space(spa_normal_class(ztest_spa)); /* * Before we kill off ztest, make sure that the config is updated. * See comment above spa_config_sync(). */ mutex_enter(&spa_namespace_lock); spa_config_sync(ztest_spa, B_FALSE, B_FALSE); mutex_exit(&spa_namespace_lock); (void) kill(getpid(), SIGKILL); } static uint64_t ztest_random(uint64_t range) { uint64_t r; ASSERT3S(ztest_fd_rand, >=, 0); if (range == 0) return (0); if (read(ztest_fd_rand, &r, sizeof (r)) != sizeof (r)) fatal(1, "short read from /dev/urandom"); return (r % range); } /* ARGSUSED */ static void ztest_record_enospc(const char *s) { ztest_shared->zs_enospc_count++; } static uint64_t ztest_get_ashift(void) { if (ztest_opts.zo_ashift == 0) return (SPA_MINBLOCKSHIFT + ztest_random(5)); return (ztest_opts.zo_ashift); } static nvlist_t * make_vdev_file(char *path, char *aux, char *pool, size_t size, uint64_t ashift) { char *pathbuf; uint64_t vdev; nvlist_t *file; pathbuf = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); if (ashift == 0) ashift = ztest_get_ashift(); if (path == NULL) { path = pathbuf; if (aux != NULL) { vdev = ztest_shared->zs_vdev_aux; (void) snprintf(path, MAXPATHLEN, ztest_aux_template, ztest_opts.zo_dir, pool == NULL ? ztest_opts.zo_pool : pool, aux, vdev); } else { vdev = ztest_shared->zs_vdev_next_leaf++; (void) snprintf(path, MAXPATHLEN, ztest_dev_template, ztest_opts.zo_dir, pool == NULL ? ztest_opts.zo_pool : pool, vdev); } } if (size != 0) { int fd = open(path, O_RDWR | O_CREAT | O_TRUNC, 0666); if (fd == -1) fatal(1, "can't open %s", path); if (ftruncate(fd, size) != 0) fatal(1, "can't ftruncate %s", path); (void) close(fd); } VERIFY(nvlist_alloc(&file, NV_UNIQUE_NAME, 0) == 0); VERIFY(nvlist_add_string(file, ZPOOL_CONFIG_TYPE, VDEV_TYPE_FILE) == 0); VERIFY(nvlist_add_string(file, ZPOOL_CONFIG_PATH, path) == 0); VERIFY(nvlist_add_uint64(file, ZPOOL_CONFIG_ASHIFT, ashift) == 0); umem_free(pathbuf, MAXPATHLEN); return (file); } static nvlist_t * make_vdev_raidz(char *path, char *aux, char *pool, size_t size, uint64_t ashift, int r) { nvlist_t *raidz, **child; int c; if (r < 2) return (make_vdev_file(path, aux, pool, size, ashift)); child = umem_alloc(r * sizeof (nvlist_t *), UMEM_NOFAIL); for (c = 0; c < r; c++) child[c] = make_vdev_file(path, aux, pool, size, ashift); VERIFY(nvlist_alloc(&raidz, NV_UNIQUE_NAME, 0) == 0); VERIFY(nvlist_add_string(raidz, ZPOOL_CONFIG_TYPE, VDEV_TYPE_RAIDZ) == 0); VERIFY(nvlist_add_uint64(raidz, ZPOOL_CONFIG_NPARITY, ztest_opts.zo_raidz_parity) == 0); VERIFY(nvlist_add_nvlist_array(raidz, ZPOOL_CONFIG_CHILDREN, child, r) == 0); for (c = 0; c < r; c++) nvlist_free(child[c]); umem_free(child, r * sizeof (nvlist_t *)); return (raidz); } static nvlist_t * make_vdev_mirror(char *path, char *aux, char *pool, size_t size, uint64_t ashift, int r, int m) { nvlist_t *mirror, **child; int c; if (m < 1) return (make_vdev_raidz(path, aux, pool, size, ashift, r)); child = umem_alloc(m * sizeof (nvlist_t *), UMEM_NOFAIL); for (c = 0; c < m; c++) child[c] = make_vdev_raidz(path, aux, pool, size, ashift, r); VERIFY(nvlist_alloc(&mirror, NV_UNIQUE_NAME, 0) == 0); VERIFY(nvlist_add_string(mirror, ZPOOL_CONFIG_TYPE, VDEV_TYPE_MIRROR) == 0); VERIFY(nvlist_add_nvlist_array(mirror, ZPOOL_CONFIG_CHILDREN, child, m) == 0); for (c = 0; c < m; c++) nvlist_free(child[c]); umem_free(child, m * sizeof (nvlist_t *)); return (mirror); } static nvlist_t * make_vdev_root(char *path, char *aux, char *pool, size_t size, uint64_t ashift, int log, int r, int m, int t) { nvlist_t *root, **child; int c; ASSERT(t > 0); child = umem_alloc(t * sizeof (nvlist_t *), UMEM_NOFAIL); for (c = 0; c < t; c++) { child[c] = make_vdev_mirror(path, aux, pool, size, ashift, r, m); VERIFY(nvlist_add_uint64(child[c], ZPOOL_CONFIG_IS_LOG, log) == 0); } VERIFY(nvlist_alloc(&root, NV_UNIQUE_NAME, 0) == 0); VERIFY(nvlist_add_string(root, ZPOOL_CONFIG_TYPE, VDEV_TYPE_ROOT) == 0); VERIFY(nvlist_add_nvlist_array(root, aux ? aux : ZPOOL_CONFIG_CHILDREN, child, t) == 0); for (c = 0; c < t; c++) nvlist_free(child[c]); umem_free(child, t * sizeof (nvlist_t *)); return (root); } /* * Find a random spa version. Returns back a random spa version in the * range [initial_version, SPA_VERSION_FEATURES]. */ static uint64_t ztest_random_spa_version(uint64_t initial_version) { uint64_t version = initial_version; if (version <= SPA_VERSION_BEFORE_FEATURES) { version = version + ztest_random(SPA_VERSION_BEFORE_FEATURES - version + 1); } if (version > SPA_VERSION_BEFORE_FEATURES) version = SPA_VERSION_FEATURES; ASSERT(SPA_VERSION_IS_SUPPORTED(version)); return (version); } static int ztest_random_blocksize(void) { /* * Choose a block size >= the ashift. * If the SPA supports new MAXBLOCKSIZE, test up to 1MB blocks. */ int maxbs = SPA_OLD_MAXBLOCKSHIFT; if (spa_maxblocksize(ztest_spa) == SPA_MAXBLOCKSIZE) maxbs = 20; uint64_t block_shift = ztest_random(maxbs - ztest_spa->spa_max_ashift + 1); return (1 << (SPA_MINBLOCKSHIFT + block_shift)); } static int ztest_random_dnodesize(void) { int slots; int max_slots = spa_maxdnodesize(ztest_spa) >> DNODE_SHIFT; if (max_slots == DNODE_MIN_SLOTS) return (DNODE_MIN_SIZE); /* * Weight the random distribution more heavily toward smaller * dnode sizes since that is more likely to reflect real-world * usage. */ ASSERT3U(max_slots, >, 4); switch (ztest_random(10)) { case 0: slots = 5 + ztest_random(max_slots - 4); break; case 1 ... 4: slots = 2 + ztest_random(3); break; default: slots = 1; break; } return (slots << DNODE_SHIFT); } static int ztest_random_ibshift(void) { return (DN_MIN_INDBLKSHIFT + ztest_random(DN_MAX_INDBLKSHIFT - DN_MIN_INDBLKSHIFT + 1)); } static uint64_t ztest_random_vdev_top(spa_t *spa, boolean_t log_ok) { uint64_t top; vdev_t *rvd = spa->spa_root_vdev; vdev_t *tvd; ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0); do { top = ztest_random(rvd->vdev_children); tvd = rvd->vdev_child[top]; } while (tvd->vdev_ishole || (tvd->vdev_islog && !log_ok) || tvd->vdev_mg == NULL || tvd->vdev_mg->mg_class == NULL); return (top); } static uint64_t ztest_random_dsl_prop(zfs_prop_t prop) { uint64_t value; do { value = zfs_prop_random_value(prop, ztest_random(-1ULL)); } while (prop == ZFS_PROP_CHECKSUM && value == ZIO_CHECKSUM_OFF); return (value); } static int ztest_dsl_prop_set_uint64(char *osname, zfs_prop_t prop, uint64_t value, boolean_t inherit) { const char *propname = zfs_prop_to_name(prop); const char *valname; char *setpoint; uint64_t curval; int error; error = dsl_prop_set_int(osname, propname, (inherit ? ZPROP_SRC_NONE : ZPROP_SRC_LOCAL), value); if (error == ENOSPC) { ztest_record_enospc(FTAG); return (error); } ASSERT0(error); setpoint = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); VERIFY0(dsl_prop_get_integer(osname, propname, &curval, setpoint)); if (ztest_opts.zo_verbose >= 6) { int err; err = zfs_prop_index_to_string(prop, curval, &valname); if (err) (void) printf("%s %s = %llu at '%s'\n", osname, propname, (unsigned long long)curval, setpoint); else (void) printf("%s %s = %s at '%s'\n", osname, propname, valname, setpoint); } umem_free(setpoint, MAXPATHLEN); return (error); } static int ztest_spa_prop_set_uint64(zpool_prop_t prop, uint64_t value) { spa_t *spa = ztest_spa; nvlist_t *props = NULL; int error; VERIFY(nvlist_alloc(&props, NV_UNIQUE_NAME, 0) == 0); VERIFY(nvlist_add_uint64(props, zpool_prop_to_name(prop), value) == 0); error = spa_prop_set(spa, props); nvlist_free(props); if (error == ENOSPC) { ztest_record_enospc(FTAG); return (error); } ASSERT0(error); return (error); } static int ztest_dmu_objset_own(const char *name, dmu_objset_type_t type, boolean_t readonly, boolean_t decrypt, void *tag, objset_t **osp) { int err; err = dmu_objset_own(name, type, readonly, decrypt, tag, osp); if (decrypt && err == EACCES) { char ddname[ZFS_MAX_DATASET_NAME_LEN]; dsl_crypto_params_t *dcp; nvlist_t *crypto_args = fnvlist_alloc(); char *cp = NULL; /* spa_keystore_load_wkey() expects a dsl dir name */ strcpy(ddname, name); cp = strchr(ddname, '@'); if (cp != NULL) *cp = '\0'; fnvlist_add_uint8_array(crypto_args, "wkeydata", (uint8_t *)ztest_wkeydata, WRAPPING_KEY_LEN); VERIFY0(dsl_crypto_params_create_nvlist(DCP_CMD_NONE, NULL, crypto_args, &dcp)); err = spa_keystore_load_wkey(ddname, dcp, B_FALSE); dsl_crypto_params_free(dcp, B_FALSE); fnvlist_free(crypto_args); if (err != 0) return (err); err = dmu_objset_own(name, type, readonly, decrypt, tag, osp); } return (err); } /* * Object and range lock mechanics */ typedef struct { list_node_t z_lnode; refcount_t z_refcnt; uint64_t z_object; zfs_rlock_t z_range_lock; } ztest_znode_t; typedef struct { rl_t *z_rl; ztest_znode_t *z_ztznode; } ztest_zrl_t; static ztest_znode_t * ztest_znode_init(uint64_t object) { ztest_znode_t *zp = umem_alloc(sizeof (*zp), UMEM_NOFAIL); list_link_init(&zp->z_lnode); refcount_create(&zp->z_refcnt); zp->z_object = object; zfs_rlock_init(&zp->z_range_lock); return (zp); } static void ztest_znode_fini(ztest_znode_t *zp) { ASSERT(refcount_is_zero(&zp->z_refcnt)); zfs_rlock_destroy(&zp->z_range_lock); zp->z_object = 0; refcount_destroy(&zp->z_refcnt); list_link_init(&zp->z_lnode); umem_free(zp, sizeof (*zp)); } static void ztest_zll_init(zll_t *zll) { mutex_init(&zll->z_lock, NULL, MUTEX_DEFAULT, NULL); list_create(&zll->z_list, sizeof (ztest_znode_t), offsetof(ztest_znode_t, z_lnode)); } static void ztest_zll_destroy(zll_t *zll) { list_destroy(&zll->z_list); mutex_destroy(&zll->z_lock); } #define RL_TAG "range_lock" static ztest_znode_t * ztest_znode_get(ztest_ds_t *zd, uint64_t object) { zll_t *zll = &zd->zd_range_lock[object & (ZTEST_OBJECT_LOCKS - 1)]; ztest_znode_t *zp = NULL; mutex_enter(&zll->z_lock); for (zp = list_head(&zll->z_list); (zp); zp = list_next(&zll->z_list, zp)) { if (zp->z_object == object) { refcount_add(&zp->z_refcnt, RL_TAG); break; } } if (zp == NULL) { zp = ztest_znode_init(object); refcount_add(&zp->z_refcnt, RL_TAG); list_insert_head(&zll->z_list, zp); } mutex_exit(&zll->z_lock); return (zp); } static void ztest_znode_put(ztest_ds_t *zd, ztest_znode_t *zp) { zll_t *zll = NULL; ASSERT3U(zp->z_object, !=, 0); zll = &zd->zd_range_lock[zp->z_object & (ZTEST_OBJECT_LOCKS - 1)]; mutex_enter(&zll->z_lock); refcount_remove(&zp->z_refcnt, RL_TAG); if (refcount_is_zero(&zp->z_refcnt)) { list_remove(&zll->z_list, zp); ztest_znode_fini(zp); } mutex_exit(&zll->z_lock); } static void ztest_rll_init(rll_t *rll) { rll->rll_writer = NULL; rll->rll_readers = 0; mutex_init(&rll->rll_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&rll->rll_cv, NULL, CV_DEFAULT, NULL); } static void ztest_rll_destroy(rll_t *rll) { ASSERT(rll->rll_writer == NULL); ASSERT(rll->rll_readers == 0); mutex_destroy(&rll->rll_lock); cv_destroy(&rll->rll_cv); } static void ztest_rll_lock(rll_t *rll, rl_type_t type) { mutex_enter(&rll->rll_lock); if (type == RL_READER) { while (rll->rll_writer != NULL) (void) cv_wait(&rll->rll_cv, &rll->rll_lock); rll->rll_readers++; } else { while (rll->rll_writer != NULL || rll->rll_readers) (void) cv_wait(&rll->rll_cv, &rll->rll_lock); rll->rll_writer = curthread; } mutex_exit(&rll->rll_lock); } static void ztest_rll_unlock(rll_t *rll) { mutex_enter(&rll->rll_lock); if (rll->rll_writer) { ASSERT(rll->rll_readers == 0); rll->rll_writer = NULL; } else { ASSERT(rll->rll_readers != 0); ASSERT(rll->rll_writer == NULL); rll->rll_readers--; } if (rll->rll_writer == NULL && rll->rll_readers == 0) cv_broadcast(&rll->rll_cv); mutex_exit(&rll->rll_lock); } static void ztest_object_lock(ztest_ds_t *zd, uint64_t object, rl_type_t type) { rll_t *rll = &zd->zd_object_lock[object & (ZTEST_OBJECT_LOCKS - 1)]; ztest_rll_lock(rll, type); } static void ztest_object_unlock(ztest_ds_t *zd, uint64_t object) { rll_t *rll = &zd->zd_object_lock[object & (ZTEST_OBJECT_LOCKS - 1)]; ztest_rll_unlock(rll); } static ztest_zrl_t * ztest_zrl_init(rl_t *rl, ztest_znode_t *zp) { ztest_zrl_t *zrl = umem_alloc(sizeof (*zrl), UMEM_NOFAIL); zrl->z_rl = rl; zrl->z_ztznode = zp; return (zrl); } static void ztest_zrl_fini(ztest_zrl_t *zrl) { umem_free(zrl, sizeof (*zrl)); } static ztest_zrl_t * ztest_range_lock(ztest_ds_t *zd, uint64_t object, uint64_t offset, uint64_t size, rl_type_t type) { ztest_znode_t *zp = ztest_znode_get(zd, object); rl_t *rl = zfs_range_lock(&zp->z_range_lock, offset, size, type); return (ztest_zrl_init(rl, zp)); } static void ztest_range_unlock(ztest_ds_t *zd, ztest_zrl_t *zrl) { zfs_range_unlock(zrl->z_rl); ztest_znode_put(zd, zrl->z_ztznode); ztest_zrl_fini(zrl); } static void ztest_zd_init(ztest_ds_t *zd, ztest_shared_ds_t *szd, objset_t *os) { zd->zd_os = os; zd->zd_zilog = dmu_objset_zil(os); zd->zd_shared = szd; dmu_objset_name(os, zd->zd_name); int l; if (zd->zd_shared != NULL) zd->zd_shared->zd_seq = 0; VERIFY(rwlock_init(&zd->zd_zilog_lock, USYNC_THREAD, NULL) == 0); mutex_init(&zd->zd_dirobj_lock, NULL, MUTEX_DEFAULT, NULL); for (l = 0; l < ZTEST_OBJECT_LOCKS; l++) ztest_rll_init(&zd->zd_object_lock[l]); for (l = 0; l < ZTEST_RANGE_LOCKS; l++) ztest_zll_init(&zd->zd_range_lock[l]); } static void ztest_zd_fini(ztest_ds_t *zd) { int l; mutex_destroy(&zd->zd_dirobj_lock); (void) rwlock_destroy(&zd->zd_zilog_lock); for (l = 0; l < ZTEST_OBJECT_LOCKS; l++) ztest_rll_destroy(&zd->zd_object_lock[l]); for (l = 0; l < ZTEST_RANGE_LOCKS; l++) ztest_zll_destroy(&zd->zd_range_lock[l]); } #define TXG_MIGHTWAIT (ztest_random(10) == 0 ? TXG_NOWAIT : TXG_WAIT) static uint64_t ztest_tx_assign(dmu_tx_t *tx, uint64_t txg_how, const char *tag) { uint64_t txg; int error; /* * Attempt to assign tx to some transaction group. */ error = dmu_tx_assign(tx, txg_how); if (error) { if (error == ERESTART) { ASSERT(txg_how == TXG_NOWAIT); dmu_tx_wait(tx); } else { ASSERT3U(error, ==, ENOSPC); ztest_record_enospc(tag); } dmu_tx_abort(tx); return (0); } txg = dmu_tx_get_txg(tx); ASSERT(txg != 0); return (txg); } static void ztest_pattern_set(void *buf, uint64_t size, uint64_t value) { uint64_t *ip = buf; uint64_t *ip_end = (uint64_t *)((uintptr_t)buf + (uintptr_t)size); while (ip < ip_end) *ip++ = value; } #ifndef NDEBUG static boolean_t ztest_pattern_match(void *buf, uint64_t size, uint64_t value) { uint64_t *ip = buf; uint64_t *ip_end = (uint64_t *)((uintptr_t)buf + (uintptr_t)size); uint64_t diff = 0; while (ip < ip_end) diff |= (value - *ip++); return (diff == 0); } #endif static void ztest_bt_generate(ztest_block_tag_t *bt, objset_t *os, uint64_t object, uint64_t dnodesize, uint64_t offset, uint64_t gen, uint64_t txg, uint64_t crtxg) { bt->bt_magic = BT_MAGIC; bt->bt_objset = dmu_objset_id(os); bt->bt_object = object; bt->bt_dnodesize = dnodesize; bt->bt_offset = offset; bt->bt_gen = gen; bt->bt_txg = txg; bt->bt_crtxg = crtxg; } static void ztest_bt_verify(ztest_block_tag_t *bt, objset_t *os, uint64_t object, uint64_t dnodesize, uint64_t offset, uint64_t gen, uint64_t txg, uint64_t crtxg) { ASSERT3U(bt->bt_magic, ==, BT_MAGIC); ASSERT3U(bt->bt_objset, ==, dmu_objset_id(os)); ASSERT3U(bt->bt_object, ==, object); ASSERT3U(bt->bt_dnodesize, ==, dnodesize); ASSERT3U(bt->bt_offset, ==, offset); ASSERT3U(bt->bt_gen, <=, gen); ASSERT3U(bt->bt_txg, <=, txg); ASSERT3U(bt->bt_crtxg, ==, crtxg); } static ztest_block_tag_t * ztest_bt_bonus(dmu_buf_t *db) { dmu_object_info_t doi; ztest_block_tag_t *bt; dmu_object_info_from_db(db, &doi); ASSERT3U(doi.doi_bonus_size, <=, db->db_size); ASSERT3U(doi.doi_bonus_size, >=, sizeof (*bt)); bt = (void *)((char *)db->db_data + doi.doi_bonus_size - sizeof (*bt)); return (bt); } /* * Generate a token to fill up unused bonus buffer space. Try to make * it unique to the object, generation, and offset to verify that data * is not getting overwritten by data from other dnodes. */ #define ZTEST_BONUS_FILL_TOKEN(obj, ds, gen, offset) \ (((ds) << 48) | ((gen) << 32) | ((obj) << 8) | (offset)) /* * Fill up the unused bonus buffer region before the block tag with a * verifiable pattern. Filling the whole bonus area with non-zero data * helps ensure that all dnode traversal code properly skips the * interior regions of large dnodes. */ void ztest_fill_unused_bonus(dmu_buf_t *db, void *end, uint64_t obj, objset_t *os, uint64_t gen) { uint64_t *bonusp; ASSERT(IS_P2ALIGNED((char *)end - (char *)db->db_data, 8)); for (bonusp = db->db_data; bonusp < (uint64_t *)end; bonusp++) { uint64_t token = ZTEST_BONUS_FILL_TOKEN(obj, dmu_objset_id(os), gen, bonusp - (uint64_t *)db->db_data); *bonusp = token; } } /* * Verify that the unused area of a bonus buffer is filled with the * expected tokens. */ void ztest_verify_unused_bonus(dmu_buf_t *db, void *end, uint64_t obj, objset_t *os, uint64_t gen) { uint64_t *bonusp; for (bonusp = db->db_data; bonusp < (uint64_t *)end; bonusp++) { uint64_t token = ZTEST_BONUS_FILL_TOKEN(obj, dmu_objset_id(os), gen, bonusp - (uint64_t *)db->db_data); VERIFY3U(*bonusp, ==, token); } } /* * ZIL logging ops */ #define lrz_type lr_mode #define lrz_blocksize lr_uid #define lrz_ibshift lr_gid #define lrz_bonustype lr_rdev #define lrz_dnodesize lr_crtime[1] static void ztest_log_create(ztest_ds_t *zd, dmu_tx_t *tx, lr_create_t *lr) { char *name = (void *)(lr + 1); /* name follows lr */ size_t namesize = strlen(name) + 1; itx_t *itx; if (zil_replaying(zd->zd_zilog, tx)) return; itx = zil_itx_create(TX_CREATE, sizeof (*lr) + namesize); bcopy(&lr->lr_common + 1, &itx->itx_lr + 1, sizeof (*lr) + namesize - sizeof (lr_t)); zil_itx_assign(zd->zd_zilog, itx, tx); } static void ztest_log_remove(ztest_ds_t *zd, dmu_tx_t *tx, lr_remove_t *lr, uint64_t object) { char *name = (void *)(lr + 1); /* name follows lr */ size_t namesize = strlen(name) + 1; itx_t *itx; if (zil_replaying(zd->zd_zilog, tx)) return; itx = zil_itx_create(TX_REMOVE, sizeof (*lr) + namesize); bcopy(&lr->lr_common + 1, &itx->itx_lr + 1, sizeof (*lr) + namesize - sizeof (lr_t)); itx->itx_oid = object; zil_itx_assign(zd->zd_zilog, itx, tx); } static void ztest_log_write(ztest_ds_t *zd, dmu_tx_t *tx, lr_write_t *lr) { itx_t *itx; itx_wr_state_t write_state = ztest_random(WR_NUM_STATES); if (zil_replaying(zd->zd_zilog, tx)) return; if (lr->lr_length > ZIL_MAX_LOG_DATA) write_state = WR_INDIRECT; itx = zil_itx_create(TX_WRITE, sizeof (*lr) + (write_state == WR_COPIED ? lr->lr_length : 0)); if (write_state == WR_COPIED && dmu_read(zd->zd_os, lr->lr_foid, lr->lr_offset, lr->lr_length, ((lr_write_t *)&itx->itx_lr) + 1, DMU_READ_NO_PREFETCH) != 0) { zil_itx_destroy(itx); itx = zil_itx_create(TX_WRITE, sizeof (*lr)); write_state = WR_NEED_COPY; } itx->itx_private = zd; itx->itx_wr_state = write_state; itx->itx_sync = (ztest_random(8) == 0); bcopy(&lr->lr_common + 1, &itx->itx_lr + 1, sizeof (*lr) - sizeof (lr_t)); zil_itx_assign(zd->zd_zilog, itx, tx); } static void ztest_log_truncate(ztest_ds_t *zd, dmu_tx_t *tx, lr_truncate_t *lr) { itx_t *itx; if (zil_replaying(zd->zd_zilog, tx)) return; itx = zil_itx_create(TX_TRUNCATE, sizeof (*lr)); bcopy(&lr->lr_common + 1, &itx->itx_lr + 1, sizeof (*lr) - sizeof (lr_t)); itx->itx_sync = B_FALSE; zil_itx_assign(zd->zd_zilog, itx, tx); } static void ztest_log_setattr(ztest_ds_t *zd, dmu_tx_t *tx, lr_setattr_t *lr) { itx_t *itx; if (zil_replaying(zd->zd_zilog, tx)) return; itx = zil_itx_create(TX_SETATTR, sizeof (*lr)); bcopy(&lr->lr_common + 1, &itx->itx_lr + 1, sizeof (*lr) - sizeof (lr_t)); itx->itx_sync = B_FALSE; zil_itx_assign(zd->zd_zilog, itx, tx); } /* * ZIL replay ops */ static int ztest_replay_create(void *arg1, void *arg2, boolean_t byteswap) { ztest_ds_t *zd = arg1; lr_create_t *lr = arg2; char *name = (void *)(lr + 1); /* name follows lr */ objset_t *os = zd->zd_os; ztest_block_tag_t *bbt; dmu_buf_t *db; dmu_tx_t *tx; uint64_t txg; int error = 0; int bonuslen; if (byteswap) byteswap_uint64_array(lr, sizeof (*lr)); ASSERT(lr->lr_doid == ZTEST_DIROBJ); ASSERT(name[0] != '\0'); tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, lr->lr_doid, B_TRUE, name); if (lr->lrz_type == DMU_OT_ZAP_OTHER) { dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, B_TRUE, NULL); } else { dmu_tx_hold_bonus(tx, DMU_NEW_OBJECT); } txg = ztest_tx_assign(tx, TXG_WAIT, FTAG); if (txg == 0) return (ENOSPC); ASSERT(dmu_objset_zil(os)->zl_replay == !!lr->lr_foid); bonuslen = DN_BONUS_SIZE(lr->lrz_dnodesize); if (lr->lrz_type == DMU_OT_ZAP_OTHER) { if (lr->lr_foid == 0) { lr->lr_foid = zap_create_dnsize(os, lr->lrz_type, lr->lrz_bonustype, bonuslen, lr->lrz_dnodesize, tx); } else { error = zap_create_claim_dnsize(os, lr->lr_foid, lr->lrz_type, lr->lrz_bonustype, bonuslen, lr->lrz_dnodesize, tx); } } else { if (lr->lr_foid == 0) { lr->lr_foid = dmu_object_alloc_dnsize(os, lr->lrz_type, 0, lr->lrz_bonustype, bonuslen, lr->lrz_dnodesize, tx); } else { error = dmu_object_claim_dnsize(os, lr->lr_foid, lr->lrz_type, 0, lr->lrz_bonustype, bonuslen, lr->lrz_dnodesize, tx); } } if (error) { ASSERT3U(error, ==, EEXIST); ASSERT(zd->zd_zilog->zl_replay); dmu_tx_commit(tx); return (error); } ASSERT(lr->lr_foid != 0); if (lr->lrz_type != DMU_OT_ZAP_OTHER) VERIFY3U(0, ==, dmu_object_set_blocksize(os, lr->lr_foid, lr->lrz_blocksize, lr->lrz_ibshift, tx)); VERIFY3U(0, ==, dmu_bonus_hold(os, lr->lr_foid, FTAG, &db)); bbt = ztest_bt_bonus(db); dmu_buf_will_dirty(db, tx); ztest_bt_generate(bbt, os, lr->lr_foid, lr->lrz_dnodesize, -1ULL, lr->lr_gen, txg, txg); ztest_fill_unused_bonus(db, bbt, lr->lr_foid, os, lr->lr_gen); dmu_buf_rele(db, FTAG); VERIFY3U(0, ==, zap_add(os, lr->lr_doid, name, sizeof (uint64_t), 1, &lr->lr_foid, tx)); (void) ztest_log_create(zd, tx, lr); dmu_tx_commit(tx); return (0); } static int ztest_replay_remove(void *arg1, void *arg2, boolean_t byteswap) { ztest_ds_t *zd = arg1; lr_remove_t *lr = arg2; char *name = (void *)(lr + 1); /* name follows lr */ objset_t *os = zd->zd_os; dmu_object_info_t doi; dmu_tx_t *tx; uint64_t object, txg; if (byteswap) byteswap_uint64_array(lr, sizeof (*lr)); ASSERT(lr->lr_doid == ZTEST_DIROBJ); ASSERT(name[0] != '\0'); VERIFY3U(0, ==, zap_lookup(os, lr->lr_doid, name, sizeof (object), 1, &object)); ASSERT(object != 0); ztest_object_lock(zd, object, RL_WRITER); VERIFY3U(0, ==, dmu_object_info(os, object, &doi)); tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, lr->lr_doid, B_FALSE, name); dmu_tx_hold_free(tx, object, 0, DMU_OBJECT_END); txg = ztest_tx_assign(tx, TXG_WAIT, FTAG); if (txg == 0) { ztest_object_unlock(zd, object); return (ENOSPC); } if (doi.doi_type == DMU_OT_ZAP_OTHER) { VERIFY3U(0, ==, zap_destroy(os, object, tx)); } else { VERIFY3U(0, ==, dmu_object_free(os, object, tx)); } VERIFY3U(0, ==, zap_remove(os, lr->lr_doid, name, tx)); (void) ztest_log_remove(zd, tx, lr, object); dmu_tx_commit(tx); ztest_object_unlock(zd, object); return (0); } static int ztest_replay_write(void *arg1, void *arg2, boolean_t byteswap) { ztest_ds_t *zd = arg1; lr_write_t *lr = arg2; objset_t *os = zd->zd_os; void *data = lr + 1; /* data follows lr */ uint64_t offset, length; ztest_block_tag_t *bt = data; ztest_block_tag_t *bbt; uint64_t gen, txg, lrtxg, crtxg; dmu_object_info_t doi; dmu_tx_t *tx; dmu_buf_t *db; arc_buf_t *abuf = NULL; ztest_zrl_t *rl; if (byteswap) byteswap_uint64_array(lr, sizeof (*lr)); offset = lr->lr_offset; length = lr->lr_length; /* If it's a dmu_sync() block, write the whole block */ if (lr->lr_common.lrc_reclen == sizeof (lr_write_t)) { uint64_t blocksize = BP_GET_LSIZE(&lr->lr_blkptr); if (length < blocksize) { offset -= offset % blocksize; length = blocksize; } } if (bt->bt_magic == BSWAP_64(BT_MAGIC)) byteswap_uint64_array(bt, sizeof (*bt)); if (bt->bt_magic != BT_MAGIC) bt = NULL; ztest_object_lock(zd, lr->lr_foid, RL_READER); rl = ztest_range_lock(zd, lr->lr_foid, offset, length, RL_WRITER); VERIFY3U(0, ==, dmu_bonus_hold(os, lr->lr_foid, FTAG, &db)); dmu_object_info_from_db(db, &doi); bbt = ztest_bt_bonus(db); ASSERT3U(bbt->bt_magic, ==, BT_MAGIC); gen = bbt->bt_gen; crtxg = bbt->bt_crtxg; lrtxg = lr->lr_common.lrc_txg; tx = dmu_tx_create(os); dmu_tx_hold_write(tx, lr->lr_foid, offset, length); if (ztest_random(8) == 0 && length == doi.doi_data_block_size && P2PHASE(offset, length) == 0) abuf = dmu_request_arcbuf(db, length); txg = ztest_tx_assign(tx, TXG_WAIT, FTAG); if (txg == 0) { if (abuf != NULL) dmu_return_arcbuf(abuf); dmu_buf_rele(db, FTAG); ztest_range_unlock(zd, rl); ztest_object_unlock(zd, lr->lr_foid); return (ENOSPC); } if (bt != NULL) { /* * Usually, verify the old data before writing new data -- * but not always, because we also want to verify correct * behavior when the data was not recently read into cache. */ ASSERT(offset % doi.doi_data_block_size == 0); if (ztest_random(4) != 0) { int prefetch = ztest_random(2) ? DMU_READ_PREFETCH : DMU_READ_NO_PREFETCH; ztest_block_tag_t rbt; VERIFY(dmu_read(os, lr->lr_foid, offset, sizeof (rbt), &rbt, prefetch) == 0); if (rbt.bt_magic == BT_MAGIC) { ztest_bt_verify(&rbt, os, lr->lr_foid, 0, offset, gen, txg, crtxg); } } /* * Writes can appear to be newer than the bonus buffer because * the ztest_get_data() callback does a dmu_read() of the * open-context data, which may be different than the data * as it was when the write was generated. */ if (zd->zd_zilog->zl_replay) { ztest_bt_verify(bt, os, lr->lr_foid, 0, offset, MAX(gen, bt->bt_gen), MAX(txg, lrtxg), bt->bt_crtxg); } /* * Set the bt's gen/txg to the bonus buffer's gen/txg * so that all of the usual ASSERTs will work. */ ztest_bt_generate(bt, os, lr->lr_foid, 0, offset, gen, txg, crtxg); } if (abuf == NULL) { dmu_write(os, lr->lr_foid, offset, length, data, tx); } else { bcopy(data, abuf->b_data, length); dmu_assign_arcbuf_by_dbuf(db, offset, abuf, tx); } (void) ztest_log_write(zd, tx, lr); dmu_buf_rele(db, FTAG); dmu_tx_commit(tx); ztest_range_unlock(zd, rl); ztest_object_unlock(zd, lr->lr_foid); return (0); } static int ztest_replay_truncate(void *arg1, void *arg2, boolean_t byteswap) { ztest_ds_t *zd = arg1; lr_truncate_t *lr = arg2; objset_t *os = zd->zd_os; dmu_tx_t *tx; uint64_t txg; ztest_zrl_t *rl; if (byteswap) byteswap_uint64_array(lr, sizeof (*lr)); ztest_object_lock(zd, lr->lr_foid, RL_READER); rl = ztest_range_lock(zd, lr->lr_foid, lr->lr_offset, lr->lr_length, RL_WRITER); tx = dmu_tx_create(os); dmu_tx_hold_free(tx, lr->lr_foid, lr->lr_offset, lr->lr_length); txg = ztest_tx_assign(tx, TXG_WAIT, FTAG); if (txg == 0) { ztest_range_unlock(zd, rl); ztest_object_unlock(zd, lr->lr_foid); return (ENOSPC); } VERIFY(dmu_free_range(os, lr->lr_foid, lr->lr_offset, lr->lr_length, tx) == 0); (void) ztest_log_truncate(zd, tx, lr); dmu_tx_commit(tx); ztest_range_unlock(zd, rl); ztest_object_unlock(zd, lr->lr_foid); return (0); } static int ztest_replay_setattr(void *arg1, void *arg2, boolean_t byteswap) { ztest_ds_t *zd = arg1; lr_setattr_t *lr = arg2; objset_t *os = zd->zd_os; dmu_tx_t *tx; dmu_buf_t *db; ztest_block_tag_t *bbt; uint64_t txg, lrtxg, crtxg, dnodesize; if (byteswap) byteswap_uint64_array(lr, sizeof (*lr)); ztest_object_lock(zd, lr->lr_foid, RL_WRITER); VERIFY3U(0, ==, dmu_bonus_hold(os, lr->lr_foid, FTAG, &db)); tx = dmu_tx_create(os); dmu_tx_hold_bonus(tx, lr->lr_foid); txg = ztest_tx_assign(tx, TXG_WAIT, FTAG); if (txg == 0) { dmu_buf_rele(db, FTAG); ztest_object_unlock(zd, lr->lr_foid); return (ENOSPC); } bbt = ztest_bt_bonus(db); ASSERT3U(bbt->bt_magic, ==, BT_MAGIC); crtxg = bbt->bt_crtxg; lrtxg = lr->lr_common.lrc_txg; dnodesize = bbt->bt_dnodesize; if (zd->zd_zilog->zl_replay) { ASSERT(lr->lr_size != 0); ASSERT(lr->lr_mode != 0); ASSERT(lrtxg != 0); } else { /* * Randomly change the size and increment the generation. */ lr->lr_size = (ztest_random(db->db_size / sizeof (*bbt)) + 1) * sizeof (*bbt); lr->lr_mode = bbt->bt_gen + 1; ASSERT(lrtxg == 0); } /* * Verify that the current bonus buffer is not newer than our txg. */ ztest_bt_verify(bbt, os, lr->lr_foid, dnodesize, -1ULL, lr->lr_mode, MAX(txg, lrtxg), crtxg); dmu_buf_will_dirty(db, tx); ASSERT3U(lr->lr_size, >=, sizeof (*bbt)); ASSERT3U(lr->lr_size, <=, db->db_size); VERIFY0(dmu_set_bonus(db, lr->lr_size, tx)); bbt = ztest_bt_bonus(db); ztest_bt_generate(bbt, os, lr->lr_foid, dnodesize, -1ULL, lr->lr_mode, txg, crtxg); ztest_fill_unused_bonus(db, bbt, lr->lr_foid, os, bbt->bt_gen); dmu_buf_rele(db, FTAG); (void) ztest_log_setattr(zd, tx, lr); dmu_tx_commit(tx); ztest_object_unlock(zd, lr->lr_foid); return (0); } zil_replay_func_t *ztest_replay_vector[TX_MAX_TYPE] = { NULL, /* 0 no such transaction type */ ztest_replay_create, /* TX_CREATE */ NULL, /* TX_MKDIR */ NULL, /* TX_MKXATTR */ NULL, /* TX_SYMLINK */ ztest_replay_remove, /* TX_REMOVE */ NULL, /* TX_RMDIR */ NULL, /* TX_LINK */ NULL, /* TX_RENAME */ ztest_replay_write, /* TX_WRITE */ ztest_replay_truncate, /* TX_TRUNCATE */ ztest_replay_setattr, /* TX_SETATTR */ NULL, /* TX_ACL */ NULL, /* TX_CREATE_ACL */ NULL, /* TX_CREATE_ATTR */ NULL, /* TX_CREATE_ACL_ATTR */ NULL, /* TX_MKDIR_ACL */ NULL, /* TX_MKDIR_ATTR */ NULL, /* TX_MKDIR_ACL_ATTR */ NULL, /* TX_WRITE2 */ }; /* * ZIL get_data callbacks */ typedef struct ztest_zgd_private { ztest_ds_t *z_zd; ztest_zrl_t *z_rl; uint64_t z_object; } ztest_zgd_private_t; static void ztest_get_done(zgd_t *zgd, int error) { ztest_zgd_private_t *zzp = zgd->zgd_private; ztest_ds_t *zd = zzp->z_zd; uint64_t object = zzp->z_object; if (zgd->zgd_db) dmu_buf_rele(zgd->zgd_db, zgd); ztest_range_unlock(zd, zzp->z_rl); ztest_object_unlock(zd, object); if (error == 0 && zgd->zgd_bp) - zil_add_block(zgd->zgd_zilog, zgd->zgd_bp); + zil_lwb_add_block(zgd->zgd_lwb, zgd->zgd_bp); umem_free(zgd, sizeof (*zgd)); umem_free(zzp, sizeof (*zzp)); } static int -ztest_get_data(void *arg, lr_write_t *lr, char *buf, zio_t *zio) +ztest_get_data(void *arg, lr_write_t *lr, char *buf, struct lwb *lwb, + zio_t *zio) { ztest_ds_t *zd = arg; objset_t *os = zd->zd_os; uint64_t object = lr->lr_foid; uint64_t offset = lr->lr_offset; uint64_t size = lr->lr_length; uint64_t txg = lr->lr_common.lrc_txg; uint64_t crtxg; dmu_object_info_t doi; dmu_buf_t *db; zgd_t *zgd; int error; ztest_zgd_private_t *zgd_private; + ASSERT3P(lwb, !=, NULL); + ASSERT3P(zio, !=, NULL); + ASSERT3U(size, !=, 0); + ztest_object_lock(zd, object, RL_READER); error = dmu_bonus_hold(os, object, FTAG, &db); if (error) { ztest_object_unlock(zd, object); return (error); } crtxg = ztest_bt_bonus(db)->bt_crtxg; if (crtxg == 0 || crtxg > txg) { dmu_buf_rele(db, FTAG); ztest_object_unlock(zd, object); return (ENOENT); } dmu_object_info_from_db(db, &doi); dmu_buf_rele(db, FTAG); db = NULL; zgd = umem_zalloc(sizeof (*zgd), UMEM_NOFAIL); - zgd->zgd_zilog = zd->zd_zilog; + zgd->zgd_lwb = lwb; zgd_private = umem_zalloc(sizeof (ztest_zgd_private_t), UMEM_NOFAIL); zgd_private->z_zd = zd; zgd_private->z_object = object; zgd->zgd_private = zgd_private; if (buf != NULL) { /* immediate write */ zgd_private->z_rl = ztest_range_lock(zd, object, offset, size, RL_READER); zgd->zgd_rl = zgd_private->z_rl->z_rl; error = dmu_read(os, object, offset, size, buf, DMU_READ_NO_PREFETCH); ASSERT(error == 0); } else { size = doi.doi_data_block_size; if (ISP2(size)) { offset = P2ALIGN(offset, size); } else { ASSERT(offset < size); offset = 0; } zgd_private->z_rl = ztest_range_lock(zd, object, offset, size, RL_READER); zgd->zgd_rl = zgd_private->z_rl->z_rl; error = dmu_buf_hold(os, object, offset, zgd, &db, DMU_READ_NO_PREFETCH); if (error == 0) { blkptr_t *bp = &lr->lr_blkptr; zgd->zgd_db = db; zgd->zgd_bp = bp; ASSERT(db->db_offset == offset); ASSERT(db->db_size == size); error = dmu_sync(zio, lr->lr_common.lrc_txg, ztest_get_done, zgd); if (error == 0) return (0); } } ztest_get_done(zgd, error); return (error); } static void * ztest_lr_alloc(size_t lrsize, char *name) { char *lr; size_t namesize = name ? strlen(name) + 1 : 0; lr = umem_zalloc(lrsize + namesize, UMEM_NOFAIL); if (name) bcopy(name, lr + lrsize, namesize); return (lr); } void ztest_lr_free(void *lr, size_t lrsize, char *name) { size_t namesize = name ? strlen(name) + 1 : 0; umem_free(lr, lrsize + namesize); } /* * Lookup a bunch of objects. Returns the number of objects not found. */ static int ztest_lookup(ztest_ds_t *zd, ztest_od_t *od, int count) { int missing = 0; int error; int i; ASSERT(MUTEX_HELD(&zd->zd_dirobj_lock)); for (i = 0; i < count; i++, od++) { od->od_object = 0; error = zap_lookup(zd->zd_os, od->od_dir, od->od_name, sizeof (uint64_t), 1, &od->od_object); if (error) { ASSERT(error == ENOENT); ASSERT(od->od_object == 0); missing++; } else { dmu_buf_t *db; ztest_block_tag_t *bbt; dmu_object_info_t doi; ASSERT(od->od_object != 0); ASSERT(missing == 0); /* there should be no gaps */ ztest_object_lock(zd, od->od_object, RL_READER); VERIFY3U(0, ==, dmu_bonus_hold(zd->zd_os, od->od_object, FTAG, &db)); dmu_object_info_from_db(db, &doi); bbt = ztest_bt_bonus(db); ASSERT3U(bbt->bt_magic, ==, BT_MAGIC); od->od_type = doi.doi_type; od->od_blocksize = doi.doi_data_block_size; od->od_gen = bbt->bt_gen; dmu_buf_rele(db, FTAG); ztest_object_unlock(zd, od->od_object); } } return (missing); } static int ztest_create(ztest_ds_t *zd, ztest_od_t *od, int count) { int missing = 0; int i; ASSERT(MUTEX_HELD(&zd->zd_dirobj_lock)); for (i = 0; i < count; i++, od++) { if (missing) { od->od_object = 0; missing++; continue; } lr_create_t *lr = ztest_lr_alloc(sizeof (*lr), od->od_name); lr->lr_doid = od->od_dir; lr->lr_foid = 0; /* 0 to allocate, > 0 to claim */ lr->lrz_type = od->od_crtype; lr->lrz_blocksize = od->od_crblocksize; lr->lrz_ibshift = ztest_random_ibshift(); lr->lrz_bonustype = DMU_OT_UINT64_OTHER; lr->lrz_dnodesize = od->od_crdnodesize; lr->lr_gen = od->od_crgen; lr->lr_crtime[0] = time(NULL); if (ztest_replay_create(zd, lr, B_FALSE) != 0) { ASSERT(missing == 0); od->od_object = 0; missing++; } else { od->od_object = lr->lr_foid; od->od_type = od->od_crtype; od->od_blocksize = od->od_crblocksize; od->od_gen = od->od_crgen; ASSERT(od->od_object != 0); } ztest_lr_free(lr, sizeof (*lr), od->od_name); } return (missing); } static int ztest_remove(ztest_ds_t *zd, ztest_od_t *od, int count) { int missing = 0; int error; int i; ASSERT(MUTEX_HELD(&zd->zd_dirobj_lock)); od += count - 1; for (i = count - 1; i >= 0; i--, od--) { if (missing) { missing++; continue; } /* * No object was found. */ if (od->od_object == 0) continue; lr_remove_t *lr = ztest_lr_alloc(sizeof (*lr), od->od_name); lr->lr_doid = od->od_dir; if ((error = ztest_replay_remove(zd, lr, B_FALSE)) != 0) { ASSERT3U(error, ==, ENOSPC); missing++; } else { od->od_object = 0; } ztest_lr_free(lr, sizeof (*lr), od->od_name); } return (missing); } static int ztest_write(ztest_ds_t *zd, uint64_t object, uint64_t offset, uint64_t size, void *data) { lr_write_t *lr; int error; lr = ztest_lr_alloc(sizeof (*lr) + size, NULL); lr->lr_foid = object; lr->lr_offset = offset; lr->lr_length = size; lr->lr_blkoff = 0; BP_ZERO(&lr->lr_blkptr); bcopy(data, lr + 1, size); error = ztest_replay_write(zd, lr, B_FALSE); ztest_lr_free(lr, sizeof (*lr) + size, NULL); return (error); } static int ztest_truncate(ztest_ds_t *zd, uint64_t object, uint64_t offset, uint64_t size) { lr_truncate_t *lr; int error; lr = ztest_lr_alloc(sizeof (*lr), NULL); lr->lr_foid = object; lr->lr_offset = offset; lr->lr_length = size; error = ztest_replay_truncate(zd, lr, B_FALSE); ztest_lr_free(lr, sizeof (*lr), NULL); return (error); } static int ztest_setattr(ztest_ds_t *zd, uint64_t object) { lr_setattr_t *lr; int error; lr = ztest_lr_alloc(sizeof (*lr), NULL); lr->lr_foid = object; lr->lr_size = 0; lr->lr_mode = 0; error = ztest_replay_setattr(zd, lr, B_FALSE); ztest_lr_free(lr, sizeof (*lr), NULL); return (error); } static void ztest_prealloc(ztest_ds_t *zd, uint64_t object, uint64_t offset, uint64_t size) { objset_t *os = zd->zd_os; dmu_tx_t *tx; uint64_t txg; ztest_zrl_t *rl; txg_wait_synced(dmu_objset_pool(os), 0); ztest_object_lock(zd, object, RL_READER); rl = ztest_range_lock(zd, object, offset, size, RL_WRITER); tx = dmu_tx_create(os); dmu_tx_hold_write(tx, object, offset, size); txg = ztest_tx_assign(tx, TXG_WAIT, FTAG); if (txg != 0) { dmu_prealloc(os, object, offset, size, tx); dmu_tx_commit(tx); txg_wait_synced(dmu_objset_pool(os), txg); } else { (void) dmu_free_long_range(os, object, offset, size); } ztest_range_unlock(zd, rl); ztest_object_unlock(zd, object); } static void ztest_io(ztest_ds_t *zd, uint64_t object, uint64_t offset) { int err; ztest_block_tag_t wbt; dmu_object_info_t doi; enum ztest_io_type io_type; uint64_t blocksize; void *data; VERIFY(dmu_object_info(zd->zd_os, object, &doi) == 0); blocksize = doi.doi_data_block_size; data = umem_alloc(blocksize, UMEM_NOFAIL); /* * Pick an i/o type at random, biased toward writing block tags. */ io_type = ztest_random(ZTEST_IO_TYPES); if (ztest_random(2) == 0) io_type = ZTEST_IO_WRITE_TAG; (void) rw_rdlock(&zd->zd_zilog_lock); switch (io_type) { case ZTEST_IO_WRITE_TAG: ztest_bt_generate(&wbt, zd->zd_os, object, doi.doi_dnodesize, offset, 0, 0, 0); (void) ztest_write(zd, object, offset, sizeof (wbt), &wbt); break; case ZTEST_IO_WRITE_PATTERN: (void) memset(data, 'a' + (object + offset) % 5, blocksize); if (ztest_random(2) == 0) { /* * Induce fletcher2 collisions to ensure that * zio_ddt_collision() detects and resolves them * when using fletcher2-verify for deduplication. */ ((uint64_t *)data)[0] ^= 1ULL << 63; ((uint64_t *)data)[4] ^= 1ULL << 63; } (void) ztest_write(zd, object, offset, blocksize, data); break; case ZTEST_IO_WRITE_ZEROES: bzero(data, blocksize); (void) ztest_write(zd, object, offset, blocksize, data); break; case ZTEST_IO_TRUNCATE: (void) ztest_truncate(zd, object, offset, blocksize); break; case ZTEST_IO_SETATTR: (void) ztest_setattr(zd, object); break; default: break; case ZTEST_IO_REWRITE: (void) rw_rdlock(&ztest_name_lock); err = ztest_dsl_prop_set_uint64(zd->zd_name, ZFS_PROP_CHECKSUM, spa_dedup_checksum(ztest_spa), B_FALSE); VERIFY(err == 0 || err == ENOSPC); err = ztest_dsl_prop_set_uint64(zd->zd_name, ZFS_PROP_COMPRESSION, ztest_random_dsl_prop(ZFS_PROP_COMPRESSION), B_FALSE); VERIFY(err == 0 || err == ENOSPC); (void) rw_unlock(&ztest_name_lock); VERIFY0(dmu_read(zd->zd_os, object, offset, blocksize, data, DMU_READ_NO_PREFETCH)); (void) ztest_write(zd, object, offset, blocksize, data); break; } (void) rw_unlock(&zd->zd_zilog_lock); umem_free(data, blocksize); } /* * Initialize an object description template. */ static void ztest_od_init(ztest_od_t *od, uint64_t id, char *tag, uint64_t index, dmu_object_type_t type, uint64_t blocksize, uint64_t dnodesize, uint64_t gen) { od->od_dir = ZTEST_DIROBJ; od->od_object = 0; od->od_crtype = type; od->od_crblocksize = blocksize ? blocksize : ztest_random_blocksize(); od->od_crdnodesize = dnodesize ? dnodesize : ztest_random_dnodesize(); od->od_crgen = gen; od->od_type = DMU_OT_NONE; od->od_blocksize = 0; od->od_gen = 0; (void) snprintf(od->od_name, sizeof (od->od_name), "%s(%lld)[%llu]", tag, (longlong_t)id, (u_longlong_t)index); } /* * Lookup or create the objects for a test using the od template. * If the objects do not all exist, or if 'remove' is specified, * remove any existing objects and create new ones. Otherwise, * use the existing objects. */ static int ztest_object_init(ztest_ds_t *zd, ztest_od_t *od, size_t size, boolean_t remove) { int count = size / sizeof (*od); int rv = 0; mutex_enter(&zd->zd_dirobj_lock); if ((ztest_lookup(zd, od, count) != 0 || remove) && (ztest_remove(zd, od, count) != 0 || ztest_create(zd, od, count) != 0)) rv = -1; zd->zd_od = od; mutex_exit(&zd->zd_dirobj_lock); return (rv); } /* ARGSUSED */ void ztest_zil_commit(ztest_ds_t *zd, uint64_t id) { zilog_t *zilog = zd->zd_zilog; (void) rw_rdlock(&zd->zd_zilog_lock); zil_commit(zilog, ztest_random(ZTEST_OBJECTS)); /* * Remember the committed values in zd, which is in parent/child * shared memory. If we die, the next iteration of ztest_run() * will verify that the log really does contain this record. */ mutex_enter(&zilog->zl_lock); ASSERT(zd->zd_shared != NULL); ASSERT3U(zd->zd_shared->zd_seq, <=, zilog->zl_commit_lr_seq); zd->zd_shared->zd_seq = zilog->zl_commit_lr_seq; mutex_exit(&zilog->zl_lock); (void) rw_unlock(&zd->zd_zilog_lock); } /* * This function is designed to simulate the operations that occur during a * mount/unmount operation. We hold the dataset across these operations in an * attempt to expose any implicit assumptions about ZIL management. */ /* ARGSUSED */ void ztest_zil_remount(ztest_ds_t *zd, uint64_t id) { objset_t *os = zd->zd_os; /* * We grab the zd_dirobj_lock to ensure that no other thread is * updating the zil (i.e. adding in-memory log records) and the * zd_zilog_lock to block any I/O. */ mutex_enter(&zd->zd_dirobj_lock); (void) rw_wrlock(&zd->zd_zilog_lock); /* zfsvfs_teardown() */ zil_close(zd->zd_zilog); /* zfsvfs_setup() */ VERIFY(zil_open(os, ztest_get_data) == zd->zd_zilog); zil_replay(os, zd, ztest_replay_vector); (void) rw_unlock(&zd->zd_zilog_lock); mutex_exit(&zd->zd_dirobj_lock); } /* * Verify that we can't destroy an active pool, create an existing pool, * or create a pool with a bad vdev spec. */ /* ARGSUSED */ void ztest_spa_create_destroy(ztest_ds_t *zd, uint64_t id) { ztest_shared_opts_t *zo = &ztest_opts; spa_t *spa; nvlist_t *nvroot; if (zo->zo_mmp_test) return; /* * Attempt to create using a bad file. */ nvroot = make_vdev_root("/dev/bogus", NULL, NULL, 0, 0, 0, 0, 0, 1); VERIFY3U(ENOENT, ==, spa_create("ztest_bad_file", nvroot, NULL, NULL, NULL)); nvlist_free(nvroot); /* * Attempt to create using a bad mirror. */ nvroot = make_vdev_root("/dev/bogus", NULL, NULL, 0, 0, 0, 0, 2, 1); VERIFY3U(ENOENT, ==, spa_create("ztest_bad_mirror", nvroot, NULL, NULL, NULL)); nvlist_free(nvroot); /* * Attempt to create an existing pool. It shouldn't matter * what's in the nvroot; we should fail with EEXIST. */ (void) rw_rdlock(&ztest_name_lock); nvroot = make_vdev_root("/dev/bogus", NULL, NULL, 0, 0, 0, 0, 0, 1); VERIFY3U(EEXIST, ==, spa_create(zo->zo_pool, nvroot, NULL, NULL, NULL)); nvlist_free(nvroot); VERIFY3U(0, ==, spa_open(zo->zo_pool, &spa, FTAG)); VERIFY3U(EBUSY, ==, spa_destroy(zo->zo_pool)); spa_close(spa, FTAG); (void) rw_unlock(&ztest_name_lock); } /* * Start and then stop the MMP threads to ensure the startup and shutdown code * works properly. Actual protection and property-related code tested via ZTS. */ /* ARGSUSED */ void ztest_mmp_enable_disable(ztest_ds_t *zd, uint64_t id) { ztest_shared_opts_t *zo = &ztest_opts; spa_t *spa = ztest_spa; if (zo->zo_mmp_test) return; spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); mutex_enter(&spa->spa_props_lock); if (!spa_multihost(spa)) { spa->spa_multihost = B_TRUE; mmp_thread_start(spa); } mutex_exit(&spa->spa_props_lock); spa_config_exit(spa, SCL_CONFIG, FTAG); txg_wait_synced(spa_get_dsl(spa), 0); mmp_signal_all_threads(); txg_wait_synced(spa_get_dsl(spa), 0); spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); mutex_enter(&spa->spa_props_lock); if (spa_multihost(spa)) { mmp_thread_stop(spa); spa->spa_multihost = B_FALSE; } mutex_exit(&spa->spa_props_lock); spa_config_exit(spa, SCL_CONFIG, FTAG); } /* ARGSUSED */ void ztest_spa_upgrade(ztest_ds_t *zd, uint64_t id) { spa_t *spa; uint64_t initial_version = SPA_VERSION_INITIAL; uint64_t version, newversion; nvlist_t *nvroot, *props; char *name; if (ztest_opts.zo_mmp_test) return; mutex_enter(&ztest_vdev_lock); name = kmem_asprintf("%s_upgrade", ztest_opts.zo_pool); /* * Clean up from previous runs. */ (void) spa_destroy(name); nvroot = make_vdev_root(NULL, NULL, name, ztest_opts.zo_vdev_size, 0, 0, ztest_opts.zo_raidz, ztest_opts.zo_mirrors, 1); /* * If we're configuring a RAIDZ device then make sure that the * the initial version is capable of supporting that feature. */ switch (ztest_opts.zo_raidz_parity) { case 0: case 1: initial_version = SPA_VERSION_INITIAL; break; case 2: initial_version = SPA_VERSION_RAIDZ2; break; case 3: initial_version = SPA_VERSION_RAIDZ3; break; } /* * Create a pool with a spa version that can be upgraded. Pick * a value between initial_version and SPA_VERSION_BEFORE_FEATURES. */ do { version = ztest_random_spa_version(initial_version); } while (version > SPA_VERSION_BEFORE_FEATURES); props = fnvlist_alloc(); fnvlist_add_uint64(props, zpool_prop_to_name(ZPOOL_PROP_VERSION), version); VERIFY3S(spa_create(name, nvroot, props, NULL, NULL), ==, 0); fnvlist_free(nvroot); fnvlist_free(props); VERIFY3S(spa_open(name, &spa, FTAG), ==, 0); VERIFY3U(spa_version(spa), ==, version); newversion = ztest_random_spa_version(version + 1); if (ztest_opts.zo_verbose >= 4) { (void) printf("upgrading spa version from %llu to %llu\n", (u_longlong_t)version, (u_longlong_t)newversion); } spa_upgrade(spa, newversion); VERIFY3U(spa_version(spa), >, version); VERIFY3U(spa_version(spa), ==, fnvlist_lookup_uint64(spa->spa_config, zpool_prop_to_name(ZPOOL_PROP_VERSION))); spa_close(spa, FTAG); strfree(name); mutex_exit(&ztest_vdev_lock); } static vdev_t * vdev_lookup_by_path(vdev_t *vd, const char *path) { vdev_t *mvd; int c; if (vd->vdev_path != NULL && strcmp(path, vd->vdev_path) == 0) return (vd); for (c = 0; c < vd->vdev_children; c++) if ((mvd = vdev_lookup_by_path(vd->vdev_child[c], path)) != NULL) return (mvd); return (NULL); } /* * Find the first available hole which can be used as a top-level. */ int find_vdev_hole(spa_t *spa) { vdev_t *rvd = spa->spa_root_vdev; int c; ASSERT(spa_config_held(spa, SCL_VDEV, RW_READER) == SCL_VDEV); for (c = 0; c < rvd->vdev_children; c++) { vdev_t *cvd = rvd->vdev_child[c]; if (cvd->vdev_ishole) break; } return (c); } /* * Verify that vdev_add() works as expected. */ /* ARGSUSED */ void ztest_vdev_add_remove(ztest_ds_t *zd, uint64_t id) { ztest_shared_t *zs = ztest_shared; spa_t *spa = ztest_spa; uint64_t leaves; uint64_t guid; nvlist_t *nvroot; int error; if (ztest_opts.zo_mmp_test) return; mutex_enter(&ztest_vdev_lock); leaves = MAX(zs->zs_mirrors + zs->zs_splits, 1) * ztest_opts.zo_raidz; spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); ztest_shared->zs_vdev_next_leaf = find_vdev_hole(spa) * leaves; /* * If we have slogs then remove them 1/4 of the time. */ if (spa_has_slogs(spa) && ztest_random(4) == 0) { /* * Grab the guid from the head of the log class rotor. */ guid = spa_log_class(spa)->mc_rotor->mg_vd->vdev_guid; spa_config_exit(spa, SCL_VDEV, FTAG); /* * We have to grab the zs_name_lock as writer to * prevent a race between removing a slog (dmu_objset_find) * and destroying a dataset. Removing the slog will * grab a reference on the dataset which may cause * dsl_destroy_head() to fail with EBUSY thus * leaving the dataset in an inconsistent state. */ rw_wrlock(&ztest_name_lock); error = spa_vdev_remove(spa, guid, B_FALSE); rw_unlock(&ztest_name_lock); if (error && error != EEXIST) fatal(0, "spa_vdev_remove() = %d", error); } else { spa_config_exit(spa, SCL_VDEV, FTAG); /* * Make 1/4 of the devices be log devices. */ nvroot = make_vdev_root(NULL, NULL, NULL, ztest_opts.zo_vdev_size, 0, ztest_random(4) == 0, ztest_opts.zo_raidz, zs->zs_mirrors, 1); error = spa_vdev_add(spa, nvroot); nvlist_free(nvroot); if (error == ENOSPC) ztest_record_enospc("spa_vdev_add"); else if (error != 0) fatal(0, "spa_vdev_add() = %d", error); } mutex_exit(&ztest_vdev_lock); } /* * Verify that adding/removing aux devices (l2arc, hot spare) works as expected. */ /* ARGSUSED */ void ztest_vdev_aux_add_remove(ztest_ds_t *zd, uint64_t id) { ztest_shared_t *zs = ztest_shared; spa_t *spa = ztest_spa; vdev_t *rvd = spa->spa_root_vdev; spa_aux_vdev_t *sav; char *aux; char *path; uint64_t guid = 0; int error; if (ztest_opts.zo_mmp_test) return; path = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); if (ztest_random(2) == 0) { sav = &spa->spa_spares; aux = ZPOOL_CONFIG_SPARES; } else { sav = &spa->spa_l2cache; aux = ZPOOL_CONFIG_L2CACHE; } mutex_enter(&ztest_vdev_lock); spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); if (sav->sav_count != 0 && ztest_random(4) == 0) { /* * Pick a random device to remove. */ guid = sav->sav_vdevs[ztest_random(sav->sav_count)]->vdev_guid; } else { /* * Find an unused device we can add. */ zs->zs_vdev_aux = 0; for (;;) { int c; (void) snprintf(path, MAXPATHLEN, ztest_aux_template, ztest_opts.zo_dir, ztest_opts.zo_pool, aux, zs->zs_vdev_aux); for (c = 0; c < sav->sav_count; c++) if (strcmp(sav->sav_vdevs[c]->vdev_path, path) == 0) break; if (c == sav->sav_count && vdev_lookup_by_path(rvd, path) == NULL) break; zs->zs_vdev_aux++; } } spa_config_exit(spa, SCL_VDEV, FTAG); if (guid == 0) { /* * Add a new device. */ nvlist_t *nvroot = make_vdev_root(NULL, aux, NULL, (ztest_opts.zo_vdev_size * 5) / 4, 0, 0, 0, 0, 1); error = spa_vdev_add(spa, nvroot); if (error != 0) fatal(0, "spa_vdev_add(%p) = %d", nvroot, error); nvlist_free(nvroot); } else { /* * Remove an existing device. Sometimes, dirty its * vdev state first to make sure we handle removal * of devices that have pending state changes. */ if (ztest_random(2) == 0) (void) vdev_online(spa, guid, 0, NULL); error = spa_vdev_remove(spa, guid, B_FALSE); if (error != 0 && error != EBUSY) fatal(0, "spa_vdev_remove(%llu) = %d", guid, error); } mutex_exit(&ztest_vdev_lock); umem_free(path, MAXPATHLEN); } /* * split a pool if it has mirror tlvdevs */ /* ARGSUSED */ void ztest_split_pool(ztest_ds_t *zd, uint64_t id) { ztest_shared_t *zs = ztest_shared; spa_t *spa = ztest_spa; vdev_t *rvd = spa->spa_root_vdev; nvlist_t *tree, **child, *config, *split, **schild; uint_t c, children, schildren = 0, lastlogid = 0; int error = 0; if (ztest_opts.zo_mmp_test) return; mutex_enter(&ztest_vdev_lock); /* ensure we have a useable config; mirrors of raidz aren't supported */ if (zs->zs_mirrors < 3 || ztest_opts.zo_raidz > 1) { mutex_exit(&ztest_vdev_lock); return; } /* clean up the old pool, if any */ (void) spa_destroy("splitp"); spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); /* generate a config from the existing config */ mutex_enter(&spa->spa_props_lock); VERIFY(nvlist_lookup_nvlist(spa->spa_config, ZPOOL_CONFIG_VDEV_TREE, &tree) == 0); mutex_exit(&spa->spa_props_lock); VERIFY(nvlist_lookup_nvlist_array(tree, ZPOOL_CONFIG_CHILDREN, &child, &children) == 0); schild = malloc(rvd->vdev_children * sizeof (nvlist_t *)); for (c = 0; c < children; c++) { vdev_t *tvd = rvd->vdev_child[c]; nvlist_t **mchild; uint_t mchildren; if (tvd->vdev_islog || tvd->vdev_ops == &vdev_hole_ops) { VERIFY(nvlist_alloc(&schild[schildren], NV_UNIQUE_NAME, 0) == 0); VERIFY(nvlist_add_string(schild[schildren], ZPOOL_CONFIG_TYPE, VDEV_TYPE_HOLE) == 0); VERIFY(nvlist_add_uint64(schild[schildren], ZPOOL_CONFIG_IS_HOLE, 1) == 0); if (lastlogid == 0) lastlogid = schildren; ++schildren; continue; } lastlogid = 0; VERIFY(nvlist_lookup_nvlist_array(child[c], ZPOOL_CONFIG_CHILDREN, &mchild, &mchildren) == 0); VERIFY(nvlist_dup(mchild[0], &schild[schildren++], 0) == 0); } /* OK, create a config that can be used to split */ VERIFY(nvlist_alloc(&split, NV_UNIQUE_NAME, 0) == 0); VERIFY(nvlist_add_string(split, ZPOOL_CONFIG_TYPE, VDEV_TYPE_ROOT) == 0); VERIFY(nvlist_add_nvlist_array(split, ZPOOL_CONFIG_CHILDREN, schild, lastlogid != 0 ? lastlogid : schildren) == 0); VERIFY(nvlist_alloc(&config, NV_UNIQUE_NAME, 0) == 0); VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, split) == 0); for (c = 0; c < schildren; c++) nvlist_free(schild[c]); free(schild); nvlist_free(split); spa_config_exit(spa, SCL_VDEV, FTAG); (void) rw_wrlock(&ztest_name_lock); error = spa_vdev_split_mirror(spa, "splitp", config, NULL, B_FALSE); (void) rw_unlock(&ztest_name_lock); nvlist_free(config); if (error == 0) { (void) printf("successful split - results:\n"); mutex_enter(&spa_namespace_lock); show_pool_stats(spa); show_pool_stats(spa_lookup("splitp")); mutex_exit(&spa_namespace_lock); ++zs->zs_splits; --zs->zs_mirrors; } mutex_exit(&ztest_vdev_lock); } /* * Verify that we can attach and detach devices. */ /* ARGSUSED */ void ztest_vdev_attach_detach(ztest_ds_t *zd, uint64_t id) { ztest_shared_t *zs = ztest_shared; spa_t *spa = ztest_spa; spa_aux_vdev_t *sav = &spa->spa_spares; vdev_t *rvd = spa->spa_root_vdev; vdev_t *oldvd, *newvd, *pvd; nvlist_t *root; uint64_t leaves; uint64_t leaf, top; uint64_t ashift = ztest_get_ashift(); uint64_t oldguid, pguid; uint64_t oldsize, newsize; char *oldpath, *newpath; int replacing; int oldvd_has_siblings = B_FALSE; int newvd_is_spare = B_FALSE; int oldvd_is_log; int error, expected_error; if (ztest_opts.zo_mmp_test) return; oldpath = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); newpath = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); mutex_enter(&ztest_vdev_lock); leaves = MAX(zs->zs_mirrors, 1) * ztest_opts.zo_raidz; spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); /* * Decide whether to do an attach or a replace. */ replacing = ztest_random(2); /* * Pick a random top-level vdev. */ top = ztest_random_vdev_top(spa, B_TRUE); /* * Pick a random leaf within it. */ leaf = ztest_random(leaves); /* * Locate this vdev. */ oldvd = rvd->vdev_child[top]; if (zs->zs_mirrors >= 1) { ASSERT(oldvd->vdev_ops == &vdev_mirror_ops); ASSERT(oldvd->vdev_children >= zs->zs_mirrors); oldvd = oldvd->vdev_child[leaf / ztest_opts.zo_raidz]; } if (ztest_opts.zo_raidz > 1) { ASSERT(oldvd->vdev_ops == &vdev_raidz_ops); ASSERT(oldvd->vdev_children == ztest_opts.zo_raidz); oldvd = oldvd->vdev_child[leaf % ztest_opts.zo_raidz]; } /* * If we're already doing an attach or replace, oldvd may be a * mirror vdev -- in which case, pick a random child. */ while (oldvd->vdev_children != 0) { oldvd_has_siblings = B_TRUE; ASSERT(oldvd->vdev_children >= 2); oldvd = oldvd->vdev_child[ztest_random(oldvd->vdev_children)]; } oldguid = oldvd->vdev_guid; oldsize = vdev_get_min_asize(oldvd); oldvd_is_log = oldvd->vdev_top->vdev_islog; (void) strcpy(oldpath, oldvd->vdev_path); pvd = oldvd->vdev_parent; pguid = pvd->vdev_guid; /* * If oldvd has siblings, then half of the time, detach it. */ if (oldvd_has_siblings && ztest_random(2) == 0) { spa_config_exit(spa, SCL_VDEV, FTAG); error = spa_vdev_detach(spa, oldguid, pguid, B_FALSE); if (error != 0 && error != ENODEV && error != EBUSY && error != ENOTSUP) fatal(0, "detach (%s) returned %d", oldpath, error); goto out; } /* * For the new vdev, choose with equal probability between the two * standard paths (ending in either 'a' or 'b') or a random hot spare. */ if (sav->sav_count != 0 && ztest_random(3) == 0) { newvd = sav->sav_vdevs[ztest_random(sav->sav_count)]; newvd_is_spare = B_TRUE; (void) strcpy(newpath, newvd->vdev_path); } else { (void) snprintf(newpath, MAXPATHLEN, ztest_dev_template, ztest_opts.zo_dir, ztest_opts.zo_pool, top * leaves + leaf); if (ztest_random(2) == 0) newpath[strlen(newpath) - 1] = 'b'; newvd = vdev_lookup_by_path(rvd, newpath); } if (newvd) { newsize = vdev_get_min_asize(newvd); } else { /* * Make newsize a little bigger or smaller than oldsize. * If it's smaller, the attach should fail. * If it's larger, and we're doing a replace, * we should get dynamic LUN growth when we're done. */ newsize = 10 * oldsize / (9 + ztest_random(3)); } /* * If pvd is not a mirror or root, the attach should fail with ENOTSUP, * unless it's a replace; in that case any non-replacing parent is OK. * * If newvd is already part of the pool, it should fail with EBUSY. * * If newvd is too small, it should fail with EOVERFLOW. */ if (pvd->vdev_ops != &vdev_mirror_ops && pvd->vdev_ops != &vdev_root_ops && (!replacing || pvd->vdev_ops == &vdev_replacing_ops || pvd->vdev_ops == &vdev_spare_ops)) expected_error = ENOTSUP; else if (newvd_is_spare && (!replacing || oldvd_is_log)) expected_error = ENOTSUP; else if (newvd == oldvd) expected_error = replacing ? 0 : EBUSY; else if (vdev_lookup_by_path(rvd, newpath) != NULL) expected_error = EBUSY; else if (newsize < oldsize) expected_error = EOVERFLOW; else if (ashift > oldvd->vdev_top->vdev_ashift) expected_error = EDOM; else expected_error = 0; spa_config_exit(spa, SCL_VDEV, FTAG); /* * Build the nvlist describing newpath. */ root = make_vdev_root(newpath, NULL, NULL, newvd == NULL ? newsize : 0, ashift, 0, 0, 0, 1); error = spa_vdev_attach(spa, oldguid, root, replacing); nvlist_free(root); /* * If our parent was the replacing vdev, but the replace completed, * then instead of failing with ENOTSUP we may either succeed, * fail with ENODEV, or fail with EOVERFLOW. */ if (expected_error == ENOTSUP && (error == 0 || error == ENODEV || error == EOVERFLOW)) expected_error = error; /* * If someone grew the LUN, the replacement may be too small. */ if (error == EOVERFLOW || error == EBUSY) expected_error = error; /* XXX workaround 6690467 */ if (error != expected_error && expected_error != EBUSY) { fatal(0, "attach (%s %llu, %s %llu, %d) " "returned %d, expected %d", oldpath, oldsize, newpath, newsize, replacing, error, expected_error); } out: mutex_exit(&ztest_vdev_lock); umem_free(oldpath, MAXPATHLEN); umem_free(newpath, MAXPATHLEN); } /* * Callback function which expands the physical size of the vdev. */ vdev_t * grow_vdev(vdev_t *vd, void *arg) { ASSERTV(spa_t *spa = vd->vdev_spa); size_t *newsize = arg; size_t fsize; int fd; ASSERT(spa_config_held(spa, SCL_STATE, RW_READER) == SCL_STATE); ASSERT(vd->vdev_ops->vdev_op_leaf); if ((fd = open(vd->vdev_path, O_RDWR)) == -1) return (vd); fsize = lseek(fd, 0, SEEK_END); VERIFY(ftruncate(fd, *newsize) == 0); if (ztest_opts.zo_verbose >= 6) { (void) printf("%s grew from %lu to %lu bytes\n", vd->vdev_path, (ulong_t)fsize, (ulong_t)*newsize); } (void) close(fd); return (NULL); } /* * Callback function which expands a given vdev by calling vdev_online(). */ /* ARGSUSED */ vdev_t * online_vdev(vdev_t *vd, void *arg) { spa_t *spa = vd->vdev_spa; vdev_t *tvd = vd->vdev_top; uint64_t guid = vd->vdev_guid; uint64_t generation = spa->spa_config_generation + 1; vdev_state_t newstate = VDEV_STATE_UNKNOWN; int error; ASSERT(spa_config_held(spa, SCL_STATE, RW_READER) == SCL_STATE); ASSERT(vd->vdev_ops->vdev_op_leaf); /* Calling vdev_online will initialize the new metaslabs */ spa_config_exit(spa, SCL_STATE, spa); error = vdev_online(spa, guid, ZFS_ONLINE_EXPAND, &newstate); spa_config_enter(spa, SCL_STATE, spa, RW_READER); /* * If vdev_online returned an error or the underlying vdev_open * failed then we abort the expand. The only way to know that * vdev_open fails is by checking the returned newstate. */ if (error || newstate != VDEV_STATE_HEALTHY) { if (ztest_opts.zo_verbose >= 5) { (void) printf("Unable to expand vdev, state %llu, " "error %d\n", (u_longlong_t)newstate, error); } return (vd); } ASSERT3U(newstate, ==, VDEV_STATE_HEALTHY); /* * Since we dropped the lock we need to ensure that we're * still talking to the original vdev. It's possible this * vdev may have been detached/replaced while we were * trying to online it. */ if (generation != spa->spa_config_generation) { if (ztest_opts.zo_verbose >= 5) { (void) printf("vdev configuration has changed, " "guid %llu, state %llu, expected gen %llu, " "got gen %llu\n", (u_longlong_t)guid, (u_longlong_t)tvd->vdev_state, (u_longlong_t)generation, (u_longlong_t)spa->spa_config_generation); } return (vd); } return (NULL); } /* * Traverse the vdev tree calling the supplied function. * We continue to walk the tree until we either have walked all * children or we receive a non-NULL return from the callback. * If a NULL callback is passed, then we just return back the first * leaf vdev we encounter. */ vdev_t * vdev_walk_tree(vdev_t *vd, vdev_t *(*func)(vdev_t *, void *), void *arg) { uint_t c; if (vd->vdev_ops->vdev_op_leaf) { if (func == NULL) return (vd); else return (func(vd, arg)); } for (c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; if ((cvd = vdev_walk_tree(cvd, func, arg)) != NULL) return (cvd); } return (NULL); } /* * Verify that dynamic LUN growth works as expected. */ /* ARGSUSED */ void ztest_vdev_LUN_growth(ztest_ds_t *zd, uint64_t id) { spa_t *spa = ztest_spa; vdev_t *vd, *tvd; metaslab_class_t *mc; metaslab_group_t *mg; size_t psize, newsize; uint64_t top; uint64_t old_class_space, new_class_space, old_ms_count, new_ms_count; mutex_enter(&ztest_vdev_lock); spa_config_enter(spa, SCL_STATE, spa, RW_READER); top = ztest_random_vdev_top(spa, B_TRUE); tvd = spa->spa_root_vdev->vdev_child[top]; mg = tvd->vdev_mg; mc = mg->mg_class; old_ms_count = tvd->vdev_ms_count; old_class_space = metaslab_class_get_space(mc); /* * Determine the size of the first leaf vdev associated with * our top-level device. */ vd = vdev_walk_tree(tvd, NULL, NULL); ASSERT3P(vd, !=, NULL); ASSERT(vd->vdev_ops->vdev_op_leaf); psize = vd->vdev_psize; /* * We only try to expand the vdev if it's healthy, less than 4x its * original size, and it has a valid psize. */ if (tvd->vdev_state != VDEV_STATE_HEALTHY || psize == 0 || psize >= 4 * ztest_opts.zo_vdev_size) { spa_config_exit(spa, SCL_STATE, spa); mutex_exit(&ztest_vdev_lock); return; } ASSERT(psize > 0); newsize = psize + psize / 8; ASSERT3U(newsize, >, psize); if (ztest_opts.zo_verbose >= 6) { (void) printf("Expanding LUN %s from %lu to %lu\n", vd->vdev_path, (ulong_t)psize, (ulong_t)newsize); } /* * Growing the vdev is a two step process: * 1). expand the physical size (i.e. relabel) * 2). online the vdev to create the new metaslabs */ if (vdev_walk_tree(tvd, grow_vdev, &newsize) != NULL || vdev_walk_tree(tvd, online_vdev, NULL) != NULL || tvd->vdev_state != VDEV_STATE_HEALTHY) { if (ztest_opts.zo_verbose >= 5) { (void) printf("Could not expand LUN because " "the vdev configuration changed.\n"); } spa_config_exit(spa, SCL_STATE, spa); mutex_exit(&ztest_vdev_lock); return; } spa_config_exit(spa, SCL_STATE, spa); /* * Expanding the LUN will update the config asynchronously, * thus we must wait for the async thread to complete any * pending tasks before proceeding. */ for (;;) { boolean_t done; mutex_enter(&spa->spa_async_lock); done = (spa->spa_async_thread == NULL && !spa->spa_async_tasks); mutex_exit(&spa->spa_async_lock); if (done) break; txg_wait_synced(spa_get_dsl(spa), 0); (void) poll(NULL, 0, 100); } spa_config_enter(spa, SCL_STATE, spa, RW_READER); tvd = spa->spa_root_vdev->vdev_child[top]; new_ms_count = tvd->vdev_ms_count; new_class_space = metaslab_class_get_space(mc); if (tvd->vdev_mg != mg || mg->mg_class != mc) { if (ztest_opts.zo_verbose >= 5) { (void) printf("Could not verify LUN expansion due to " "intervening vdev offline or remove.\n"); } spa_config_exit(spa, SCL_STATE, spa); mutex_exit(&ztest_vdev_lock); return; } /* * Make sure we were able to grow the vdev. */ if (new_ms_count <= old_ms_count) fatal(0, "LUN expansion failed: ms_count %llu <= %llu\n", old_ms_count, new_ms_count); /* * Make sure we were able to grow the pool. */ if (new_class_space <= old_class_space) fatal(0, "LUN expansion failed: class_space %llu <= %llu\n", old_class_space, new_class_space); if (ztest_opts.zo_verbose >= 5) { char oldnumbuf[NN_NUMBUF_SZ], newnumbuf[NN_NUMBUF_SZ]; nicenum(old_class_space, oldnumbuf, sizeof (oldnumbuf)); nicenum(new_class_space, newnumbuf, sizeof (newnumbuf)); (void) printf("%s grew from %s to %s\n", spa->spa_name, oldnumbuf, newnumbuf); } spa_config_exit(spa, SCL_STATE, spa); mutex_exit(&ztest_vdev_lock); } /* * Verify that dmu_objset_{create,destroy,open,close} work as expected. */ /* ARGSUSED */ static void ztest_objset_create_cb(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx) { /* * Create the objects common to all ztest datasets. */ VERIFY(zap_create_claim(os, ZTEST_DIROBJ, DMU_OT_ZAP_OTHER, DMU_OT_NONE, 0, tx) == 0); } static int ztest_dataset_create(char *dsname) { int err; uint64_t rand; dsl_crypto_params_t *dcp = NULL; /* * 50% of the time, we create encrypted datasets * using a random cipher suite and a hard-coded * wrapping key. */ rand = ztest_random(2); if (rand != 0) { nvlist_t *crypto_args = fnvlist_alloc(); nvlist_t *props = fnvlist_alloc(); /* slight bias towards the default cipher suite */ rand = ztest_random(ZIO_CRYPT_FUNCTIONS); if (rand < ZIO_CRYPT_AES_128_CCM) rand = ZIO_CRYPT_ON; fnvlist_add_uint64(props, zfs_prop_to_name(ZFS_PROP_ENCRYPTION), rand); fnvlist_add_uint8_array(crypto_args, "wkeydata", (uint8_t *)ztest_wkeydata, WRAPPING_KEY_LEN); /* * These parameters aren't really used by the kernel. They * are simply stored so that userspace knows how to load * the wrapping key. */ fnvlist_add_uint64(props, zfs_prop_to_name(ZFS_PROP_KEYFORMAT), ZFS_KEYFORMAT_RAW); fnvlist_add_string(props, zfs_prop_to_name(ZFS_PROP_KEYLOCATION), "prompt"); fnvlist_add_uint64(props, zfs_prop_to_name(ZFS_PROP_PBKDF2_SALT), 0ULL); fnvlist_add_uint64(props, zfs_prop_to_name(ZFS_PROP_PBKDF2_ITERS), 0ULL); VERIFY0(dsl_crypto_params_create_nvlist(DCP_CMD_NONE, props, crypto_args, &dcp)); fnvlist_free(crypto_args); fnvlist_free(props); } err = dmu_objset_create(dsname, DMU_OST_OTHER, 0, dcp, ztest_objset_create_cb, NULL); dsl_crypto_params_free(dcp, !!err); rand = ztest_random(100); if (err || rand < 80) return (err); if (ztest_opts.zo_verbose >= 5) (void) printf("Setting dataset %s to sync always\n", dsname); return (ztest_dsl_prop_set_uint64(dsname, ZFS_PROP_SYNC, ZFS_SYNC_ALWAYS, B_FALSE)); } /* ARGSUSED */ static int ztest_objset_destroy_cb(const char *name, void *arg) { objset_t *os; dmu_object_info_t doi; int error; /* * Verify that the dataset contains a directory object. */ VERIFY0(ztest_dmu_objset_own(name, DMU_OST_OTHER, B_TRUE, B_TRUE, FTAG, &os)); error = dmu_object_info(os, ZTEST_DIROBJ, &doi); if (error != ENOENT) { /* We could have crashed in the middle of destroying it */ ASSERT0(error); ASSERT3U(doi.doi_type, ==, DMU_OT_ZAP_OTHER); ASSERT3S(doi.doi_physical_blocks_512, >=, 0); } dmu_objset_disown(os, B_TRUE, FTAG); /* * Destroy the dataset. */ if (strchr(name, '@') != NULL) { VERIFY0(dsl_destroy_snapshot(name, B_TRUE)); } else { error = dsl_destroy_head(name); /* There could be a hold on this dataset */ if (error != EBUSY) ASSERT0(error); } return (0); } static boolean_t ztest_snapshot_create(char *osname, uint64_t id) { char snapname[ZFS_MAX_DATASET_NAME_LEN]; int error; (void) snprintf(snapname, sizeof (snapname), "%llu", (u_longlong_t)id); error = dmu_objset_snapshot_one(osname, snapname); if (error == ENOSPC) { ztest_record_enospc(FTAG); return (B_FALSE); } if (error != 0 && error != EEXIST) { fatal(0, "ztest_snapshot_create(%s@%s) = %d", osname, snapname, error); } return (B_TRUE); } static boolean_t ztest_snapshot_destroy(char *osname, uint64_t id) { char snapname[ZFS_MAX_DATASET_NAME_LEN]; int error; (void) snprintf(snapname, sizeof (snapname), "%s@%llu", osname, (u_longlong_t)id); error = dsl_destroy_snapshot(snapname, B_FALSE); if (error != 0 && error != ENOENT) fatal(0, "ztest_snapshot_destroy(%s) = %d", snapname, error); return (B_TRUE); } /* ARGSUSED */ void ztest_dmu_objset_create_destroy(ztest_ds_t *zd, uint64_t id) { ztest_ds_t *zdtmp; int iters; int error; objset_t *os, *os2; char name[ZFS_MAX_DATASET_NAME_LEN]; zilog_t *zilog; int i; zdtmp = umem_alloc(sizeof (ztest_ds_t), UMEM_NOFAIL); (void) rw_rdlock(&ztest_name_lock); (void) snprintf(name, sizeof (name), "%s/temp_%llu", ztest_opts.zo_pool, (u_longlong_t)id); /* * If this dataset exists from a previous run, process its replay log * half of the time. If we don't replay it, then dsl_destroy_head() * (invoked from ztest_objset_destroy_cb()) should just throw it away. */ if (ztest_random(2) == 0 && ztest_dmu_objset_own(name, DMU_OST_OTHER, B_FALSE, B_TRUE, FTAG, &os) == 0) { ztest_zd_init(zdtmp, NULL, os); zil_replay(os, zdtmp, ztest_replay_vector); ztest_zd_fini(zdtmp); txg_wait_synced(dmu_objset_pool(os), 0); dmu_objset_disown(os, B_TRUE, FTAG); } /* * There may be an old instance of the dataset we're about to * create lying around from a previous run. If so, destroy it * and all of its snapshots. */ (void) dmu_objset_find(name, ztest_objset_destroy_cb, NULL, DS_FIND_CHILDREN | DS_FIND_SNAPSHOTS); /* * Verify that the destroyed dataset is no longer in the namespace. */ VERIFY3U(ENOENT, ==, ztest_dmu_objset_own(name, DMU_OST_OTHER, B_TRUE, B_TRUE, FTAG, &os)); /* * Verify that we can create a new dataset. */ error = ztest_dataset_create(name); if (error) { if (error == ENOSPC) { ztest_record_enospc(FTAG); goto out; } fatal(0, "dmu_objset_create(%s) = %d", name, error); } VERIFY0(ztest_dmu_objset_own(name, DMU_OST_OTHER, B_FALSE, B_TRUE, FTAG, &os)); ztest_zd_init(zdtmp, NULL, os); /* * Open the intent log for it. */ zilog = zil_open(os, ztest_get_data); /* * Put some objects in there, do a little I/O to them, * and randomly take a couple of snapshots along the way. */ iters = ztest_random(5); for (i = 0; i < iters; i++) { ztest_dmu_object_alloc_free(zdtmp, id); if (ztest_random(iters) == 0) (void) ztest_snapshot_create(name, i); } /* * Verify that we cannot create an existing dataset. */ VERIFY3U(EEXIST, ==, dmu_objset_create(name, DMU_OST_OTHER, 0, NULL, NULL, NULL)); /* * Verify that we can hold an objset that is also owned. */ VERIFY3U(0, ==, dmu_objset_hold(name, FTAG, &os2)); dmu_objset_rele(os2, FTAG); /* * Verify that we cannot own an objset that is already owned. */ VERIFY3U(EBUSY, ==, ztest_dmu_objset_own(name, DMU_OST_OTHER, B_FALSE, B_TRUE, FTAG, &os2)); zil_close(zilog); txg_wait_synced(spa_get_dsl(os->os_spa), 0); dmu_objset_disown(os, B_TRUE, FTAG); ztest_zd_fini(zdtmp); out: (void) rw_unlock(&ztest_name_lock); umem_free(zdtmp, sizeof (ztest_ds_t)); } /* * Verify that dmu_snapshot_{create,destroy,open,close} work as expected. */ void ztest_dmu_snapshot_create_destroy(ztest_ds_t *zd, uint64_t id) { (void) rw_rdlock(&ztest_name_lock); (void) ztest_snapshot_destroy(zd->zd_name, id); (void) ztest_snapshot_create(zd->zd_name, id); (void) rw_unlock(&ztest_name_lock); } /* * Cleanup non-standard snapshots and clones. */ void ztest_dsl_dataset_cleanup(char *osname, uint64_t id) { char *snap1name; char *clone1name; char *snap2name; char *clone2name; char *snap3name; int error; snap1name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); clone1name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); snap2name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); clone2name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); snap3name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); (void) snprintf(snap1name, ZFS_MAX_DATASET_NAME_LEN, "%s@s1_%llu", osname, (u_longlong_t)id); (void) snprintf(clone1name, ZFS_MAX_DATASET_NAME_LEN, "%s/c1_%llu", osname, (u_longlong_t)id); (void) snprintf(snap2name, ZFS_MAX_DATASET_NAME_LEN, "%s@s2_%llu", clone1name, (u_longlong_t)id); (void) snprintf(clone2name, ZFS_MAX_DATASET_NAME_LEN, "%s/c2_%llu", osname, (u_longlong_t)id); (void) snprintf(snap3name, ZFS_MAX_DATASET_NAME_LEN, "%s@s3_%llu", clone1name, (u_longlong_t)id); error = dsl_destroy_head(clone2name); if (error && error != ENOENT) fatal(0, "dsl_destroy_head(%s) = %d", clone2name, error); error = dsl_destroy_snapshot(snap3name, B_FALSE); if (error && error != ENOENT) fatal(0, "dsl_destroy_snapshot(%s) = %d", snap3name, error); error = dsl_destroy_snapshot(snap2name, B_FALSE); if (error && error != ENOENT) fatal(0, "dsl_destroy_snapshot(%s) = %d", snap2name, error); error = dsl_destroy_head(clone1name); if (error && error != ENOENT) fatal(0, "dsl_destroy_head(%s) = %d", clone1name, error); error = dsl_destroy_snapshot(snap1name, B_FALSE); if (error && error != ENOENT) fatal(0, "dsl_destroy_snapshot(%s) = %d", snap1name, error); umem_free(snap1name, ZFS_MAX_DATASET_NAME_LEN); umem_free(clone1name, ZFS_MAX_DATASET_NAME_LEN); umem_free(snap2name, ZFS_MAX_DATASET_NAME_LEN); umem_free(clone2name, ZFS_MAX_DATASET_NAME_LEN); umem_free(snap3name, ZFS_MAX_DATASET_NAME_LEN); } /* * Verify dsl_dataset_promote handles EBUSY */ void ztest_dsl_dataset_promote_busy(ztest_ds_t *zd, uint64_t id) { objset_t *os; char *snap1name; char *clone1name; char *snap2name; char *clone2name; char *snap3name; char *osname = zd->zd_name; int error; snap1name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); clone1name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); snap2name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); clone2name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); snap3name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); (void) rw_rdlock(&ztest_name_lock); ztest_dsl_dataset_cleanup(osname, id); (void) snprintf(snap1name, ZFS_MAX_DATASET_NAME_LEN, "%s@s1_%llu", osname, (u_longlong_t)id); (void) snprintf(clone1name, ZFS_MAX_DATASET_NAME_LEN, "%s/c1_%llu", osname, (u_longlong_t)id); (void) snprintf(snap2name, ZFS_MAX_DATASET_NAME_LEN, "%s@s2_%llu", clone1name, (u_longlong_t)id); (void) snprintf(clone2name, ZFS_MAX_DATASET_NAME_LEN, "%s/c2_%llu", osname, (u_longlong_t)id); (void) snprintf(snap3name, ZFS_MAX_DATASET_NAME_LEN, "%s@s3_%llu", clone1name, (u_longlong_t)id); error = dmu_objset_snapshot_one(osname, strchr(snap1name, '@') + 1); if (error && error != EEXIST) { if (error == ENOSPC) { ztest_record_enospc(FTAG); goto out; } fatal(0, "dmu_take_snapshot(%s) = %d", snap1name, error); } error = dmu_objset_clone(clone1name, snap1name); if (error) { if (error == ENOSPC) { ztest_record_enospc(FTAG); goto out; } fatal(0, "dmu_objset_create(%s) = %d", clone1name, error); } error = dmu_objset_snapshot_one(clone1name, strchr(snap2name, '@') + 1); if (error && error != EEXIST) { if (error == ENOSPC) { ztest_record_enospc(FTAG); goto out; } fatal(0, "dmu_open_snapshot(%s) = %d", snap2name, error); } error = dmu_objset_snapshot_one(clone1name, strchr(snap3name, '@') + 1); if (error && error != EEXIST) { if (error == ENOSPC) { ztest_record_enospc(FTAG); goto out; } fatal(0, "dmu_open_snapshot(%s) = %d", snap3name, error); } error = dmu_objset_clone(clone2name, snap3name); if (error) { if (error == ENOSPC) { ztest_record_enospc(FTAG); goto out; } fatal(0, "dmu_objset_create(%s) = %d", clone2name, error); } error = ztest_dmu_objset_own(snap2name, DMU_OST_ANY, B_TRUE, B_TRUE, FTAG, &os); if (error) fatal(0, "dmu_objset_own(%s) = %d", snap2name, error); error = dsl_dataset_promote(clone2name, NULL); if (error == ENOSPC) { dmu_objset_disown(os, B_TRUE, FTAG); ztest_record_enospc(FTAG); goto out; } if (error != EBUSY) fatal(0, "dsl_dataset_promote(%s), %d, not EBUSY", clone2name, error); dmu_objset_disown(os, B_TRUE, FTAG); out: ztest_dsl_dataset_cleanup(osname, id); (void) rw_unlock(&ztest_name_lock); umem_free(snap1name, ZFS_MAX_DATASET_NAME_LEN); umem_free(clone1name, ZFS_MAX_DATASET_NAME_LEN); umem_free(snap2name, ZFS_MAX_DATASET_NAME_LEN); umem_free(clone2name, ZFS_MAX_DATASET_NAME_LEN); umem_free(snap3name, ZFS_MAX_DATASET_NAME_LEN); } #undef OD_ARRAY_SIZE #define OD_ARRAY_SIZE 4 /* * Verify that dmu_object_{alloc,free} work as expected. */ void ztest_dmu_object_alloc_free(ztest_ds_t *zd, uint64_t id) { ztest_od_t *od; int batchsize; int size; int b; size = sizeof (ztest_od_t) * OD_ARRAY_SIZE; od = umem_alloc(size, UMEM_NOFAIL); batchsize = OD_ARRAY_SIZE; for (b = 0; b < batchsize; b++) ztest_od_init(od + b, id, FTAG, b, DMU_OT_UINT64_OTHER, 0, 0, 0); /* * Destroy the previous batch of objects, create a new batch, * and do some I/O on the new objects. */ if (ztest_object_init(zd, od, size, B_TRUE) != 0) return; while (ztest_random(4 * batchsize) != 0) ztest_io(zd, od[ztest_random(batchsize)].od_object, ztest_random(ZTEST_RANGE_LOCKS) << SPA_MAXBLOCKSHIFT); umem_free(od, size); } #undef OD_ARRAY_SIZE #define OD_ARRAY_SIZE 2 /* * Verify that dmu_{read,write} work as expected. */ void ztest_dmu_read_write(ztest_ds_t *zd, uint64_t id) { int size; ztest_od_t *od; objset_t *os = zd->zd_os; size = sizeof (ztest_od_t) * OD_ARRAY_SIZE; od = umem_alloc(size, UMEM_NOFAIL); dmu_tx_t *tx; int i, freeit, error; uint64_t n, s, txg; bufwad_t *packbuf, *bigbuf, *pack, *bigH, *bigT; uint64_t packobj, packoff, packsize, bigobj, bigoff, bigsize; uint64_t chunksize = (1000 + ztest_random(1000)) * sizeof (uint64_t); uint64_t regions = 997; uint64_t stride = 123456789ULL; uint64_t width = 40; int free_percent = 5; /* * This test uses two objects, packobj and bigobj, that are always * updated together (i.e. in the same tx) so that their contents are * in sync and can be compared. Their contents relate to each other * in a simple way: packobj is a dense array of 'bufwad' structures, * while bigobj is a sparse array of the same bufwads. Specifically, * for any index n, there are three bufwads that should be identical: * * packobj, at offset n * sizeof (bufwad_t) * bigobj, at the head of the nth chunk * bigobj, at the tail of the nth chunk * * The chunk size is arbitrary. It doesn't have to be a power of two, * and it doesn't have any relation to the object blocksize. * The only requirement is that it can hold at least two bufwads. * * Normally, we write the bufwad to each of these locations. * However, free_percent of the time we instead write zeroes to * packobj and perform a dmu_free_range() on bigobj. By comparing * bigobj to packobj, we can verify that the DMU is correctly * tracking which parts of an object are allocated and free, * and that the contents of the allocated blocks are correct. */ /* * Read the directory info. If it's the first time, set things up. */ ztest_od_init(od, id, FTAG, 0, DMU_OT_UINT64_OTHER, 0, 0, chunksize); ztest_od_init(od + 1, id, FTAG, 1, DMU_OT_UINT64_OTHER, 0, 0, chunksize); if (ztest_object_init(zd, od, size, B_FALSE) != 0) { umem_free(od, size); return; } bigobj = od[0].od_object; packobj = od[1].od_object; chunksize = od[0].od_gen; ASSERT(chunksize == od[1].od_gen); /* * Prefetch a random chunk of the big object. * Our aim here is to get some async reads in flight * for blocks that we may free below; the DMU should * handle this race correctly. */ n = ztest_random(regions) * stride + ztest_random(width); s = 1 + ztest_random(2 * width - 1); dmu_prefetch(os, bigobj, 0, n * chunksize, s * chunksize, ZIO_PRIORITY_SYNC_READ); /* * Pick a random index and compute the offsets into packobj and bigobj. */ n = ztest_random(regions) * stride + ztest_random(width); s = 1 + ztest_random(width - 1); packoff = n * sizeof (bufwad_t); packsize = s * sizeof (bufwad_t); bigoff = n * chunksize; bigsize = s * chunksize; packbuf = umem_alloc(packsize, UMEM_NOFAIL); bigbuf = umem_alloc(bigsize, UMEM_NOFAIL); /* * free_percent of the time, free a range of bigobj rather than * overwriting it. */ freeit = (ztest_random(100) < free_percent); /* * Read the current contents of our objects. */ error = dmu_read(os, packobj, packoff, packsize, packbuf, DMU_READ_PREFETCH); ASSERT0(error); error = dmu_read(os, bigobj, bigoff, bigsize, bigbuf, DMU_READ_PREFETCH); ASSERT0(error); /* * Get a tx for the mods to both packobj and bigobj. */ tx = dmu_tx_create(os); dmu_tx_hold_write(tx, packobj, packoff, packsize); if (freeit) dmu_tx_hold_free(tx, bigobj, bigoff, bigsize); else dmu_tx_hold_write(tx, bigobj, bigoff, bigsize); /* This accounts for setting the checksum/compression. */ dmu_tx_hold_bonus(tx, bigobj); txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG); if (txg == 0) { umem_free(packbuf, packsize); umem_free(bigbuf, bigsize); umem_free(od, size); return; } enum zio_checksum cksum; do { cksum = (enum zio_checksum) ztest_random_dsl_prop(ZFS_PROP_CHECKSUM); } while (cksum >= ZIO_CHECKSUM_LEGACY_FUNCTIONS); dmu_object_set_checksum(os, bigobj, cksum, tx); enum zio_compress comp; do { comp = (enum zio_compress) ztest_random_dsl_prop(ZFS_PROP_COMPRESSION); } while (comp >= ZIO_COMPRESS_LEGACY_FUNCTIONS); dmu_object_set_compress(os, bigobj, comp, tx); /* * For each index from n to n + s, verify that the existing bufwad * in packobj matches the bufwads at the head and tail of the * corresponding chunk in bigobj. Then update all three bufwads * with the new values we want to write out. */ for (i = 0; i < s; i++) { /* LINTED */ pack = (bufwad_t *)((char *)packbuf + i * sizeof (bufwad_t)); /* LINTED */ bigH = (bufwad_t *)((char *)bigbuf + i * chunksize); /* LINTED */ bigT = (bufwad_t *)((char *)bigH + chunksize) - 1; ASSERT((uintptr_t)bigH - (uintptr_t)bigbuf < bigsize); ASSERT((uintptr_t)bigT - (uintptr_t)bigbuf < bigsize); if (pack->bw_txg > txg) fatal(0, "future leak: got %llx, open txg is %llx", pack->bw_txg, txg); if (pack->bw_data != 0 && pack->bw_index != n + i) fatal(0, "wrong index: got %llx, wanted %llx+%llx", pack->bw_index, n, i); if (bcmp(pack, bigH, sizeof (bufwad_t)) != 0) fatal(0, "pack/bigH mismatch in %p/%p", pack, bigH); if (bcmp(pack, bigT, sizeof (bufwad_t)) != 0) fatal(0, "pack/bigT mismatch in %p/%p", pack, bigT); if (freeit) { bzero(pack, sizeof (bufwad_t)); } else { pack->bw_index = n + i; pack->bw_txg = txg; pack->bw_data = 1 + ztest_random(-2ULL); } *bigH = *pack; *bigT = *pack; } /* * We've verified all the old bufwads, and made new ones. * Now write them out. */ dmu_write(os, packobj, packoff, packsize, packbuf, tx); if (freeit) { if (ztest_opts.zo_verbose >= 7) { (void) printf("freeing offset %llx size %llx" " txg %llx\n", (u_longlong_t)bigoff, (u_longlong_t)bigsize, (u_longlong_t)txg); } VERIFY(0 == dmu_free_range(os, bigobj, bigoff, bigsize, tx)); } else { if (ztest_opts.zo_verbose >= 7) { (void) printf("writing offset %llx size %llx" " txg %llx\n", (u_longlong_t)bigoff, (u_longlong_t)bigsize, (u_longlong_t)txg); } dmu_write(os, bigobj, bigoff, bigsize, bigbuf, tx); } dmu_tx_commit(tx); /* * Sanity check the stuff we just wrote. */ { void *packcheck = umem_alloc(packsize, UMEM_NOFAIL); void *bigcheck = umem_alloc(bigsize, UMEM_NOFAIL); VERIFY(0 == dmu_read(os, packobj, packoff, packsize, packcheck, DMU_READ_PREFETCH)); VERIFY(0 == dmu_read(os, bigobj, bigoff, bigsize, bigcheck, DMU_READ_PREFETCH)); ASSERT(bcmp(packbuf, packcheck, packsize) == 0); ASSERT(bcmp(bigbuf, bigcheck, bigsize) == 0); umem_free(packcheck, packsize); umem_free(bigcheck, bigsize); } umem_free(packbuf, packsize); umem_free(bigbuf, bigsize); umem_free(od, size); } void compare_and_update_pbbufs(uint64_t s, bufwad_t *packbuf, bufwad_t *bigbuf, uint64_t bigsize, uint64_t n, uint64_t chunksize, uint64_t txg) { uint64_t i; bufwad_t *pack; bufwad_t *bigH; bufwad_t *bigT; /* * For each index from n to n + s, verify that the existing bufwad * in packobj matches the bufwads at the head and tail of the * corresponding chunk in bigobj. Then update all three bufwads * with the new values we want to write out. */ for (i = 0; i < s; i++) { /* LINTED */ pack = (bufwad_t *)((char *)packbuf + i * sizeof (bufwad_t)); /* LINTED */ bigH = (bufwad_t *)((char *)bigbuf + i * chunksize); /* LINTED */ bigT = (bufwad_t *)((char *)bigH + chunksize) - 1; ASSERT((uintptr_t)bigH - (uintptr_t)bigbuf < bigsize); ASSERT((uintptr_t)bigT - (uintptr_t)bigbuf < bigsize); if (pack->bw_txg > txg) fatal(0, "future leak: got %llx, open txg is %llx", pack->bw_txg, txg); if (pack->bw_data != 0 && pack->bw_index != n + i) fatal(0, "wrong index: got %llx, wanted %llx+%llx", pack->bw_index, n, i); if (bcmp(pack, bigH, sizeof (bufwad_t)) != 0) fatal(0, "pack/bigH mismatch in %p/%p", pack, bigH); if (bcmp(pack, bigT, sizeof (bufwad_t)) != 0) fatal(0, "pack/bigT mismatch in %p/%p", pack, bigT); pack->bw_index = n + i; pack->bw_txg = txg; pack->bw_data = 1 + ztest_random(-2ULL); *bigH = *pack; *bigT = *pack; } } #undef OD_ARRAY_SIZE #define OD_ARRAY_SIZE 2 void ztest_dmu_read_write_zcopy(ztest_ds_t *zd, uint64_t id) { objset_t *os = zd->zd_os; ztest_od_t *od; dmu_tx_t *tx; uint64_t i; int error; int size; uint64_t n, s, txg; bufwad_t *packbuf, *bigbuf; uint64_t packobj, packoff, packsize, bigobj, bigoff, bigsize; uint64_t blocksize = ztest_random_blocksize(); uint64_t chunksize = blocksize; uint64_t regions = 997; uint64_t stride = 123456789ULL; uint64_t width = 9; dmu_buf_t *bonus_db; arc_buf_t **bigbuf_arcbufs; dmu_object_info_t doi; size = sizeof (ztest_od_t) * OD_ARRAY_SIZE; od = umem_alloc(size, UMEM_NOFAIL); /* * This test uses two objects, packobj and bigobj, that are always * updated together (i.e. in the same tx) so that their contents are * in sync and can be compared. Their contents relate to each other * in a simple way: packobj is a dense array of 'bufwad' structures, * while bigobj is a sparse array of the same bufwads. Specifically, * for any index n, there are three bufwads that should be identical: * * packobj, at offset n * sizeof (bufwad_t) * bigobj, at the head of the nth chunk * bigobj, at the tail of the nth chunk * * The chunk size is set equal to bigobj block size so that * dmu_assign_arcbuf_by_dbuf() can be tested for object updates. */ /* * Read the directory info. If it's the first time, set things up. */ ztest_od_init(od, id, FTAG, 0, DMU_OT_UINT64_OTHER, blocksize, 0, 0); ztest_od_init(od + 1, id, FTAG, 1, DMU_OT_UINT64_OTHER, 0, 0, chunksize); if (ztest_object_init(zd, od, size, B_FALSE) != 0) { umem_free(od, size); return; } bigobj = od[0].od_object; packobj = od[1].od_object; blocksize = od[0].od_blocksize; chunksize = blocksize; ASSERT(chunksize == od[1].od_gen); VERIFY(dmu_object_info(os, bigobj, &doi) == 0); VERIFY(ISP2(doi.doi_data_block_size)); VERIFY(chunksize == doi.doi_data_block_size); VERIFY(chunksize >= 2 * sizeof (bufwad_t)); /* * Pick a random index and compute the offsets into packobj and bigobj. */ n = ztest_random(regions) * stride + ztest_random(width); s = 1 + ztest_random(width - 1); packoff = n * sizeof (bufwad_t); packsize = s * sizeof (bufwad_t); bigoff = n * chunksize; bigsize = s * chunksize; packbuf = umem_zalloc(packsize, UMEM_NOFAIL); bigbuf = umem_zalloc(bigsize, UMEM_NOFAIL); VERIFY3U(0, ==, dmu_bonus_hold(os, bigobj, FTAG, &bonus_db)); bigbuf_arcbufs = umem_zalloc(2 * s * sizeof (arc_buf_t *), UMEM_NOFAIL); /* * Iteration 0 test zcopy for DB_UNCACHED dbufs. * Iteration 1 test zcopy to already referenced dbufs. * Iteration 2 test zcopy to dirty dbuf in the same txg. * Iteration 3 test zcopy to dbuf dirty in previous txg. * Iteration 4 test zcopy when dbuf is no longer dirty. * Iteration 5 test zcopy when it can't be done. * Iteration 6 one more zcopy write. */ for (i = 0; i < 7; i++) { uint64_t j; uint64_t off; /* * In iteration 5 (i == 5) use arcbufs * that don't match bigobj blksz to test * dmu_assign_arcbuf_by_dbuf() when it can't directly * assign an arcbuf to a dbuf. */ for (j = 0; j < s; j++) { if (i != 5 || chunksize < (SPA_MINBLOCKSIZE * 2)) { bigbuf_arcbufs[j] = dmu_request_arcbuf(bonus_db, chunksize); } else { bigbuf_arcbufs[2 * j] = dmu_request_arcbuf(bonus_db, chunksize / 2); bigbuf_arcbufs[2 * j + 1] = dmu_request_arcbuf(bonus_db, chunksize / 2); } } /* * Get a tx for the mods to both packobj and bigobj. */ tx = dmu_tx_create(os); dmu_tx_hold_write(tx, packobj, packoff, packsize); dmu_tx_hold_write(tx, bigobj, bigoff, bigsize); txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG); if (txg == 0) { umem_free(packbuf, packsize); umem_free(bigbuf, bigsize); for (j = 0; j < s; j++) { if (i != 5 || chunksize < (SPA_MINBLOCKSIZE * 2)) { dmu_return_arcbuf(bigbuf_arcbufs[j]); } else { dmu_return_arcbuf( bigbuf_arcbufs[2 * j]); dmu_return_arcbuf( bigbuf_arcbufs[2 * j + 1]); } } umem_free(bigbuf_arcbufs, 2 * s * sizeof (arc_buf_t *)); umem_free(od, size); dmu_buf_rele(bonus_db, FTAG); return; } /* * 50% of the time don't read objects in the 1st iteration to * test dmu_assign_arcbuf_by_dbuf() for the case when there are * no existing dbufs for the specified offsets. */ if (i != 0 || ztest_random(2) != 0) { error = dmu_read(os, packobj, packoff, packsize, packbuf, DMU_READ_PREFETCH); ASSERT0(error); error = dmu_read(os, bigobj, bigoff, bigsize, bigbuf, DMU_READ_PREFETCH); ASSERT0(error); } compare_and_update_pbbufs(s, packbuf, bigbuf, bigsize, n, chunksize, txg); /* * We've verified all the old bufwads, and made new ones. * Now write them out. */ dmu_write(os, packobj, packoff, packsize, packbuf, tx); if (ztest_opts.zo_verbose >= 7) { (void) printf("writing offset %llx size %llx" " txg %llx\n", (u_longlong_t)bigoff, (u_longlong_t)bigsize, (u_longlong_t)txg); } for (off = bigoff, j = 0; j < s; j++, off += chunksize) { dmu_buf_t *dbt; if (i != 5 || chunksize < (SPA_MINBLOCKSIZE * 2)) { bcopy((caddr_t)bigbuf + (off - bigoff), bigbuf_arcbufs[j]->b_data, chunksize); } else { bcopy((caddr_t)bigbuf + (off - bigoff), bigbuf_arcbufs[2 * j]->b_data, chunksize / 2); bcopy((caddr_t)bigbuf + (off - bigoff) + chunksize / 2, bigbuf_arcbufs[2 * j + 1]->b_data, chunksize / 2); } if (i == 1) { VERIFY(dmu_buf_hold(os, bigobj, off, FTAG, &dbt, DMU_READ_NO_PREFETCH) == 0); } if (i != 5 || chunksize < (SPA_MINBLOCKSIZE * 2)) { dmu_assign_arcbuf_by_dbuf(bonus_db, off, bigbuf_arcbufs[j], tx); } else { dmu_assign_arcbuf_by_dbuf(bonus_db, off, bigbuf_arcbufs[2 * j], tx); dmu_assign_arcbuf_by_dbuf(bonus_db, off + chunksize / 2, bigbuf_arcbufs[2 * j + 1], tx); } if (i == 1) { dmu_buf_rele(dbt, FTAG); } } dmu_tx_commit(tx); /* * Sanity check the stuff we just wrote. */ { void *packcheck = umem_alloc(packsize, UMEM_NOFAIL); void *bigcheck = umem_alloc(bigsize, UMEM_NOFAIL); VERIFY(0 == dmu_read(os, packobj, packoff, packsize, packcheck, DMU_READ_PREFETCH)); VERIFY(0 == dmu_read(os, bigobj, bigoff, bigsize, bigcheck, DMU_READ_PREFETCH)); ASSERT(bcmp(packbuf, packcheck, packsize) == 0); ASSERT(bcmp(bigbuf, bigcheck, bigsize) == 0); umem_free(packcheck, packsize); umem_free(bigcheck, bigsize); } if (i == 2) { txg_wait_open(dmu_objset_pool(os), 0); } else if (i == 3) { txg_wait_synced(dmu_objset_pool(os), 0); } } dmu_buf_rele(bonus_db, FTAG); umem_free(packbuf, packsize); umem_free(bigbuf, bigsize); umem_free(bigbuf_arcbufs, 2 * s * sizeof (arc_buf_t *)); umem_free(od, size); } /* ARGSUSED */ void ztest_dmu_write_parallel(ztest_ds_t *zd, uint64_t id) { ztest_od_t *od; od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL); uint64_t offset = (1ULL << (ztest_random(20) + 43)) + (ztest_random(ZTEST_RANGE_LOCKS) << SPA_MAXBLOCKSHIFT); /* * Have multiple threads write to large offsets in an object * to verify that parallel writes to an object -- even to the * same blocks within the object -- doesn't cause any trouble. */ ztest_od_init(od, ID_PARALLEL, FTAG, 0, DMU_OT_UINT64_OTHER, 0, 0, 0); if (ztest_object_init(zd, od, sizeof (ztest_od_t), B_FALSE) != 0) return; while (ztest_random(10) != 0) ztest_io(zd, od->od_object, offset); umem_free(od, sizeof (ztest_od_t)); } void ztest_dmu_prealloc(ztest_ds_t *zd, uint64_t id) { ztest_od_t *od; uint64_t offset = (1ULL << (ztest_random(4) + SPA_MAXBLOCKSHIFT)) + (ztest_random(ZTEST_RANGE_LOCKS) << SPA_MAXBLOCKSHIFT); uint64_t count = ztest_random(20) + 1; uint64_t blocksize = ztest_random_blocksize(); void *data; od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL); ztest_od_init(od, id, FTAG, 0, DMU_OT_UINT64_OTHER, blocksize, 0, 0); if (ztest_object_init(zd, od, sizeof (ztest_od_t), !ztest_random(2)) != 0) { umem_free(od, sizeof (ztest_od_t)); return; } if (ztest_truncate(zd, od->od_object, offset, count * blocksize) != 0) { umem_free(od, sizeof (ztest_od_t)); return; } ztest_prealloc(zd, od->od_object, offset, count * blocksize); data = umem_zalloc(blocksize, UMEM_NOFAIL); while (ztest_random(count) != 0) { uint64_t randoff = offset + (ztest_random(count) * blocksize); if (ztest_write(zd, od->od_object, randoff, blocksize, data) != 0) break; while (ztest_random(4) != 0) ztest_io(zd, od->od_object, randoff); } umem_free(data, blocksize); umem_free(od, sizeof (ztest_od_t)); } /* * Verify that zap_{create,destroy,add,remove,update} work as expected. */ #define ZTEST_ZAP_MIN_INTS 1 #define ZTEST_ZAP_MAX_INTS 4 #define ZTEST_ZAP_MAX_PROPS 1000 void ztest_zap(ztest_ds_t *zd, uint64_t id) { objset_t *os = zd->zd_os; ztest_od_t *od; uint64_t object; uint64_t txg, last_txg; uint64_t value[ZTEST_ZAP_MAX_INTS]; uint64_t zl_ints, zl_intsize, prop; int i, ints; dmu_tx_t *tx; char propname[100], txgname[100]; int error; char *hc[2] = { "s.acl.h", ".s.open.h.hyLZlg" }; od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL); ztest_od_init(od, id, FTAG, 0, DMU_OT_ZAP_OTHER, 0, 0, 0); if (ztest_object_init(zd, od, sizeof (ztest_od_t), !ztest_random(2)) != 0) goto out; object = od->od_object; /* * Generate a known hash collision, and verify that * we can lookup and remove both entries. */ tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, object, B_TRUE, NULL); txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG); if (txg == 0) goto out; for (i = 0; i < 2; i++) { value[i] = i; VERIFY3U(0, ==, zap_add(os, object, hc[i], sizeof (uint64_t), 1, &value[i], tx)); } for (i = 0; i < 2; i++) { VERIFY3U(EEXIST, ==, zap_add(os, object, hc[i], sizeof (uint64_t), 1, &value[i], tx)); VERIFY3U(0, ==, zap_length(os, object, hc[i], &zl_intsize, &zl_ints)); ASSERT3U(zl_intsize, ==, sizeof (uint64_t)); ASSERT3U(zl_ints, ==, 1); } for (i = 0; i < 2; i++) { VERIFY3U(0, ==, zap_remove(os, object, hc[i], tx)); } dmu_tx_commit(tx); /* * Generate a buch of random entries. */ ints = MAX(ZTEST_ZAP_MIN_INTS, object % ZTEST_ZAP_MAX_INTS); prop = ztest_random(ZTEST_ZAP_MAX_PROPS); (void) sprintf(propname, "prop_%llu", (u_longlong_t)prop); (void) sprintf(txgname, "txg_%llu", (u_longlong_t)prop); bzero(value, sizeof (value)); last_txg = 0; /* * If these zap entries already exist, validate their contents. */ error = zap_length(os, object, txgname, &zl_intsize, &zl_ints); if (error == 0) { ASSERT3U(zl_intsize, ==, sizeof (uint64_t)); ASSERT3U(zl_ints, ==, 1); VERIFY(zap_lookup(os, object, txgname, zl_intsize, zl_ints, &last_txg) == 0); VERIFY(zap_length(os, object, propname, &zl_intsize, &zl_ints) == 0); ASSERT3U(zl_intsize, ==, sizeof (uint64_t)); ASSERT3U(zl_ints, ==, ints); VERIFY(zap_lookup(os, object, propname, zl_intsize, zl_ints, value) == 0); for (i = 0; i < ints; i++) { ASSERT3U(value[i], ==, last_txg + object + i); } } else { ASSERT3U(error, ==, ENOENT); } /* * Atomically update two entries in our zap object. * The first is named txg_%llu, and contains the txg * in which the property was last updated. The second * is named prop_%llu, and the nth element of its value * should be txg + object + n. */ tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, object, B_TRUE, NULL); txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG); if (txg == 0) goto out; if (last_txg > txg) fatal(0, "zap future leak: old %llu new %llu", last_txg, txg); for (i = 0; i < ints; i++) value[i] = txg + object + i; VERIFY3U(0, ==, zap_update(os, object, txgname, sizeof (uint64_t), 1, &txg, tx)); VERIFY3U(0, ==, zap_update(os, object, propname, sizeof (uint64_t), ints, value, tx)); dmu_tx_commit(tx); /* * Remove a random pair of entries. */ prop = ztest_random(ZTEST_ZAP_MAX_PROPS); (void) sprintf(propname, "prop_%llu", (u_longlong_t)prop); (void) sprintf(txgname, "txg_%llu", (u_longlong_t)prop); error = zap_length(os, object, txgname, &zl_intsize, &zl_ints); if (error == ENOENT) goto out; ASSERT0(error); tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, object, B_TRUE, NULL); txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG); if (txg == 0) goto out; VERIFY3U(0, ==, zap_remove(os, object, txgname, tx)); VERIFY3U(0, ==, zap_remove(os, object, propname, tx)); dmu_tx_commit(tx); out: umem_free(od, sizeof (ztest_od_t)); } /* * Testcase to test the upgrading of a microzap to fatzap. */ void ztest_fzap(ztest_ds_t *zd, uint64_t id) { objset_t *os = zd->zd_os; ztest_od_t *od; uint64_t object, txg; int i; od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL); ztest_od_init(od, id, FTAG, 0, DMU_OT_ZAP_OTHER, 0, 0, 0); if (ztest_object_init(zd, od, sizeof (ztest_od_t), !ztest_random(2)) != 0) goto out; object = od->od_object; /* * Add entries to this ZAP and make sure it spills over * and gets upgraded to a fatzap. Also, since we are adding * 2050 entries we should see ptrtbl growth and leaf-block split. */ for (i = 0; i < 2050; i++) { char name[ZFS_MAX_DATASET_NAME_LEN]; uint64_t value = i; dmu_tx_t *tx; int error; (void) snprintf(name, sizeof (name), "fzap-%llu-%llu", (u_longlong_t)id, (u_longlong_t)value); tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, object, B_TRUE, name); txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG); if (txg == 0) goto out; error = zap_add(os, object, name, sizeof (uint64_t), 1, &value, tx); ASSERT(error == 0 || error == EEXIST); dmu_tx_commit(tx); } out: umem_free(od, sizeof (ztest_od_t)); } /* ARGSUSED */ void ztest_zap_parallel(ztest_ds_t *zd, uint64_t id) { objset_t *os = zd->zd_os; ztest_od_t *od; uint64_t txg, object, count, wsize, wc, zl_wsize, zl_wc; dmu_tx_t *tx; int i, namelen, error; int micro = ztest_random(2); char name[20], string_value[20]; void *data; od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL); ztest_od_init(od, ID_PARALLEL, FTAG, micro, DMU_OT_ZAP_OTHER, 0, 0, 0); if (ztest_object_init(zd, od, sizeof (ztest_od_t), B_FALSE) != 0) { umem_free(od, sizeof (ztest_od_t)); return; } object = od->od_object; /* * Generate a random name of the form 'xxx.....' where each * x is a random printable character and the dots are dots. * There are 94 such characters, and the name length goes from * 6 to 20, so there are 94^3 * 15 = 12,458,760 possible names. */ namelen = ztest_random(sizeof (name) - 5) + 5 + 1; for (i = 0; i < 3; i++) name[i] = '!' + ztest_random('~' - '!' + 1); for (; i < namelen - 1; i++) name[i] = '.'; name[i] = '\0'; if ((namelen & 1) || micro) { wsize = sizeof (txg); wc = 1; data = &txg; } else { wsize = 1; wc = namelen; data = string_value; } count = -1ULL; VERIFY0(zap_count(os, object, &count)); ASSERT(count != -1ULL); /* * Select an operation: length, lookup, add, update, remove. */ i = ztest_random(5); if (i >= 2) { tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, object, B_TRUE, NULL); txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG); if (txg == 0) { umem_free(od, sizeof (ztest_od_t)); return; } bcopy(name, string_value, namelen); } else { tx = NULL; txg = 0; bzero(string_value, namelen); } switch (i) { case 0: error = zap_length(os, object, name, &zl_wsize, &zl_wc); if (error == 0) { ASSERT3U(wsize, ==, zl_wsize); ASSERT3U(wc, ==, zl_wc); } else { ASSERT3U(error, ==, ENOENT); } break; case 1: error = zap_lookup(os, object, name, wsize, wc, data); if (error == 0) { if (data == string_value && bcmp(name, data, namelen) != 0) fatal(0, "name '%s' != val '%s' len %d", name, data, namelen); } else { ASSERT3U(error, ==, ENOENT); } break; case 2: error = zap_add(os, object, name, wsize, wc, data, tx); ASSERT(error == 0 || error == EEXIST); break; case 3: VERIFY(zap_update(os, object, name, wsize, wc, data, tx) == 0); break; case 4: error = zap_remove(os, object, name, tx); ASSERT(error == 0 || error == ENOENT); break; } if (tx != NULL) dmu_tx_commit(tx); umem_free(od, sizeof (ztest_od_t)); } /* * Commit callback data. */ typedef struct ztest_cb_data { list_node_t zcd_node; uint64_t zcd_txg; int zcd_expected_err; boolean_t zcd_added; boolean_t zcd_called; spa_t *zcd_spa; } ztest_cb_data_t; /* This is the actual commit callback function */ static void ztest_commit_callback(void *arg, int error) { ztest_cb_data_t *data = arg; uint64_t synced_txg; VERIFY(data != NULL); VERIFY3S(data->zcd_expected_err, ==, error); VERIFY(!data->zcd_called); synced_txg = spa_last_synced_txg(data->zcd_spa); if (data->zcd_txg > synced_txg) fatal(0, "commit callback of txg %" PRIu64 " called prematurely" ", last synced txg = %" PRIu64 "\n", data->zcd_txg, synced_txg); data->zcd_called = B_TRUE; if (error == ECANCELED) { ASSERT0(data->zcd_txg); ASSERT(!data->zcd_added); /* * The private callback data should be destroyed here, but * since we are going to check the zcd_called field after * dmu_tx_abort(), we will destroy it there. */ return; } ASSERT(data->zcd_added); ASSERT3U(data->zcd_txg, !=, 0); (void) mutex_enter(&zcl.zcl_callbacks_lock); /* See if this cb was called more quickly */ if ((synced_txg - data->zcd_txg) < zc_min_txg_delay) zc_min_txg_delay = synced_txg - data->zcd_txg; /* Remove our callback from the list */ list_remove(&zcl.zcl_callbacks, data); (void) mutex_exit(&zcl.zcl_callbacks_lock); umem_free(data, sizeof (ztest_cb_data_t)); } /* Allocate and initialize callback data structure */ static ztest_cb_data_t * ztest_create_cb_data(objset_t *os, uint64_t txg) { ztest_cb_data_t *cb_data; cb_data = umem_zalloc(sizeof (ztest_cb_data_t), UMEM_NOFAIL); cb_data->zcd_txg = txg; cb_data->zcd_spa = dmu_objset_spa(os); list_link_init(&cb_data->zcd_node); return (cb_data); } /* * Commit callback test. */ void ztest_dmu_commit_callbacks(ztest_ds_t *zd, uint64_t id) { objset_t *os = zd->zd_os; ztest_od_t *od; dmu_tx_t *tx; ztest_cb_data_t *cb_data[3], *tmp_cb; uint64_t old_txg, txg; int i, error = 0; od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL); ztest_od_init(od, id, FTAG, 0, DMU_OT_UINT64_OTHER, 0, 0, 0); if (ztest_object_init(zd, od, sizeof (ztest_od_t), B_FALSE) != 0) { umem_free(od, sizeof (ztest_od_t)); return; } tx = dmu_tx_create(os); cb_data[0] = ztest_create_cb_data(os, 0); dmu_tx_callback_register(tx, ztest_commit_callback, cb_data[0]); dmu_tx_hold_write(tx, od->od_object, 0, sizeof (uint64_t)); /* Every once in a while, abort the transaction on purpose */ if (ztest_random(100) == 0) error = -1; if (!error) error = dmu_tx_assign(tx, TXG_NOWAIT); txg = error ? 0 : dmu_tx_get_txg(tx); cb_data[0]->zcd_txg = txg; cb_data[1] = ztest_create_cb_data(os, txg); dmu_tx_callback_register(tx, ztest_commit_callback, cb_data[1]); if (error) { /* * It's not a strict requirement to call the registered * callbacks from inside dmu_tx_abort(), but that's what * it's supposed to happen in the current implementation * so we will check for that. */ for (i = 0; i < 2; i++) { cb_data[i]->zcd_expected_err = ECANCELED; VERIFY(!cb_data[i]->zcd_called); } dmu_tx_abort(tx); for (i = 0; i < 2; i++) { VERIFY(cb_data[i]->zcd_called); umem_free(cb_data[i], sizeof (ztest_cb_data_t)); } umem_free(od, sizeof (ztest_od_t)); return; } cb_data[2] = ztest_create_cb_data(os, txg); dmu_tx_callback_register(tx, ztest_commit_callback, cb_data[2]); /* * Read existing data to make sure there isn't a future leak. */ VERIFY(0 == dmu_read(os, od->od_object, 0, sizeof (uint64_t), &old_txg, DMU_READ_PREFETCH)); if (old_txg > txg) fatal(0, "future leak: got %" PRIu64 ", open txg is %" PRIu64, old_txg, txg); dmu_write(os, od->od_object, 0, sizeof (uint64_t), &txg, tx); (void) mutex_enter(&zcl.zcl_callbacks_lock); /* * Since commit callbacks don't have any ordering requirement and since * it is theoretically possible for a commit callback to be called * after an arbitrary amount of time has elapsed since its txg has been * synced, it is difficult to reliably determine whether a commit * callback hasn't been called due to high load or due to a flawed * implementation. * * In practice, we will assume that if after a certain number of txgs a * commit callback hasn't been called, then most likely there's an * implementation bug.. */ tmp_cb = list_head(&zcl.zcl_callbacks); if (tmp_cb != NULL && tmp_cb->zcd_txg + ZTEST_COMMIT_CB_THRESH < txg) { fatal(0, "Commit callback threshold exceeded, oldest txg: %" PRIu64 ", open txg: %" PRIu64 "\n", tmp_cb->zcd_txg, txg); } /* * Let's find the place to insert our callbacks. * * Even though the list is ordered by txg, it is possible for the * insertion point to not be the end because our txg may already be * quiescing at this point and other callbacks in the open txg * (from other objsets) may have sneaked in. */ tmp_cb = list_tail(&zcl.zcl_callbacks); while (tmp_cb != NULL && tmp_cb->zcd_txg > txg) tmp_cb = list_prev(&zcl.zcl_callbacks, tmp_cb); /* Add the 3 callbacks to the list */ for (i = 0; i < 3; i++) { if (tmp_cb == NULL) list_insert_head(&zcl.zcl_callbacks, cb_data[i]); else list_insert_after(&zcl.zcl_callbacks, tmp_cb, cb_data[i]); cb_data[i]->zcd_added = B_TRUE; VERIFY(!cb_data[i]->zcd_called); tmp_cb = cb_data[i]; } zc_cb_counter += 3; (void) mutex_exit(&zcl.zcl_callbacks_lock); dmu_tx_commit(tx); umem_free(od, sizeof (ztest_od_t)); } /* * Visit each object in the dataset. Verify that its properties * are consistent what was stored in the block tag when it was created, * and that its unused bonus buffer space has not been overwritten. */ /* ARGSUSED */ void ztest_verify_dnode_bt(ztest_ds_t *zd, uint64_t id) { objset_t *os = zd->zd_os; uint64_t obj; int err = 0; for (obj = 0; err == 0; err = dmu_object_next(os, &obj, FALSE, 0)) { ztest_block_tag_t *bt = NULL; dmu_object_info_t doi; dmu_buf_t *db; if (dmu_bonus_hold(os, obj, FTAG, &db) != 0) continue; dmu_object_info_from_db(db, &doi); if (doi.doi_bonus_size >= sizeof (*bt)) bt = ztest_bt_bonus(db); if (bt && bt->bt_magic == BT_MAGIC) { ztest_bt_verify(bt, os, obj, doi.doi_dnodesize, bt->bt_offset, bt->bt_gen, bt->bt_txg, bt->bt_crtxg); ztest_verify_unused_bonus(db, bt, obj, os, bt->bt_gen); } dmu_buf_rele(db, FTAG); } } /* ARGSUSED */ void ztest_dsl_prop_get_set(ztest_ds_t *zd, uint64_t id) { zfs_prop_t proplist[] = { ZFS_PROP_CHECKSUM, ZFS_PROP_COMPRESSION, ZFS_PROP_COPIES, ZFS_PROP_DEDUP }; int p; (void) rw_rdlock(&ztest_name_lock); for (p = 0; p < sizeof (proplist) / sizeof (proplist[0]); p++) (void) ztest_dsl_prop_set_uint64(zd->zd_name, proplist[p], ztest_random_dsl_prop(proplist[p]), (int)ztest_random(2)); VERIFY0(ztest_dsl_prop_set_uint64(zd->zd_name, ZFS_PROP_RECORDSIZE, ztest_random_blocksize(), (int)ztest_random(2))); (void) rw_unlock(&ztest_name_lock); } /* ARGSUSED */ void ztest_spa_prop_get_set(ztest_ds_t *zd, uint64_t id) { nvlist_t *props = NULL; (void) rw_rdlock(&ztest_name_lock); (void) ztest_spa_prop_set_uint64(ZPOOL_PROP_DEDUPDITTO, ZIO_DEDUPDITTO_MIN + ztest_random(ZIO_DEDUPDITTO_MIN)); VERIFY0(spa_prop_get(ztest_spa, &props)); if (ztest_opts.zo_verbose >= 6) dump_nvlist(props, 4); nvlist_free(props); (void) rw_unlock(&ztest_name_lock); } static int user_release_one(const char *snapname, const char *holdname) { nvlist_t *snaps, *holds; int error; snaps = fnvlist_alloc(); holds = fnvlist_alloc(); fnvlist_add_boolean(holds, holdname); fnvlist_add_nvlist(snaps, snapname, holds); fnvlist_free(holds); error = dsl_dataset_user_release(snaps, NULL); fnvlist_free(snaps); return (error); } /* * Test snapshot hold/release and deferred destroy. */ void ztest_dmu_snapshot_hold(ztest_ds_t *zd, uint64_t id) { int error; objset_t *os = zd->zd_os; objset_t *origin; char snapname[100]; char fullname[100]; char clonename[100]; char tag[100]; char osname[ZFS_MAX_DATASET_NAME_LEN]; nvlist_t *holds; (void) rw_rdlock(&ztest_name_lock); dmu_objset_name(os, osname); (void) snprintf(snapname, sizeof (snapname), "sh1_%llu", (u_longlong_t)id); (void) snprintf(fullname, sizeof (fullname), "%s@%s", osname, snapname); (void) snprintf(clonename, sizeof (clonename), "%s/ch1_%llu", osname, (u_longlong_t)id); (void) snprintf(tag, sizeof (tag), "tag_%llu", (u_longlong_t)id); /* * Clean up from any previous run. */ error = dsl_destroy_head(clonename); if (error != ENOENT) ASSERT0(error); error = user_release_one(fullname, tag); if (error != ESRCH && error != ENOENT) ASSERT0(error); error = dsl_destroy_snapshot(fullname, B_FALSE); if (error != ENOENT) ASSERT0(error); /* * Create snapshot, clone it, mark snap for deferred destroy, * destroy clone, verify snap was also destroyed. */ error = dmu_objset_snapshot_one(osname, snapname); if (error) { if (error == ENOSPC) { ztest_record_enospc("dmu_objset_snapshot"); goto out; } fatal(0, "dmu_objset_snapshot(%s) = %d", fullname, error); } error = dmu_objset_clone(clonename, fullname); if (error) { if (error == ENOSPC) { ztest_record_enospc("dmu_objset_clone"); goto out; } fatal(0, "dmu_objset_clone(%s) = %d", clonename, error); } error = dsl_destroy_snapshot(fullname, B_TRUE); if (error) { fatal(0, "dsl_destroy_snapshot(%s, B_TRUE) = %d", fullname, error); } error = dsl_destroy_head(clonename); if (error) fatal(0, "dsl_destroy_head(%s) = %d", clonename, error); error = dmu_objset_hold(fullname, FTAG, &origin); if (error != ENOENT) fatal(0, "dmu_objset_hold(%s) = %d", fullname, error); /* * Create snapshot, add temporary hold, verify that we can't * destroy a held snapshot, mark for deferred destroy, * release hold, verify snapshot was destroyed. */ error = dmu_objset_snapshot_one(osname, snapname); if (error) { if (error == ENOSPC) { ztest_record_enospc("dmu_objset_snapshot"); goto out; } fatal(0, "dmu_objset_snapshot(%s) = %d", fullname, error); } holds = fnvlist_alloc(); fnvlist_add_string(holds, fullname, tag); error = dsl_dataset_user_hold(holds, 0, NULL); fnvlist_free(holds); if (error == ENOSPC) { ztest_record_enospc("dsl_dataset_user_hold"); goto out; } else if (error) { fatal(0, "dsl_dataset_user_hold(%s, %s) = %u", fullname, tag, error); } error = dsl_destroy_snapshot(fullname, B_FALSE); if (error != EBUSY) { fatal(0, "dsl_destroy_snapshot(%s, B_FALSE) = %d", fullname, error); } error = dsl_destroy_snapshot(fullname, B_TRUE); if (error) { fatal(0, "dsl_destroy_snapshot(%s, B_TRUE) = %d", fullname, error); } error = user_release_one(fullname, tag); if (error) fatal(0, "user_release_one(%s, %s) = %d", fullname, tag, error); VERIFY3U(dmu_objset_hold(fullname, FTAG, &origin), ==, ENOENT); out: (void) rw_unlock(&ztest_name_lock); } /* * Inject random faults into the on-disk data. */ /* ARGSUSED */ void ztest_fault_inject(ztest_ds_t *zd, uint64_t id) { ztest_shared_t *zs = ztest_shared; spa_t *spa = ztest_spa; int fd; uint64_t offset; uint64_t leaves; uint64_t bad = 0x1990c0ffeedecadeull; uint64_t top, leaf; char *path0; char *pathrand; size_t fsize; int bshift = SPA_MAXBLOCKSHIFT + 2; int iters = 1000; int maxfaults; int mirror_save; vdev_t *vd0 = NULL; uint64_t guid0 = 0; boolean_t islog = B_FALSE; path0 = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); pathrand = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); mutex_enter(&ztest_vdev_lock); maxfaults = MAXFAULTS(); leaves = MAX(zs->zs_mirrors, 1) * ztest_opts.zo_raidz; mirror_save = zs->zs_mirrors; mutex_exit(&ztest_vdev_lock); ASSERT(leaves >= 1); /* * Grab the name lock as reader. There are some operations * which don't like to have their vdevs changed while * they are in progress (i.e. spa_change_guid). Those * operations will have grabbed the name lock as writer. */ (void) rw_rdlock(&ztest_name_lock); /* * We need SCL_STATE here because we're going to look at vd0->vdev_tsd. */ spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); if (ztest_random(2) == 0) { /* * Inject errors on a normal data device or slog device. */ top = ztest_random_vdev_top(spa, B_TRUE); leaf = ztest_random(leaves) + zs->zs_splits; /* * Generate paths to the first leaf in this top-level vdev, * and to the random leaf we selected. We'll induce transient * write failures and random online/offline activity on leaf 0, * and we'll write random garbage to the randomly chosen leaf. */ (void) snprintf(path0, MAXPATHLEN, ztest_dev_template, ztest_opts.zo_dir, ztest_opts.zo_pool, top * leaves + zs->zs_splits); (void) snprintf(pathrand, MAXPATHLEN, ztest_dev_template, ztest_opts.zo_dir, ztest_opts.zo_pool, top * leaves + leaf); vd0 = vdev_lookup_by_path(spa->spa_root_vdev, path0); if (vd0 != NULL && vd0->vdev_top->vdev_islog) islog = B_TRUE; /* * If the top-level vdev needs to be resilvered * then we only allow faults on the device that is * resilvering. */ if (vd0 != NULL && maxfaults != 1 && (!vdev_resilver_needed(vd0->vdev_top, NULL, NULL) || vd0->vdev_resilver_txg != 0)) { /* * Make vd0 explicitly claim to be unreadable, * or unwriteable, or reach behind its back * and close the underlying fd. We can do this if * maxfaults == 0 because we'll fail and reexecute, * and we can do it if maxfaults >= 2 because we'll * have enough redundancy. If maxfaults == 1, the * combination of this with injection of random data * corruption below exceeds the pool's fault tolerance. */ vdev_file_t *vf = vd0->vdev_tsd; if (vf != NULL && ztest_random(3) == 0) { (void) close(vf->vf_vnode->v_fd); vf->vf_vnode->v_fd = -1; } else if (ztest_random(2) == 0) { vd0->vdev_cant_read = B_TRUE; } else { vd0->vdev_cant_write = B_TRUE; } guid0 = vd0->vdev_guid; } } else { /* * Inject errors on an l2cache device. */ spa_aux_vdev_t *sav = &spa->spa_l2cache; if (sav->sav_count == 0) { spa_config_exit(spa, SCL_STATE, FTAG); (void) rw_unlock(&ztest_name_lock); goto out; } vd0 = sav->sav_vdevs[ztest_random(sav->sav_count)]; guid0 = vd0->vdev_guid; (void) strcpy(path0, vd0->vdev_path); (void) strcpy(pathrand, vd0->vdev_path); leaf = 0; leaves = 1; maxfaults = INT_MAX; /* no limit on cache devices */ } spa_config_exit(spa, SCL_STATE, FTAG); (void) rw_unlock(&ztest_name_lock); /* * If we can tolerate two or more faults, or we're dealing * with a slog, randomly online/offline vd0. */ if ((maxfaults >= 2 || islog) && guid0 != 0) { if (ztest_random(10) < 6) { int flags = (ztest_random(2) == 0 ? ZFS_OFFLINE_TEMPORARY : 0); /* * We have to grab the zs_name_lock as writer to * prevent a race between offlining a slog and * destroying a dataset. Offlining the slog will * grab a reference on the dataset which may cause * dsl_destroy_head() to fail with EBUSY thus * leaving the dataset in an inconsistent state. */ if (islog) (void) rw_wrlock(&ztest_name_lock); VERIFY(vdev_offline(spa, guid0, flags) != EBUSY); if (islog) (void) rw_unlock(&ztest_name_lock); } else { /* * Ideally we would like to be able to randomly * call vdev_[on|off]line without holding locks * to force unpredictable failures but the side * effects of vdev_[on|off]line prevent us from * doing so. We grab the ztest_vdev_lock here to * prevent a race between injection testing and * aux_vdev removal. */ mutex_enter(&ztest_vdev_lock); (void) vdev_online(spa, guid0, 0, NULL); mutex_exit(&ztest_vdev_lock); } } if (maxfaults == 0) goto out; /* * We have at least single-fault tolerance, so inject data corruption. */ fd = open(pathrand, O_RDWR); if (fd == -1) /* we hit a gap in the device namespace */ goto out; fsize = lseek(fd, 0, SEEK_END); while (--iters != 0) { /* * The offset must be chosen carefully to ensure that * we do not inject a given logical block with errors * on two different leaf devices, because ZFS can not * tolerate that (if maxfaults==1). * * We divide each leaf into chunks of size * (# leaves * SPA_MAXBLOCKSIZE * 4). Within each chunk * there is a series of ranges to which we can inject errors. * Each range can accept errors on only a single leaf vdev. * The error injection ranges are separated by ranges * which we will not inject errors on any device (DMZs). * Each DMZ must be large enough such that a single block * can not straddle it, so that a single block can not be * a target in two different injection ranges (on different * leaf vdevs). * * For example, with 3 leaves, each chunk looks like: * 0 to 32M: injection range for leaf 0 * 32M to 64M: DMZ - no injection allowed * 64M to 96M: injection range for leaf 1 * 96M to 128M: DMZ - no injection allowed * 128M to 160M: injection range for leaf 2 * 160M to 192M: DMZ - no injection allowed */ offset = ztest_random(fsize / (leaves << bshift)) * (leaves << bshift) + (leaf << bshift) + (ztest_random(1ULL << (bshift - 1)) & -8ULL); /* * Only allow damage to the labels at one end of the vdev. * * If all labels are damaged, the device will be totally * inaccessible, which will result in loss of data, * because we also damage (parts of) the other side of * the mirror/raidz. * * Additionally, we will always have both an even and an * odd label, so that we can handle crashes in the * middle of vdev_config_sync(). */ if ((leaf & 1) == 0 && offset < VDEV_LABEL_START_SIZE) continue; /* * The two end labels are stored at the "end" of the disk, but * the end of the disk (vdev_psize) is aligned to * sizeof (vdev_label_t). */ uint64_t psize = P2ALIGN(fsize, sizeof (vdev_label_t)); if ((leaf & 1) == 1 && offset + sizeof (bad) > psize - VDEV_LABEL_END_SIZE) continue; mutex_enter(&ztest_vdev_lock); if (mirror_save != zs->zs_mirrors) { mutex_exit(&ztest_vdev_lock); (void) close(fd); goto out; } if (pwrite(fd, &bad, sizeof (bad), offset) != sizeof (bad)) fatal(1, "can't inject bad word at 0x%llx in %s", offset, pathrand); mutex_exit(&ztest_vdev_lock); if (ztest_opts.zo_verbose >= 7) (void) printf("injected bad word into %s," " offset 0x%llx\n", pathrand, (u_longlong_t)offset); } (void) close(fd); out: umem_free(path0, MAXPATHLEN); umem_free(pathrand, MAXPATHLEN); } /* * Verify that DDT repair works as expected. */ void ztest_ddt_repair(ztest_ds_t *zd, uint64_t id) { ztest_shared_t *zs = ztest_shared; spa_t *spa = ztest_spa; objset_t *os = zd->zd_os; ztest_od_t *od; uint64_t object, blocksize, txg, pattern, psize; enum zio_checksum checksum = spa_dedup_checksum(spa); dmu_buf_t *db; dmu_tx_t *tx; abd_t *abd; blkptr_t blk; int copies = 2 * ZIO_DEDUPDITTO_MIN; int i; blocksize = ztest_random_blocksize(); blocksize = MIN(blocksize, 2048); /* because we write so many */ od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL); ztest_od_init(od, id, FTAG, 0, DMU_OT_UINT64_OTHER, blocksize, 0, 0); if (ztest_object_init(zd, od, sizeof (ztest_od_t), B_FALSE) != 0) { umem_free(od, sizeof (ztest_od_t)); return; } /* * Take the name lock as writer to prevent anyone else from changing * the pool and dataset properies we need to maintain during this test. */ (void) rw_wrlock(&ztest_name_lock); if (ztest_dsl_prop_set_uint64(zd->zd_name, ZFS_PROP_DEDUP, checksum, B_FALSE) != 0 || ztest_dsl_prop_set_uint64(zd->zd_name, ZFS_PROP_COPIES, 1, B_FALSE) != 0) { (void) rw_unlock(&ztest_name_lock); umem_free(od, sizeof (ztest_od_t)); return; } dmu_objset_stats_t dds; dsl_pool_config_enter(dmu_objset_pool(os), FTAG); dmu_objset_fast_stat(os, &dds); dsl_pool_config_exit(dmu_objset_pool(os), FTAG); object = od[0].od_object; blocksize = od[0].od_blocksize; pattern = zs->zs_guid ^ dds.dds_guid; ASSERT(object != 0); tx = dmu_tx_create(os); dmu_tx_hold_write(tx, object, 0, copies * blocksize); txg = ztest_tx_assign(tx, TXG_WAIT, FTAG); if (txg == 0) { (void) rw_unlock(&ztest_name_lock); umem_free(od, sizeof (ztest_od_t)); return; } /* * Write all the copies of our block. */ for (i = 0; i < copies; i++) { uint64_t offset = i * blocksize; int error = dmu_buf_hold(os, object, offset, FTAG, &db, DMU_READ_NO_PREFETCH); if (error != 0) { fatal(B_FALSE, "dmu_buf_hold(%p, %llu, %llu) = %u", os, (long long)object, (long long) offset, error); } ASSERT(db->db_offset == offset); ASSERT(db->db_size == blocksize); ASSERT(ztest_pattern_match(db->db_data, db->db_size, pattern) || ztest_pattern_match(db->db_data, db->db_size, 0ULL)); dmu_buf_will_fill(db, tx); ztest_pattern_set(db->db_data, db->db_size, pattern); dmu_buf_rele(db, FTAG); } dmu_tx_commit(tx); txg_wait_synced(spa_get_dsl(spa), txg); /* * Find out what block we got. */ VERIFY0(dmu_buf_hold(os, object, 0, FTAG, &db, DMU_READ_NO_PREFETCH)); blk = *((dmu_buf_impl_t *)db)->db_blkptr; dmu_buf_rele(db, FTAG); /* * Damage the block. Dedup-ditto will save us when we read it later. */ psize = BP_GET_PSIZE(&blk); abd = abd_alloc_linear(psize, B_TRUE); ztest_pattern_set(abd_to_buf(abd), psize, ~pattern); (void) zio_wait(zio_rewrite(NULL, spa, 0, &blk, abd, psize, NULL, NULL, ZIO_PRIORITY_SYNC_WRITE, ZIO_FLAG_CANFAIL | ZIO_FLAG_INDUCE_DAMAGE, NULL)); abd_free(abd); (void) rw_unlock(&ztest_name_lock); umem_free(od, sizeof (ztest_od_t)); } /* * Scrub the pool. */ /* ARGSUSED */ void ztest_scrub(ztest_ds_t *zd, uint64_t id) { spa_t *spa = ztest_spa; (void) spa_scan(spa, POOL_SCAN_SCRUB); (void) poll(NULL, 0, 100); /* wait a moment, then force a restart */ (void) spa_scan(spa, POOL_SCAN_SCRUB); } /* * Change the guid for the pool. */ /* ARGSUSED */ void ztest_reguid(ztest_ds_t *zd, uint64_t id) { spa_t *spa = ztest_spa; uint64_t orig, load; int error; if (ztest_opts.zo_mmp_test) return; orig = spa_guid(spa); load = spa_load_guid(spa); (void) rw_wrlock(&ztest_name_lock); error = spa_change_guid(spa); (void) rw_unlock(&ztest_name_lock); if (error != 0) return; if (ztest_opts.zo_verbose >= 4) { (void) printf("Changed guid old %llu -> %llu\n", (u_longlong_t)orig, (u_longlong_t)spa_guid(spa)); } VERIFY3U(orig, !=, spa_guid(spa)); VERIFY3U(load, ==, spa_load_guid(spa)); } /* * Rename the pool to a different name and then rename it back. */ /* ARGSUSED */ void ztest_spa_rename(ztest_ds_t *zd, uint64_t id) { char *oldname, *newname; spa_t *spa; if (ztest_opts.zo_mmp_test) return; (void) rw_wrlock(&ztest_name_lock); oldname = ztest_opts.zo_pool; newname = umem_alloc(strlen(oldname) + 5, UMEM_NOFAIL); (void) strcpy(newname, oldname); (void) strcat(newname, "_tmp"); /* * Do the rename */ VERIFY3U(0, ==, spa_rename(oldname, newname)); /* * Try to open it under the old name, which shouldn't exist */ VERIFY3U(ENOENT, ==, spa_open(oldname, &spa, FTAG)); /* * Open it under the new name and make sure it's still the same spa_t. */ VERIFY3U(0, ==, spa_open(newname, &spa, FTAG)); ASSERT(spa == ztest_spa); spa_close(spa, FTAG); /* * Rename it back to the original */ VERIFY3U(0, ==, spa_rename(newname, oldname)); /* * Make sure it can still be opened */ VERIFY3U(0, ==, spa_open(oldname, &spa, FTAG)); ASSERT(spa == ztest_spa); spa_close(spa, FTAG); umem_free(newname, strlen(newname) + 1); (void) rw_unlock(&ztest_name_lock); } void ztest_fletcher(ztest_ds_t *zd, uint64_t id) { hrtime_t end = gethrtime() + NANOSEC; while (gethrtime() <= end) { int run_count = 100; void *buf; struct abd *abd_data, *abd_meta; uint32_t size; int *ptr; int i; zio_cksum_t zc_ref; zio_cksum_t zc_ref_byteswap; size = ztest_random_blocksize(); buf = umem_alloc(size, UMEM_NOFAIL); abd_data = abd_alloc(size, B_FALSE); abd_meta = abd_alloc(size, B_TRUE); for (i = 0, ptr = buf; i < size / sizeof (*ptr); i++, ptr++) *ptr = ztest_random(UINT_MAX); abd_copy_from_buf_off(abd_data, buf, 0, size); abd_copy_from_buf_off(abd_meta, buf, 0, size); VERIFY0(fletcher_4_impl_set("scalar")); fletcher_4_native(buf, size, NULL, &zc_ref); fletcher_4_byteswap(buf, size, NULL, &zc_ref_byteswap); VERIFY0(fletcher_4_impl_set("cycle")); while (run_count-- > 0) { zio_cksum_t zc; zio_cksum_t zc_byteswap; fletcher_4_byteswap(buf, size, NULL, &zc_byteswap); fletcher_4_native(buf, size, NULL, &zc); VERIFY0(bcmp(&zc, &zc_ref, sizeof (zc))); VERIFY0(bcmp(&zc_byteswap, &zc_ref_byteswap, sizeof (zc_byteswap))); /* Test ABD - data */ abd_fletcher_4_byteswap(abd_data, size, NULL, &zc_byteswap); abd_fletcher_4_native(abd_data, size, NULL, &zc); VERIFY0(bcmp(&zc, &zc_ref, sizeof (zc))); VERIFY0(bcmp(&zc_byteswap, &zc_ref_byteswap, sizeof (zc_byteswap))); /* Test ABD - metadata */ abd_fletcher_4_byteswap(abd_meta, size, NULL, &zc_byteswap); abd_fletcher_4_native(abd_meta, size, NULL, &zc); VERIFY0(bcmp(&zc, &zc_ref, sizeof (zc))); VERIFY0(bcmp(&zc_byteswap, &zc_ref_byteswap, sizeof (zc_byteswap))); } umem_free(buf, size); abd_free(abd_data); abd_free(abd_meta); } } void ztest_fletcher_incr(ztest_ds_t *zd, uint64_t id) { void *buf; size_t size; int *ptr; int i; zio_cksum_t zc_ref; zio_cksum_t zc_ref_bswap; hrtime_t end = gethrtime() + NANOSEC; while (gethrtime() <= end) { int run_count = 100; size = ztest_random_blocksize(); buf = umem_alloc(size, UMEM_NOFAIL); for (i = 0, ptr = buf; i < size / sizeof (*ptr); i++, ptr++) *ptr = ztest_random(UINT_MAX); VERIFY0(fletcher_4_impl_set("scalar")); fletcher_4_native(buf, size, NULL, &zc_ref); fletcher_4_byteswap(buf, size, NULL, &zc_ref_bswap); VERIFY0(fletcher_4_impl_set("cycle")); while (run_count-- > 0) { zio_cksum_t zc; zio_cksum_t zc_bswap; size_t pos = 0; ZIO_SET_CHECKSUM(&zc, 0, 0, 0, 0); ZIO_SET_CHECKSUM(&zc_bswap, 0, 0, 0, 0); while (pos < size) { size_t inc = 64 * ztest_random(size / 67); /* sometimes add few bytes to test non-simd */ if (ztest_random(100) < 10) inc += P2ALIGN(ztest_random(64), sizeof (uint32_t)); if (inc > (size - pos)) inc = size - pos; fletcher_4_incremental_native(buf + pos, inc, &zc); fletcher_4_incremental_byteswap(buf + pos, inc, &zc_bswap); pos += inc; } VERIFY3U(pos, ==, size); VERIFY(ZIO_CHECKSUM_EQUAL(zc, zc_ref)); VERIFY(ZIO_CHECKSUM_EQUAL(zc_bswap, zc_ref_bswap)); /* * verify if incremental on the whole buffer is * equivalent to non-incremental version */ ZIO_SET_CHECKSUM(&zc, 0, 0, 0, 0); ZIO_SET_CHECKSUM(&zc_bswap, 0, 0, 0, 0); fletcher_4_incremental_native(buf, size, &zc); fletcher_4_incremental_byteswap(buf, size, &zc_bswap); VERIFY(ZIO_CHECKSUM_EQUAL(zc, zc_ref)); VERIFY(ZIO_CHECKSUM_EQUAL(zc_bswap, zc_ref_bswap)); } umem_free(buf, size); } } static int ztest_check_path(char *path) { struct stat s; /* return true on success */ return (!stat(path, &s)); } static void ztest_get_zdb_bin(char *bin, int len) { char *zdb_path; /* * Try to use ZDB_PATH and in-tree zdb path. If not successful, just * let popen to search through PATH. */ if ((zdb_path = getenv("ZDB_PATH"))) { strlcpy(bin, zdb_path, len); /* In env */ if (!ztest_check_path(bin)) { ztest_dump_core = 0; fatal(1, "invalid ZDB_PATH '%s'", bin); } return; } VERIFY(realpath(getexecname(), bin) != NULL); if (strstr(bin, "/ztest/")) { strstr(bin, "/ztest/")[0] = '\0'; /* In-tree */ strcat(bin, "/zdb/zdb"); if (ztest_check_path(bin)) return; } strcpy(bin, "zdb"); } /* * Verify pool integrity by running zdb. */ static void ztest_run_zdb(char *pool) { int status; char *bin; char *zdb; char *zbuf; const int len = MAXPATHLEN + MAXNAMELEN + 20; FILE *fp; bin = umem_alloc(len, UMEM_NOFAIL); zdb = umem_alloc(len, UMEM_NOFAIL); zbuf = umem_alloc(1024, UMEM_NOFAIL); ztest_get_zdb_bin(bin, len); (void) sprintf(zdb, "%s -bcc%s%s -G -d -U %s %s", bin, ztest_opts.zo_verbose >= 3 ? "s" : "", ztest_opts.zo_verbose >= 4 ? "v" : "", spa_config_path, pool); if (ztest_opts.zo_verbose >= 5) (void) printf("Executing %s\n", strstr(zdb, "zdb ")); fp = popen(zdb, "r"); while (fgets(zbuf, 1024, fp) != NULL) if (ztest_opts.zo_verbose >= 3) (void) printf("%s", zbuf); status = pclose(fp); if (status == 0) goto out; ztest_dump_core = 0; if (WIFEXITED(status)) fatal(0, "'%s' exit code %d", zdb, WEXITSTATUS(status)); else fatal(0, "'%s' died with signal %d", zdb, WTERMSIG(status)); out: umem_free(bin, len); umem_free(zdb, len); umem_free(zbuf, 1024); } static void ztest_walk_pool_directory(char *header) { spa_t *spa = NULL; if (ztest_opts.zo_verbose >= 6) (void) printf("%s\n", header); mutex_enter(&spa_namespace_lock); while ((spa = spa_next(spa)) != NULL) if (ztest_opts.zo_verbose >= 6) (void) printf("\t%s\n", spa_name(spa)); mutex_exit(&spa_namespace_lock); } static void ztest_spa_import_export(char *oldname, char *newname) { nvlist_t *config, *newconfig; uint64_t pool_guid; spa_t *spa; int error; if (ztest_opts.zo_verbose >= 4) { (void) printf("import/export: old = %s, new = %s\n", oldname, newname); } /* * Clean up from previous runs. */ (void) spa_destroy(newname); /* * Get the pool's configuration and guid. */ VERIFY3U(0, ==, spa_open(oldname, &spa, FTAG)); /* * Kick off a scrub to tickle scrub/export races. */ if (ztest_random(2) == 0) (void) spa_scan(spa, POOL_SCAN_SCRUB); pool_guid = spa_guid(spa); spa_close(spa, FTAG); ztest_walk_pool_directory("pools before export"); /* * Export it. */ VERIFY3U(0, ==, spa_export(oldname, &config, B_FALSE, B_FALSE)); ztest_walk_pool_directory("pools after export"); /* * Try to import it. */ newconfig = spa_tryimport(config); ASSERT(newconfig != NULL); nvlist_free(newconfig); /* * Import it under the new name. */ error = spa_import(newname, config, NULL, 0); if (error != 0) { dump_nvlist(config, 0); fatal(B_FALSE, "couldn't import pool %s as %s: error %u", oldname, newname, error); } ztest_walk_pool_directory("pools after import"); /* * Try to import it again -- should fail with EEXIST. */ VERIFY3U(EEXIST, ==, spa_import(newname, config, NULL, 0)); /* * Try to import it under a different name -- should fail with EEXIST. */ VERIFY3U(EEXIST, ==, spa_import(oldname, config, NULL, 0)); /* * Verify that the pool is no longer visible under the old name. */ VERIFY3U(ENOENT, ==, spa_open(oldname, &spa, FTAG)); /* * Verify that we can open and close the pool using the new name. */ VERIFY3U(0, ==, spa_open(newname, &spa, FTAG)); ASSERT(pool_guid == spa_guid(spa)); spa_close(spa, FTAG); nvlist_free(config); } static void ztest_resume(spa_t *spa) { if (spa_suspended(spa) && ztest_opts.zo_verbose >= 6) (void) printf("resuming from suspended state\n"); spa_vdev_state_enter(spa, SCL_NONE); vdev_clear(spa, NULL); (void) spa_vdev_state_exit(spa, NULL, 0); (void) zio_resume(spa); } static void ztest_resume_thread(void *arg) { spa_t *spa = arg; while (!ztest_exiting) { if (spa_suspended(spa)) ztest_resume(spa); (void) poll(NULL, 0, 100); /* * Periodically change the zfs_compressed_arc_enabled setting. */ if (ztest_random(10) == 0) zfs_compressed_arc_enabled = ztest_random(2); /* * Periodically change the zfs_abd_scatter_enabled setting. */ if (ztest_random(10) == 0) zfs_abd_scatter_enabled = ztest_random(2); } thread_exit(); } #define GRACE 300 #if 0 static void ztest_deadman_alarm(int sig) { fatal(0, "failed to complete within %d seconds of deadline", GRACE); } #endif static void ztest_execute(int test, ztest_info_t *zi, uint64_t id) { ztest_ds_t *zd = &ztest_ds[id % ztest_opts.zo_datasets]; ztest_shared_callstate_t *zc = ZTEST_GET_SHARED_CALLSTATE(test); hrtime_t functime = gethrtime(); int i; for (i = 0; i < zi->zi_iters; i++) zi->zi_func(zd, id); functime = gethrtime() - functime; atomic_add_64(&zc->zc_count, 1); atomic_add_64(&zc->zc_time, functime); if (ztest_opts.zo_verbose >= 4) (void) printf("%6.2f sec in %s\n", (double)functime / NANOSEC, zi->zi_funcname); } static void ztest_thread(void *arg) { int rand; uint64_t id = (uintptr_t)arg; ztest_shared_t *zs = ztest_shared; uint64_t call_next; hrtime_t now; ztest_info_t *zi; ztest_shared_callstate_t *zc; while ((now = gethrtime()) < zs->zs_thread_stop) { /* * See if it's time to force a crash. */ if (now > zs->zs_thread_kill) ztest_kill(zs); /* * If we're getting ENOSPC with some regularity, stop. */ if (zs->zs_enospc_count > 10) break; /* * Pick a random function to execute. */ rand = ztest_random(ZTEST_FUNCS); zi = &ztest_info[rand]; zc = ZTEST_GET_SHARED_CALLSTATE(rand); call_next = zc->zc_next; if (now >= call_next && atomic_cas_64(&zc->zc_next, call_next, call_next + ztest_random(2 * zi->zi_interval[0] + 1)) == call_next) { ztest_execute(rand, zi, id); } } thread_exit(); } static void ztest_dataset_name(char *dsname, char *pool, int d) { (void) snprintf(dsname, ZFS_MAX_DATASET_NAME_LEN, "%s/ds_%d", pool, d); } static void ztest_dataset_destroy(int d) { char name[ZFS_MAX_DATASET_NAME_LEN]; int t; ztest_dataset_name(name, ztest_opts.zo_pool, d); if (ztest_opts.zo_verbose >= 3) (void) printf("Destroying %s to free up space\n", name); /* * Cleanup any non-standard clones and snapshots. In general, * ztest thread t operates on dataset (t % zopt_datasets), * so there may be more than one thing to clean up. */ for (t = d; t < ztest_opts.zo_threads; t += ztest_opts.zo_datasets) ztest_dsl_dataset_cleanup(name, t); (void) dmu_objset_find(name, ztest_objset_destroy_cb, NULL, DS_FIND_SNAPSHOTS | DS_FIND_CHILDREN); } static void ztest_dataset_dirobj_verify(ztest_ds_t *zd) { uint64_t usedobjs, dirobjs, scratch; /* * ZTEST_DIROBJ is the object directory for the entire dataset. * Therefore, the number of objects in use should equal the * number of ZTEST_DIROBJ entries, +1 for ZTEST_DIROBJ itself. * If not, we have an object leak. * * Note that we can only check this in ztest_dataset_open(), * when the open-context and syncing-context values agree. * That's because zap_count() returns the open-context value, * while dmu_objset_space() returns the rootbp fill count. */ VERIFY3U(0, ==, zap_count(zd->zd_os, ZTEST_DIROBJ, &dirobjs)); dmu_objset_space(zd->zd_os, &scratch, &scratch, &usedobjs, &scratch); ASSERT3U(dirobjs + 1, ==, usedobjs); } static int ztest_dataset_open(int d) { ztest_ds_t *zd = &ztest_ds[d]; uint64_t committed_seq = ZTEST_GET_SHARED_DS(d)->zd_seq; objset_t *os; zilog_t *zilog; char name[ZFS_MAX_DATASET_NAME_LEN]; int error; ztest_dataset_name(name, ztest_opts.zo_pool, d); (void) rw_rdlock(&ztest_name_lock); error = ztest_dataset_create(name); if (error == ENOSPC) { (void) rw_unlock(&ztest_name_lock); ztest_record_enospc(FTAG); return (error); } ASSERT(error == 0 || error == EEXIST); VERIFY0(ztest_dmu_objset_own(name, DMU_OST_OTHER, B_FALSE, B_TRUE, zd, &os)); (void) rw_unlock(&ztest_name_lock); ztest_zd_init(zd, ZTEST_GET_SHARED_DS(d), os); zilog = zd->zd_zilog; if (zilog->zl_header->zh_claim_lr_seq != 0 && zilog->zl_header->zh_claim_lr_seq < committed_seq) fatal(0, "missing log records: claimed %llu < committed %llu", zilog->zl_header->zh_claim_lr_seq, committed_seq); ztest_dataset_dirobj_verify(zd); zil_replay(os, zd, ztest_replay_vector); ztest_dataset_dirobj_verify(zd); if (ztest_opts.zo_verbose >= 6) (void) printf("%s replay %llu blocks, %llu records, seq %llu\n", zd->zd_name, (u_longlong_t)zilog->zl_parse_blk_count, (u_longlong_t)zilog->zl_parse_lr_count, (u_longlong_t)zilog->zl_replaying_seq); zilog = zil_open(os, ztest_get_data); if (zilog->zl_replaying_seq != 0 && zilog->zl_replaying_seq < committed_seq) fatal(0, "missing log records: replayed %llu < committed %llu", zilog->zl_replaying_seq, committed_seq); return (0); } static void ztest_dataset_close(int d) { ztest_ds_t *zd = &ztest_ds[d]; zil_close(zd->zd_zilog); txg_wait_synced(spa_get_dsl(zd->zd_os->os_spa), 0); dmu_objset_disown(zd->zd_os, B_TRUE, zd); ztest_zd_fini(zd); } /* * Kick off threads to run tests on all datasets in parallel. */ static void ztest_run(ztest_shared_t *zs) { spa_t *spa; objset_t *os; kthread_t *resume_thread; kthread_t **run_threads; uint64_t object; int error; int t, d; ztest_exiting = B_FALSE; /* * Initialize parent/child shared state. */ mutex_init(&ztest_vdev_lock, NULL, MUTEX_DEFAULT, NULL); VERIFY(rwlock_init(&ztest_name_lock, USYNC_THREAD, NULL) == 0); zs->zs_thread_start = gethrtime(); zs->zs_thread_stop = zs->zs_thread_start + ztest_opts.zo_passtime * NANOSEC; zs->zs_thread_stop = MIN(zs->zs_thread_stop, zs->zs_proc_stop); zs->zs_thread_kill = zs->zs_thread_stop; if (ztest_random(100) < ztest_opts.zo_killrate) { zs->zs_thread_kill -= ztest_random(ztest_opts.zo_passtime * NANOSEC); } mutex_init(&zcl.zcl_callbacks_lock, NULL, MUTEX_DEFAULT, NULL); list_create(&zcl.zcl_callbacks, sizeof (ztest_cb_data_t), offsetof(ztest_cb_data_t, zcd_node)); /* * Open our pool. */ kernel_init(FREAD | FWRITE); VERIFY0(spa_open(ztest_opts.zo_pool, &spa, FTAG)); spa->spa_debug = B_TRUE; metaslab_preload_limit = ztest_random(20) + 1; ztest_spa = spa; dmu_objset_stats_t dds; VERIFY0(ztest_dmu_objset_own(ztest_opts.zo_pool, DMU_OST_ANY, B_TRUE, B_TRUE, FTAG, &os)); dsl_pool_config_enter(dmu_objset_pool(os), FTAG); dmu_objset_fast_stat(os, &dds); dsl_pool_config_exit(dmu_objset_pool(os), FTAG); zs->zs_guid = dds.dds_guid; dmu_objset_disown(os, B_TRUE, FTAG); spa->spa_dedup_ditto = 2 * ZIO_DEDUPDITTO_MIN; /* * We don't expect the pool to suspend unless maxfaults == 0, * in which case ztest_fault_inject() temporarily takes away * the only valid replica. */ if (MAXFAULTS() == 0) spa->spa_failmode = ZIO_FAILURE_MODE_WAIT; else spa->spa_failmode = ZIO_FAILURE_MODE_PANIC; /* * Create a thread to periodically resume suspended I/O. */ resume_thread = thread_create(NULL, 0, ztest_resume_thread, spa, 0, NULL, TS_RUN | TS_JOINABLE, defclsyspri); #if 0 /* * Set a deadman alarm to abort() if we hang. */ signal(SIGALRM, ztest_deadman_alarm); alarm((zs->zs_thread_stop - zs->zs_thread_start) / NANOSEC + GRACE); #endif /* * Verify that we can safely inquire about about any object, * whether it's allocated or not. To make it interesting, * we probe a 5-wide window around each power of two. * This hits all edge cases, including zero and the max. */ for (t = 0; t < 64; t++) { for (d = -5; d <= 5; d++) { error = dmu_object_info(spa->spa_meta_objset, (1ULL << t) + d, NULL); ASSERT(error == 0 || error == ENOENT || error == EINVAL); } } /* * If we got any ENOSPC errors on the previous run, destroy something. */ if (zs->zs_enospc_count != 0) { int d = ztest_random(ztest_opts.zo_datasets); ztest_dataset_destroy(d); } zs->zs_enospc_count = 0; run_threads = umem_zalloc(ztest_opts.zo_threads * sizeof (kthread_t *), UMEM_NOFAIL); if (ztest_opts.zo_verbose >= 4) (void) printf("starting main threads...\n"); /* * Kick off all the tests that run in parallel. */ for (t = 0; t < ztest_opts.zo_threads; t++) { if (t < ztest_opts.zo_datasets && ztest_dataset_open(t) != 0) { umem_free(run_threads, ztest_opts.zo_threads * sizeof (kthread_t *)); return; } run_threads[t] = thread_create(NULL, 0, ztest_thread, (void *)(uintptr_t)t, 0, NULL, TS_RUN | TS_JOINABLE, defclsyspri); } /* * Wait for all of the tests to complete. We go in reverse order * so we don't close datasets while threads are still using them. */ for (t = ztest_opts.zo_threads - 1; t >= 0; t--) { VERIFY0(thread_join(run_threads[t])); if (t < ztest_opts.zo_datasets) ztest_dataset_close(t); } txg_wait_synced(spa_get_dsl(spa), 0); zs->zs_alloc = metaslab_class_get_alloc(spa_normal_class(spa)); zs->zs_space = metaslab_class_get_space(spa_normal_class(spa)); umem_free(run_threads, ztest_opts.zo_threads * sizeof (kthread_t *)); /* Kill the resume thread */ ztest_exiting = B_TRUE; VERIFY0(thread_join(resume_thread)); ztest_resume(spa); /* * Right before closing the pool, kick off a bunch of async I/O; * spa_close() should wait for it to complete. */ for (object = 1; object < 50; object++) { dmu_prefetch(spa->spa_meta_objset, object, 0, 0, 1ULL << 20, ZIO_PRIORITY_SYNC_READ); } /* Verify that at least one commit cb was called in a timely fashion */ if (zc_cb_counter >= ZTEST_COMMIT_CB_MIN_REG) VERIFY0(zc_min_txg_delay); spa_close(spa, FTAG); /* * Verify that we can loop over all pools. */ mutex_enter(&spa_namespace_lock); for (spa = spa_next(NULL); spa != NULL; spa = spa_next(spa)) if (ztest_opts.zo_verbose > 3) (void) printf("spa_next: found %s\n", spa_name(spa)); mutex_exit(&spa_namespace_lock); /* * Verify that we can export the pool and reimport it under a * different name. */ if ((ztest_random(2) == 0) && !ztest_opts.zo_mmp_test) { char name[ZFS_MAX_DATASET_NAME_LEN]; (void) snprintf(name, sizeof (name), "%s_import", ztest_opts.zo_pool); ztest_spa_import_export(ztest_opts.zo_pool, name); ztest_spa_import_export(name, ztest_opts.zo_pool); } kernel_fini(); list_destroy(&zcl.zcl_callbacks); mutex_destroy(&zcl.zcl_callbacks_lock); (void) rwlock_destroy(&ztest_name_lock); mutex_destroy(&ztest_vdev_lock); } static void ztest_freeze(void) { ztest_ds_t *zd = &ztest_ds[0]; spa_t *spa; int numloops = 0; if (ztest_opts.zo_verbose >= 3) (void) printf("testing spa_freeze()...\n"); kernel_init(FREAD | FWRITE); VERIFY3U(0, ==, spa_open(ztest_opts.zo_pool, &spa, FTAG)); VERIFY3U(0, ==, ztest_dataset_open(0)); spa->spa_debug = B_TRUE; ztest_spa = spa; /* * Force the first log block to be transactionally allocated. * We have to do this before we freeze the pool -- otherwise * the log chain won't be anchored. */ while (BP_IS_HOLE(&zd->zd_zilog->zl_header->zh_log)) { ztest_dmu_object_alloc_free(zd, 0); zil_commit(zd->zd_zilog, 0); } txg_wait_synced(spa_get_dsl(spa), 0); /* * Freeze the pool. This stops spa_sync() from doing anything, * so that the only way to record changes from now on is the ZIL. */ spa_freeze(spa); /* * Because it is hard to predict how much space a write will actually * require beforehand, we leave ourselves some fudge space to write over * capacity. */ uint64_t capacity = metaslab_class_get_space(spa_normal_class(spa)) / 2; /* * Run tests that generate log records but don't alter the pool config * or depend on DSL sync tasks (snapshots, objset create/destroy, etc). * We do a txg_wait_synced() after each iteration to force the txg * to increase well beyond the last synced value in the uberblock. * The ZIL should be OK with that. * * Run a random number of times less than zo_maxloops and ensure we do * not run out of space on the pool. */ while (ztest_random(10) != 0 && numloops++ < ztest_opts.zo_maxloops && metaslab_class_get_alloc(spa_normal_class(spa)) < capacity) { ztest_od_t od; ztest_od_init(&od, 0, FTAG, 0, DMU_OT_UINT64_OTHER, 0, 0, 0); VERIFY0(ztest_object_init(zd, &od, sizeof (od), B_FALSE)); ztest_io(zd, od.od_object, ztest_random(ZTEST_RANGE_LOCKS) << SPA_MAXBLOCKSHIFT); txg_wait_synced(spa_get_dsl(spa), 0); } /* * Commit all of the changes we just generated. */ zil_commit(zd->zd_zilog, 0); txg_wait_synced(spa_get_dsl(spa), 0); /* * Close our dataset and close the pool. */ ztest_dataset_close(0); spa_close(spa, FTAG); kernel_fini(); /* * Open and close the pool and dataset to induce log replay. */ kernel_init(FREAD | FWRITE); VERIFY3U(0, ==, spa_open(ztest_opts.zo_pool, &spa, FTAG)); ASSERT(spa_freeze_txg(spa) == UINT64_MAX); VERIFY3U(0, ==, ztest_dataset_open(0)); spa->spa_debug = B_TRUE; ztest_spa = spa; txg_wait_synced(spa_get_dsl(spa), 0); ztest_dataset_close(0); ztest_reguid(NULL, 0); spa_close(spa, FTAG); kernel_fini(); } void print_time(hrtime_t t, char *timebuf) { hrtime_t s = t / NANOSEC; hrtime_t m = s / 60; hrtime_t h = m / 60; hrtime_t d = h / 24; s -= m * 60; m -= h * 60; h -= d * 24; timebuf[0] = '\0'; if (d) (void) sprintf(timebuf, "%llud%02lluh%02llum%02llus", d, h, m, s); else if (h) (void) sprintf(timebuf, "%lluh%02llum%02llus", h, m, s); else if (m) (void) sprintf(timebuf, "%llum%02llus", m, s); else (void) sprintf(timebuf, "%llus", s); } static nvlist_t * make_random_props(void) { nvlist_t *props; VERIFY(nvlist_alloc(&props, NV_UNIQUE_NAME, 0) == 0); if (ztest_random(2) == 0) return (props); VERIFY(nvlist_add_uint64(props, "autoreplace", 1) == 0); return (props); } /* * Import a storage pool with the given name. */ static void ztest_import(ztest_shared_t *zs) { libzfs_handle_t *hdl; importargs_t args = { 0 }; spa_t *spa; nvlist_t *cfg = NULL; int nsearch = 1; char *searchdirs[nsearch]; char *name = ztest_opts.zo_pool; int flags = ZFS_IMPORT_MISSING_LOG; int error; mutex_init(&ztest_vdev_lock, NULL, MUTEX_DEFAULT, NULL); VERIFY(rwlock_init(&ztest_name_lock, USYNC_THREAD, NULL) == 0); kernel_init(FREAD | FWRITE); hdl = libzfs_init(); searchdirs[0] = ztest_opts.zo_dir; args.paths = nsearch; args.path = searchdirs; args.can_be_active = B_FALSE; error = zpool_tryimport(hdl, name, &cfg, &args); if (error) (void) fatal(0, "No pools found\n"); VERIFY0(spa_import(name, cfg, NULL, flags)); VERIFY0(spa_open(name, &spa, FTAG)); zs->zs_metaslab_sz = 1ULL << spa->spa_root_vdev->vdev_child[0]->vdev_ms_shift; spa_close(spa, FTAG); libzfs_fini(hdl); kernel_fini(); if (!ztest_opts.zo_mmp_test) { ztest_run_zdb(ztest_opts.zo_pool); ztest_freeze(); ztest_run_zdb(ztest_opts.zo_pool); } (void) rwlock_destroy(&ztest_name_lock); mutex_destroy(&ztest_vdev_lock); } /* * Create a storage pool with the given name and initial vdev size. * Then test spa_freeze() functionality. */ static void ztest_init(ztest_shared_t *zs) { spa_t *spa; nvlist_t *nvroot, *props; int i; mutex_init(&ztest_vdev_lock, NULL, MUTEX_DEFAULT, NULL); VERIFY(rwlock_init(&ztest_name_lock, USYNC_THREAD, NULL) == 0); kernel_init(FREAD | FWRITE); /* * Create the storage pool. */ (void) spa_destroy(ztest_opts.zo_pool); ztest_shared->zs_vdev_next_leaf = 0; zs->zs_splits = 0; zs->zs_mirrors = ztest_opts.zo_mirrors; nvroot = make_vdev_root(NULL, NULL, NULL, ztest_opts.zo_vdev_size, 0, 0, ztest_opts.zo_raidz, zs->zs_mirrors, 1); props = make_random_props(); for (i = 0; i < SPA_FEATURES; i++) { char *buf; VERIFY3S(-1, !=, asprintf(&buf, "feature@%s", spa_feature_table[i].fi_uname)); VERIFY3U(0, ==, nvlist_add_uint64(props, buf, 0)); free(buf); } VERIFY3U(0, ==, spa_create(ztest_opts.zo_pool, nvroot, props, NULL, NULL)); nvlist_free(nvroot); nvlist_free(props); VERIFY3U(0, ==, spa_open(ztest_opts.zo_pool, &spa, FTAG)); zs->zs_metaslab_sz = 1ULL << spa->spa_root_vdev->vdev_child[0]->vdev_ms_shift; spa_close(spa, FTAG); kernel_fini(); if (!ztest_opts.zo_mmp_test) { ztest_run_zdb(ztest_opts.zo_pool); ztest_freeze(); ztest_run_zdb(ztest_opts.zo_pool); } (void) rwlock_destroy(&ztest_name_lock); mutex_destroy(&ztest_vdev_lock); } static void setup_data_fd(void) { static char ztest_name_data[] = "/tmp/ztest.data.XXXXXX"; ztest_fd_data = mkstemp(ztest_name_data); ASSERT3S(ztest_fd_data, >=, 0); (void) unlink(ztest_name_data); } static int shared_data_size(ztest_shared_hdr_t *hdr) { int size; size = hdr->zh_hdr_size; size += hdr->zh_opts_size; size += hdr->zh_size; size += hdr->zh_stats_size * hdr->zh_stats_count; size += hdr->zh_ds_size * hdr->zh_ds_count; return (size); } static void setup_hdr(void) { int size; ztest_shared_hdr_t *hdr; hdr = (void *)mmap(0, P2ROUNDUP(sizeof (*hdr), getpagesize()), PROT_READ | PROT_WRITE, MAP_SHARED, ztest_fd_data, 0); ASSERT(hdr != MAP_FAILED); VERIFY3U(0, ==, ftruncate(ztest_fd_data, sizeof (ztest_shared_hdr_t))); hdr->zh_hdr_size = sizeof (ztest_shared_hdr_t); hdr->zh_opts_size = sizeof (ztest_shared_opts_t); hdr->zh_size = sizeof (ztest_shared_t); hdr->zh_stats_size = sizeof (ztest_shared_callstate_t); hdr->zh_stats_count = ZTEST_FUNCS; hdr->zh_ds_size = sizeof (ztest_shared_ds_t); hdr->zh_ds_count = ztest_opts.zo_datasets; size = shared_data_size(hdr); VERIFY3U(0, ==, ftruncate(ztest_fd_data, size)); (void) munmap((caddr_t)hdr, P2ROUNDUP(sizeof (*hdr), getpagesize())); } static void setup_data(void) { int size, offset; ztest_shared_hdr_t *hdr; uint8_t *buf; hdr = (void *)mmap(0, P2ROUNDUP(sizeof (*hdr), getpagesize()), PROT_READ, MAP_SHARED, ztest_fd_data, 0); ASSERT(hdr != MAP_FAILED); size = shared_data_size(hdr); (void) munmap((caddr_t)hdr, P2ROUNDUP(sizeof (*hdr), getpagesize())); hdr = ztest_shared_hdr = (void *)mmap(0, P2ROUNDUP(size, getpagesize()), PROT_READ | PROT_WRITE, MAP_SHARED, ztest_fd_data, 0); ASSERT(hdr != MAP_FAILED); buf = (uint8_t *)hdr; offset = hdr->zh_hdr_size; ztest_shared_opts = (void *)&buf[offset]; offset += hdr->zh_opts_size; ztest_shared = (void *)&buf[offset]; offset += hdr->zh_size; ztest_shared_callstate = (void *)&buf[offset]; offset += hdr->zh_stats_size * hdr->zh_stats_count; ztest_shared_ds = (void *)&buf[offset]; } static boolean_t exec_child(char *cmd, char *libpath, boolean_t ignorekill, int *statusp) { pid_t pid; int status; char *cmdbuf = NULL; pid = fork(); if (cmd == NULL) { cmdbuf = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); (void) strlcpy(cmdbuf, getexecname(), MAXPATHLEN); cmd = cmdbuf; } if (pid == -1) fatal(1, "fork failed"); if (pid == 0) { /* child */ char *emptyargv[2] = { cmd, NULL }; char fd_data_str[12]; struct rlimit rl = { 1024, 1024 }; (void) setrlimit(RLIMIT_NOFILE, &rl); (void) close(ztest_fd_rand); VERIFY(11 >= snprintf(fd_data_str, 12, "%d", ztest_fd_data)); VERIFY(0 == setenv("ZTEST_FD_DATA", fd_data_str, 1)); (void) enable_extended_FILE_stdio(-1, -1); if (libpath != NULL) VERIFY(0 == setenv("LD_LIBRARY_PATH", libpath, 1)); (void) execv(cmd, emptyargv); ztest_dump_core = B_FALSE; fatal(B_TRUE, "exec failed: %s", cmd); } if (cmdbuf != NULL) { umem_free(cmdbuf, MAXPATHLEN); cmd = NULL; } while (waitpid(pid, &status, 0) != pid) continue; if (statusp != NULL) *statusp = status; if (WIFEXITED(status)) { if (WEXITSTATUS(status) != 0) { (void) fprintf(stderr, "child exited with code %d\n", WEXITSTATUS(status)); exit(2); } return (B_FALSE); } else if (WIFSIGNALED(status)) { if (!ignorekill || WTERMSIG(status) != SIGKILL) { (void) fprintf(stderr, "child died with signal %d\n", WTERMSIG(status)); exit(3); } return (B_TRUE); } else { (void) fprintf(stderr, "something strange happened to child\n"); exit(4); /* NOTREACHED */ } } static void ztest_run_init(void) { int i; ztest_shared_t *zs = ztest_shared; /* * Blow away any existing copy of zpool.cache */ (void) remove(spa_config_path); if (ztest_opts.zo_init == 0) { if (ztest_opts.zo_verbose >= 1) (void) printf("Importing pool %s\n", ztest_opts.zo_pool); ztest_import(zs); return; } /* * Create and initialize our storage pool. */ for (i = 1; i <= ztest_opts.zo_init; i++) { bzero(zs, sizeof (ztest_shared_t)); if (ztest_opts.zo_verbose >= 3 && ztest_opts.zo_init != 1) { (void) printf("ztest_init(), pass %d\n", i); } ztest_init(zs); } } int main(int argc, char **argv) { int kills = 0; int iters = 0; int older = 0; int newer = 0; ztest_shared_t *zs; ztest_info_t *zi; ztest_shared_callstate_t *zc; char timebuf[100]; char numbuf[NN_NUMBUF_SZ]; spa_t *spa; char *cmd; boolean_t hasalt; int f; char *fd_data_str = getenv("ZTEST_FD_DATA"); struct sigaction action; (void) setvbuf(stdout, NULL, _IOLBF, 0); dprintf_setup(&argc, argv); action.sa_handler = sig_handler; sigemptyset(&action.sa_mask); action.sa_flags = 0; if (sigaction(SIGSEGV, &action, NULL) < 0) { (void) fprintf(stderr, "ztest: cannot catch SIGSEGV: %s.\n", strerror(errno)); exit(EXIT_FAILURE); } if (sigaction(SIGABRT, &action, NULL) < 0) { (void) fprintf(stderr, "ztest: cannot catch SIGABRT: %s.\n", strerror(errno)); exit(EXIT_FAILURE); } ztest_fd_rand = open("/dev/urandom", O_RDONLY); ASSERT3S(ztest_fd_rand, >=, 0); if (!fd_data_str) { process_options(argc, argv); setup_data_fd(); setup_hdr(); setup_data(); bcopy(&ztest_opts, ztest_shared_opts, sizeof (*ztest_shared_opts)); } else { ztest_fd_data = atoi(fd_data_str); setup_data(); bcopy(ztest_shared_opts, &ztest_opts, sizeof (ztest_opts)); } ASSERT3U(ztest_opts.zo_datasets, ==, ztest_shared_hdr->zh_ds_count); /* Override location of zpool.cache */ VERIFY(asprintf((char **)&spa_config_path, "%s/zpool.cache", ztest_opts.zo_dir) != -1); ztest_ds = umem_alloc(ztest_opts.zo_datasets * sizeof (ztest_ds_t), UMEM_NOFAIL); zs = ztest_shared; if (fd_data_str) { metaslab_gang_bang = ztest_opts.zo_metaslab_gang_bang; metaslab_df_alloc_threshold = zs->zs_metaslab_df_alloc_threshold; if (zs->zs_do_init) ztest_run_init(); else ztest_run(zs); exit(0); } hasalt = (strlen(ztest_opts.zo_alt_ztest) != 0); if (ztest_opts.zo_verbose >= 1) { (void) printf("%llu vdevs, %d datasets, %d threads," " %llu seconds...\n", (u_longlong_t)ztest_opts.zo_vdevs, ztest_opts.zo_datasets, ztest_opts.zo_threads, (u_longlong_t)ztest_opts.zo_time); } cmd = umem_alloc(MAXNAMELEN, UMEM_NOFAIL); (void) strlcpy(cmd, getexecname(), MAXNAMELEN); zs->zs_do_init = B_TRUE; if (strlen(ztest_opts.zo_alt_ztest) != 0) { if (ztest_opts.zo_verbose >= 1) { (void) printf("Executing older ztest for " "initialization: %s\n", ztest_opts.zo_alt_ztest); } VERIFY(!exec_child(ztest_opts.zo_alt_ztest, ztest_opts.zo_alt_libpath, B_FALSE, NULL)); } else { VERIFY(!exec_child(NULL, NULL, B_FALSE, NULL)); } zs->zs_do_init = B_FALSE; zs->zs_proc_start = gethrtime(); zs->zs_proc_stop = zs->zs_proc_start + ztest_opts.zo_time * NANOSEC; for (f = 0; f < ZTEST_FUNCS; f++) { zi = &ztest_info[f]; zc = ZTEST_GET_SHARED_CALLSTATE(f); if (zs->zs_proc_start + zi->zi_interval[0] > zs->zs_proc_stop) zc->zc_next = UINT64_MAX; else zc->zc_next = zs->zs_proc_start + ztest_random(2 * zi->zi_interval[0] + 1); } /* * Run the tests in a loop. These tests include fault injection * to verify that self-healing data works, and forced crashes * to verify that we never lose on-disk consistency. */ while (gethrtime() < zs->zs_proc_stop) { int status; boolean_t killed; /* * Initialize the workload counters for each function. */ for (f = 0; f < ZTEST_FUNCS; f++) { zc = ZTEST_GET_SHARED_CALLSTATE(f); zc->zc_count = 0; zc->zc_time = 0; } /* Set the allocation switch size */ zs->zs_metaslab_df_alloc_threshold = ztest_random(zs->zs_metaslab_sz / 4) + 1; if (!hasalt || ztest_random(2) == 0) { if (hasalt && ztest_opts.zo_verbose >= 1) { (void) printf("Executing newer ztest: %s\n", cmd); } newer++; killed = exec_child(cmd, NULL, B_TRUE, &status); } else { if (hasalt && ztest_opts.zo_verbose >= 1) { (void) printf("Executing older ztest: %s\n", ztest_opts.zo_alt_ztest); } older++; killed = exec_child(ztest_opts.zo_alt_ztest, ztest_opts.zo_alt_libpath, B_TRUE, &status); } if (killed) kills++; iters++; if (ztest_opts.zo_verbose >= 1) { hrtime_t now = gethrtime(); now = MIN(now, zs->zs_proc_stop); print_time(zs->zs_proc_stop - now, timebuf); nicenum(zs->zs_space, numbuf, sizeof (numbuf)); (void) printf("Pass %3d, %8s, %3llu ENOSPC, " "%4.1f%% of %5s used, %3.0f%% done, %8s to go\n", iters, WIFEXITED(status) ? "Complete" : "SIGKILL", (u_longlong_t)zs->zs_enospc_count, 100.0 * zs->zs_alloc / zs->zs_space, numbuf, 100.0 * (now - zs->zs_proc_start) / (ztest_opts.zo_time * NANOSEC), timebuf); } if (ztest_opts.zo_verbose >= 2) { (void) printf("\nWorkload summary:\n\n"); (void) printf("%7s %9s %s\n", "Calls", "Time", "Function"); (void) printf("%7s %9s %s\n", "-----", "----", "--------"); for (f = 0; f < ZTEST_FUNCS; f++) { zi = &ztest_info[f]; zc = ZTEST_GET_SHARED_CALLSTATE(f); print_time(zc->zc_time, timebuf); (void) printf("%7llu %9s %s\n", (u_longlong_t)zc->zc_count, timebuf, zi->zi_funcname); } (void) printf("\n"); } /* * It's possible that we killed a child during a rename test, * in which case we'll have a 'ztest_tmp' pool lying around * instead of 'ztest'. Do a blind rename in case this happened. */ kernel_init(FREAD); if (spa_open(ztest_opts.zo_pool, &spa, FTAG) == 0) { spa_close(spa, FTAG); } else { char tmpname[ZFS_MAX_DATASET_NAME_LEN]; kernel_fini(); kernel_init(FREAD | FWRITE); (void) snprintf(tmpname, sizeof (tmpname), "%s_tmp", ztest_opts.zo_pool); (void) spa_rename(tmpname, ztest_opts.zo_pool); } kernel_fini(); if (!ztest_opts.zo_mmp_test) ztest_run_zdb(ztest_opts.zo_pool); } if (ztest_opts.zo_verbose >= 1) { if (hasalt) { (void) printf("%d runs of older ztest: %s\n", older, ztest_opts.zo_alt_ztest); (void) printf("%d runs of newer ztest: %s\n", newer, cmd); } (void) printf("%d killed, %d completed, %.0f%% kill rate\n", kills, iters - kills, (100.0 * kills) / MAX(1, iters)); } umem_free(cmd, MAXNAMELEN); return (0); } diff --git a/include/sys/dmu.h b/include/sys/dmu.h index 8a92919003ed..f252a9cb63c0 100644 --- a/include/sys/dmu.h +++ b/include/sys/dmu.h @@ -1,1026 +1,1026 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2017 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. */ /* 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; 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). */ #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(ot) (((ot) & DMU_OT_NEWTYPE) ? \ ((ot) & DMU_OT_METADATA) : \ dmu_ot[(int)(ot)].ot_metadata) #define DMU_OT_IS_ENCRYPTED(ot) (((ot) & DMU_OT_NEWTYPE) ? \ ((ot) & DMU_OT_ENCRYPTED) : \ dmu_ot[(int)(ot)].ot_encrypt) /* * 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[(int)(ot)].ot_byteswap) 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_METDATA() 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; typedef enum txg_how { TXG_WAIT = 1, TXG_NOWAIT, TXG_WAITED, } txg_how_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) /* * Zap prefix for object accounting in DMU_{USER,GROUP}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, void *tag, objset_t **osp); int dmu_objset_own(const char *name, dmu_objset_type_t type, boolean_t readonly, boolean_t key_required, void *tag, objset_t **osp); void dmu_objset_rele(objset_t *os, void *tag); void dmu_objset_disown(objset_t *os, boolean_t key_required, 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 dmu_objset_clone(const char *name, const char *origin); 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_snapshot_tmp(const char *, const char *, int); int dmu_objset_find(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_STATS "DDT-statistics" #define DMU_POOL_CREATION_VERSION "creation_version" #define DMU_POOL_SCAN "scan" #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" /* * 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_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); 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, dmu_tx_t *txp); /* * 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 allcated 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); /* * 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); int dmu_object_dirty_raw(objset_t *os, uint64_t object, dmu_tx_t *tx); 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); /* * 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 void dmu_write_policy(objset_t *os, dnode_t *dn, int level, int wp, struct zio_prop *zp); /* * 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_read() to * read db_data, 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_impl(objset_t *os, uint64_t object, void *tag, uint32_t flags, dmu_buf_t **dbp); int dmu_bonus_hold(objset_t *os, uint64_t object, void *tag, dmu_buf_t **); 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, void *tag, dmu_buf_t **dbp); int dmu_spill_hold_by_dnode(dnode_t *dn, uint32_t flags, void *tag, dmu_buf_t **dbp); int dmu_spill_hold_existing(dmu_buf_t *bonus, 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, void *tag, dmu_buf_t **, int flags); int dmu_buf_hold_by_dnode(dnode_t *dn, uint64_t offset, void *tag, dmu_buf_t **dbp, int flags); /* * 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, 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, void *tag); void dmu_buf_rele(dmu_buf_t *db, void *tag); uint64_t dmu_buf_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, void *tag, int *numbufsp, dmu_buf_t ***dbpp); void dmu_buf_rele_array(dmu_buf_t **, int numbufs, 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; /* * 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. */ /*ARGSUSED*/ 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) { 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); /* * 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); dnode_t *dmu_buf_dnode_enter(dmu_buf_t *db); void dmu_buf_dnode_exit(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_change_crypt_params(dmu_buf_t *db, 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_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, enum txg_how txg_how); 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. */ 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); /* * 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_range_raw(objset_t *os, uint64_t object, uint64_t offset, uint64_t size); int dmu_free_long_object(objset_t *os, uint64_t object); int dmu_free_long_object_raw(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. */ #define DMU_READ_PREFETCH 0 /* prefetch */ #define DMU_READ_NO_PREFETCH 1 /* don't prefetch */ #define DMU_READ_NO_DECRYPT 2 /* don't decrypt */ int dmu_read(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, void *buf, uint32_t flags); int dmu_read_by_dnode(dnode_t *dn, uint64_t offset, uint64_t size, void *buf, uint32_t flags); void dmu_write(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, const void *buf, dmu_tx_t *tx); void dmu_write_by_dnode(dnode_t *dn, uint64_t offset, uint64_t size, const void *buf, dmu_tx_t *tx); void dmu_prealloc(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, dmu_tx_t *tx); #ifdef _KERNEL #include int dmu_read_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size); int dmu_read_uio_dbuf(dmu_buf_t *zdb, struct uio *uio, uint64_t size); int dmu_read_uio_dnode(dnode_t *dn, struct uio *uio, uint64_t size); int dmu_write_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size, dmu_tx_t *tx); int dmu_write_uio_dbuf(dmu_buf_t *zdb, struct uio *uio, uint64_t size, dmu_tx_t *tx); int dmu_write_uio_dnode(dnode_t *dn, struct uio *uio, uint64_t size, dmu_tx_t *tx); #endif struct arc_buf *dmu_request_arcbuf(dmu_buf_t *handle, int size); void dmu_return_arcbuf(struct arc_buf *buf); void dmu_assign_arcbuf_by_dnode(dnode_t *dn, uint64_t offset, struct arc_buf *buf, dmu_tx_t *tx); void dmu_assign_arcbuf_by_dbuf(dmu_buf_t *handle, uint64_t offset, struct arc_buf *buf, dmu_tx_t *tx); #define dmu_assign_arcbuf dmu_assign_arcbuf_by_dbuf void dmu_convert_to_raw(dmu_buf_t *handle, boolean_t byteorder, const uint8_t *salt, const uint8_t *iv, const uint8_t *mac, dmu_tx_t *tx); void dmu_copy_from_buf(objset_t *os, uint64_t object, uint64_t offset, dmu_buf_t *handle, dmu_tx_t *tx); #ifdef HAVE_UIO_ZEROCOPY int dmu_xuio_init(struct xuio *uio, int niov); void dmu_xuio_fini(struct xuio *uio); int dmu_xuio_add(struct xuio *uio, struct arc_buf *abuf, offset_t off, size_t n); int dmu_xuio_cnt(struct xuio *uio); struct arc_buf *dmu_xuio_arcbuf(struct xuio *uio, int i); void dmu_xuio_clear(struct xuio *uio, int i); #endif /* HAVE_UIO_ZEROCOPY */ void xuio_stat_wbuf_copied(void); void xuio_stat_wbuf_nocopy(void); extern int zfs_prefetch_disable; extern int 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); 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_encrypt; char *ot_name; } dmu_object_type_info_t; typedef const struct dmu_object_byteswap_info { arc_byteswap_func_t ob_func; char *ob_name; } dmu_object_byteswap_info_t; extern const dmu_object_type_info_t dmu_ot[DMU_OT_NUMTYPES]; extern const 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. This is specifically optimized for zfs_getattr(). */ 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 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; char dds_origin[ZFS_MAX_DATASET_NAME_LEN]; } 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 */ timestruc_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_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, 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 int objset_used_cb_t(dmu_object_type_t bonustype, void *bonus, uint64_t *userp, uint64_t *groupp); extern void dmu_objset_register_type(dmu_objset_type_t ost, objset_used_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 zilog *zgd_zilog; + struct lwb *zgd_lwb; struct blkptr *zgd_bp; dmu_buf_t *zgd_db; struct rl *zgd_rl; 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); /* * 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, struct vnode *vp, offset_t *offp); /* CRC64 table */ #define ZFS_CRC64_POLY 0xC96C5795D7870F42ULL /* ECMA-182, reflected form */ extern uint64_t zfs_crc64_table[256]; extern int zfs_mdcomp_disable; #ifdef __cplusplus } #endif #endif /* _SYS_DMU_H */ diff --git a/include/sys/trace_zil.h b/include/sys/trace_zil.h index eaa391270d76..ff16c8686cc4 100644 --- a/include/sys/trace_zil.h +++ b/include/sys/trace_zil.h @@ -1,129 +1,221 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ #if defined(_KERNEL) && defined(HAVE_DECLARE_EVENT_CLASS) #undef TRACE_SYSTEM #define TRACE_SYSTEM zfs #undef TRACE_SYSTEM_VAR #define TRACE_SYSTEM_VAR zfs_zil #if !defined(_TRACE_ZIL_H) || defined(TRACE_HEADER_MULTI_READ) #define _TRACE_ZIL_H #include #include +#define ZILOG_TP_STRUCT_ENTRY \ + __field(uint64_t, zl_lr_seq) \ + __field(uint64_t, zl_commit_lr_seq) \ + __field(uint64_t, zl_destroy_txg) \ + __field(uint64_t, zl_replaying_seq) \ + __field(uint32_t, zl_suspend) \ + __field(uint8_t, zl_suspending) \ + __field(uint8_t, zl_keep_first) \ + __field(uint8_t, zl_replay) \ + __field(uint8_t, zl_stop_sync) \ + __field(uint8_t, zl_logbias) \ + __field(uint8_t, zl_sync) \ + __field(int, zl_parse_error) \ + __field(uint64_t, zl_parse_blk_seq) \ + __field(uint64_t, zl_parse_lr_seq) \ + __field(uint64_t, zl_parse_blk_count) \ + __field(uint64_t, zl_parse_lr_count) \ + __field(uint64_t, zl_cur_used) \ + __field(clock_t, zl_replay_time) \ + __field(uint64_t, zl_replay_blks) + +#define ZILOG_TP_FAST_ASSIGN \ + __entry->zl_lr_seq = zilog->zl_lr_seq; \ + __entry->zl_commit_lr_seq = zilog->zl_commit_lr_seq; \ + __entry->zl_destroy_txg = zilog->zl_destroy_txg; \ + __entry->zl_replaying_seq = zilog->zl_replaying_seq; \ + __entry->zl_suspend = zilog->zl_suspend; \ + __entry->zl_suspending = zilog->zl_suspending; \ + __entry->zl_keep_first = zilog->zl_keep_first; \ + __entry->zl_replay = zilog->zl_replay; \ + __entry->zl_stop_sync = zilog->zl_stop_sync; \ + __entry->zl_logbias = zilog->zl_logbias; \ + __entry->zl_sync = zilog->zl_sync; \ + __entry->zl_parse_error = zilog->zl_parse_error; \ + __entry->zl_parse_blk_seq = zilog->zl_parse_blk_seq; \ + __entry->zl_parse_lr_seq = zilog->zl_parse_lr_seq; \ + __entry->zl_parse_blk_count = zilog->zl_parse_blk_count;\ + __entry->zl_parse_lr_count = zilog->zl_parse_lr_count; \ + __entry->zl_cur_used = zilog->zl_cur_used; \ + __entry->zl_replay_time = zilog->zl_replay_time; \ + __entry->zl_replay_blks = zilog->zl_replay_blks; + +#define ZILOG_TP_PRINTK_FMT \ + "zl { lr_seq %llu commit_lr_seq %llu destroy_txg %llu " \ + "replaying_seq %llu suspend %u suspending %u keep_first %u " \ + "replay %u stop_sync %u logbias %u sync %u " \ + "parse_error %u parse_blk_seq %llu parse_lr_seq %llu " \ + "parse_blk_count %llu parse_lr_count %llu " \ + "cur_used %llu replay_time %lu replay_blks %llu }" + +#define ZILOG_TP_PRINTK_ARGS \ + __entry->zl_lr_seq, __entry->zl_commit_lr_seq, \ + __entry->zl_destroy_txg, __entry->zl_replaying_seq, \ + __entry->zl_suspend, __entry->zl_suspending, \ + __entry->zl_keep_first, __entry->zl_replay, \ + __entry->zl_stop_sync, __entry->zl_logbias, __entry->zl_sync, \ + __entry->zl_parse_error, __entry->zl_parse_blk_seq, \ + __entry->zl_parse_lr_seq, __entry->zl_parse_blk_count, \ + __entry->zl_parse_lr_count, __entry->zl_cur_used, \ + __entry->zl_replay_time, __entry->zl_replay_blks + +#define ITX_TP_STRUCT_ENTRY \ + __field(itx_wr_state_t, itx_wr_state) \ + __field(uint8_t, itx_sync) \ + __field(zil_callback_t, itx_callback) \ + __field(void *, itx_callback_data) \ + __field(uint64_t, itx_oid) \ + \ + __field(uint64_t, lrc_txtype) \ + __field(uint64_t, lrc_reclen) \ + __field(uint64_t, lrc_txg) \ + __field(uint64_t, lrc_seq) + +#define ITX_TP_FAST_ASSIGN \ + __entry->itx_wr_state = itx->itx_wr_state; \ + __entry->itx_sync = itx->itx_sync; \ + __entry->itx_callback = itx->itx_callback; \ + __entry->itx_callback_data = itx->itx_callback_data; \ + __entry->itx_oid = itx->itx_oid; \ + \ + __entry->lrc_txtype = itx->itx_lr.lrc_txtype; \ + __entry->lrc_reclen = itx->itx_lr.lrc_reclen; \ + __entry->lrc_txg = itx->itx_lr.lrc_txg; \ + __entry->lrc_seq = itx->itx_lr.lrc_seq; + +#define ITX_TP_PRINTK_FMT \ + "itx { wr_state %u sync %u callback %p callback_data %p oid %llu" \ + " { txtype %llu reclen %llu txg %llu seq %llu } }" + +#define ITX_TP_PRINTK_ARGS \ + __entry->itx_wr_state, __entry->itx_sync, __entry->itx_callback,\ + __entry->itx_callback_data, __entry->itx_oid, \ + __entry->lrc_txtype, __entry->lrc_reclen, __entry->lrc_txg, \ + __entry->lrc_seq + +#define ZCW_TP_STRUCT_ENTRY \ + __field(lwb_t *, zcw_lwb) \ + __field(boolean_t, zcw_done) \ + __field(int, zcw_zio_error) \ + +#define ZCW_TP_FAST_ASSIGN \ + __entry->zcw_lwb = zcw->zcw_lwb; \ + __entry->zcw_done = zcw->zcw_done; \ + __entry->zcw_zio_error = zcw->zcw_zio_error; + +#define ZCW_TP_PRINTK_FMT \ + "zcw { lwb %p done %u error %u }" + +#define ZCW_TP_PRINTK_ARGS \ + __entry->zcw_lwb, __entry->zcw_done, __entry->zcw_zio_error + /* - * Generic support for one argument tracepoints of the form: + * Generic support for two argument tracepoints of the form: * - * DTRACE_PROBE1(..., - * zilog_t *, ...); + * DTRACE_PROBE2(..., + * zilog_t *, ..., + * itx_t *, ...); */ /* BEGIN CSTYLED */ -DECLARE_EVENT_CLASS(zfs_zil_class, - TP_PROTO(zilog_t *zilog), - TP_ARGS(zilog), +DECLARE_EVENT_CLASS(zfs_zil_process_itx_class, + TP_PROTO(zilog_t *zilog, itx_t *itx), + TP_ARGS(zilog, itx), TP_STRUCT__entry( - __field(uint64_t, zl_lr_seq) - __field(uint64_t, zl_commit_lr_seq) - __field(uint64_t, zl_destroy_txg) - __field(uint64_t, zl_replaying_seq) - __field(uint32_t, zl_suspend) - __field(uint8_t, zl_suspending) - __field(uint8_t, zl_keep_first) - __field(uint8_t, zl_replay) - __field(uint8_t, zl_stop_sync) - __field(uint8_t, zl_writer) - __field(uint8_t, zl_logbias) - __field(uint8_t, zl_sync) - __field(int, zl_parse_error) - __field(uint64_t, zl_parse_blk_seq) - __field(uint64_t, zl_parse_lr_seq) - __field(uint64_t, zl_parse_blk_count) - __field(uint64_t, zl_parse_lr_count) - __field(uint64_t, zl_next_batch) - __field(uint64_t, zl_com_batch) - __field(uint64_t, zl_cur_used) - __field(clock_t, zl_replay_time) - __field(uint64_t, zl_replay_blks) + ZILOG_TP_STRUCT_ENTRY + ITX_TP_STRUCT_ENTRY ), TP_fast_assign( - __entry->zl_lr_seq = zilog->zl_lr_seq; - __entry->zl_commit_lr_seq = zilog->zl_commit_lr_seq; - __entry->zl_destroy_txg = zilog->zl_destroy_txg; - __entry->zl_replaying_seq = zilog->zl_replaying_seq; - __entry->zl_suspend = zilog->zl_suspend; - __entry->zl_suspending = zilog->zl_suspending; - __entry->zl_keep_first = zilog->zl_keep_first; - __entry->zl_replay = zilog->zl_replay; - __entry->zl_stop_sync = zilog->zl_stop_sync; - __entry->zl_writer = zilog->zl_writer; - __entry->zl_logbias = zilog->zl_logbias; - __entry->zl_sync = zilog->zl_sync; - __entry->zl_parse_error = zilog->zl_parse_error; - __entry->zl_parse_blk_seq = zilog->zl_parse_blk_seq; - __entry->zl_parse_lr_seq = zilog->zl_parse_lr_seq; - __entry->zl_parse_blk_count = zilog->zl_parse_blk_count; - __entry->zl_parse_lr_count = zilog->zl_parse_lr_count; - __entry->zl_next_batch = zilog->zl_next_batch; - __entry->zl_com_batch = zilog->zl_com_batch; - __entry->zl_cur_used = zilog->zl_cur_used; - __entry->zl_replay_time = zilog->zl_replay_time; - __entry->zl_replay_blks = zilog->zl_replay_blks; + ZILOG_TP_FAST_ASSIGN + ITX_TP_FAST_ASSIGN ), - TP_printk("zl { lr_seq %llu commit_lr_seq %llu destroy_txg %llu " - "replaying_seq %llu suspend %u suspending %u keep_first %u " - "replay %u stop_sync %u writer %u logbias %u sync %u " - "parse_error %u parse_blk_seq %llu parse_lr_seq %llu " - "parse_blk_count %llu parse_lr_count %llu next_batch %llu " - "com_batch %llu cur_used %llu replay_time %lu replay_blks %llu }", - __entry->zl_lr_seq, __entry->zl_commit_lr_seq, - __entry->zl_destroy_txg, __entry->zl_replaying_seq, - __entry->zl_suspend, __entry->zl_suspending, __entry->zl_keep_first, - __entry->zl_replay, __entry->zl_stop_sync, __entry->zl_writer, - __entry->zl_logbias, __entry->zl_sync, __entry->zl_parse_error, - __entry->zl_parse_blk_seq, __entry->zl_parse_lr_seq, - __entry->zl_parse_blk_count, __entry->zl_parse_lr_count, - __entry->zl_next_batch, __entry->zl_com_batch, __entry->zl_cur_used, - __entry->zl_replay_time, __entry->zl_replay_blks) + TP_printk( + ZILOG_TP_PRINTK_FMT " " ITX_TP_PRINTK_FMT, + ZILOG_TP_PRINTK_ARGS, ITX_TP_PRINTK_ARGS) ); /* END CSTYLED */ /* BEGIN CSTYLED */ -#define DEFINE_ZIL_EVENT(name) \ -DEFINE_EVENT(zfs_zil_class, name, \ - TP_PROTO(zilog_t *zilog), \ - TP_ARGS(zilog)) -DEFINE_ZIL_EVENT(zfs_zil__cw1); -DEFINE_ZIL_EVENT(zfs_zil__cw2); +#define DEFINE_ZIL_PROCESS_ITX_EVENT(name) \ +DEFINE_EVENT(zfs_zil_process_itx_class, name, \ + TP_PROTO(zilog_t *zilog, itx_t *itx), \ + TP_ARGS(zilog, itx)) +DEFINE_ZIL_PROCESS_ITX_EVENT(zfs_zil__process__commit__itx); +DEFINE_ZIL_PROCESS_ITX_EVENT(zfs_zil__process__normal__itx); +/* END CSTYLED */ + +/* + * Generic support for two argument tracepoints of the form: + * + * DTRACE_PROBE2(..., + * zilog_t *, ..., + * zil_commit_waiter_t *, ...); + */ +/* BEGIN CSTYLED */ +DECLARE_EVENT_CLASS(zfs_zil_commit_io_error_class, + TP_PROTO(zilog_t *zilog, zil_commit_waiter_t *zcw), + TP_ARGS(zilog, zcw), + TP_STRUCT__entry( + ZILOG_TP_STRUCT_ENTRY + ZCW_TP_STRUCT_ENTRY + ), + TP_fast_assign( + ZILOG_TP_FAST_ASSIGN + ZCW_TP_FAST_ASSIGN + ), + TP_printk( + ZILOG_TP_PRINTK_FMT " " ZCW_TP_PRINTK_FMT, + ZILOG_TP_PRINTK_ARGS, ZCW_TP_PRINTK_ARGS) +); + +/* BEGIN CSTYLED */ +#define DEFINE_ZIL_COMMIT_IO_ERROR_EVENT(name) \ +DEFINE_EVENT(zfs_zil_commit_io_error_class, name, \ + TP_PROTO(zilog_t *zilog, zil_commit_waiter_t *zcw), \ + TP_ARGS(zilog, zcw)) +DEFINE_ZIL_COMMIT_IO_ERROR_EVENT(zfs_zil__commit__io__error); /* END CSTYLED */ #endif /* _TRACE_ZIL_H */ #undef TRACE_INCLUDE_PATH #undef TRACE_INCLUDE_FILE #define TRACE_INCLUDE_PATH sys #define TRACE_INCLUDE_FILE trace_zil #include #endif /* _KERNEL && HAVE_DECLARE_EVENT_CLASS */ diff --git a/include/sys/zil.h b/include/sys/zil.h index 442fee4a1f8f..5513c7ce8ed9 100644 --- a/include/sys/zil.h +++ b/include/sys/zil.h @@ -1,519 +1,523 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2012, 2017 by Delphix. All rights reserved. */ /* Portions Copyright 2010 Robert Milkowski */ #ifndef _SYS_ZIL_H #define _SYS_ZIL_H #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif struct dsl_pool; struct dsl_dataset; +struct lwb; /* * Intent log format: * * Each objset has its own intent log. The log header (zil_header_t) * for objset N's intent log is kept in the Nth object of the SPA's * intent_log objset. The log header points to a chain of log blocks, * each of which contains log records (i.e., transactions) followed by * a log block trailer (zil_trailer_t). The format of a log record * depends on the record (or transaction) type, but all records begin * with a common structure that defines the type, length, and txg. */ /* * Intent log header - this on disk structure holds fields to manage * the log. All fields are 64 bit to easily handle cross architectures. */ typedef struct zil_header { uint64_t zh_claim_txg; /* txg in which log blocks were claimed */ uint64_t zh_replay_seq; /* highest replayed sequence number */ blkptr_t zh_log; /* log chain */ uint64_t zh_claim_blk_seq; /* highest claimed block sequence number */ uint64_t zh_flags; /* header flags */ uint64_t zh_claim_lr_seq; /* highest claimed lr sequence number */ uint64_t zh_pad[3]; } zil_header_t; /* * zh_flags bit settings */ #define ZIL_REPLAY_NEEDED 0x1 /* replay needed - internal only */ #define ZIL_CLAIM_LR_SEQ_VALID 0x2 /* zh_claim_lr_seq field is valid */ /* * Log block chaining. * * Log blocks are chained together. Originally they were chained at the * end of the block. For performance reasons the chain was moved to the * beginning of the block which allows writes for only the data being used. * The older position is supported for backwards compatability. * * The zio_eck_t contains a zec_cksum which for the intent log is * the sequence number of this log block. A seq of 0 is invalid. * The zec_cksum is checked by the SPA against the sequence * number passed in the blk_cksum field of the blkptr_t */ typedef struct zil_chain { uint64_t zc_pad; blkptr_t zc_next_blk; /* next block in chain */ uint64_t zc_nused; /* bytes in log block used */ zio_eck_t zc_eck; /* block trailer */ } zil_chain_t; #define ZIL_MIN_BLKSZ 4096ULL /* * ziltest is by and large an ugly hack, but very useful in * checking replay without tedious work. * When running ziltest we want to keep all itx's and so maintain * a single list in the zl_itxg[] that uses a high txg: ZILTEST_TXG * We subtract TXG_CONCURRENT_STATES to allow for common code. */ #define ZILTEST_TXG (UINT64_MAX - TXG_CONCURRENT_STATES) /* * The words of a log block checksum. */ #define ZIL_ZC_GUID_0 0 #define ZIL_ZC_GUID_1 1 #define ZIL_ZC_OBJSET 2 #define ZIL_ZC_SEQ 3 typedef enum zil_create { Z_FILE, Z_DIR, Z_XATTRDIR, } zil_create_t; /* * size of xvattr log section. * its composed of lr_attr_t + xvattr bitmap + 2 64 bit timestamps * for create time and a single 64 bit integer for all of the attributes, * and 4 64 bit integers (32 bytes) for the scanstamp. * */ #define ZIL_XVAT_SIZE(mapsize) \ sizeof (lr_attr_t) + (sizeof (uint32_t) * (mapsize - 1)) + \ (sizeof (uint64_t) * 7) /* * Size of ACL in log. The ACE data is padded out to properly align * on 8 byte boundary. */ #define ZIL_ACE_LENGTH(x) (roundup(x, sizeof (uint64_t))) /* * Intent log transaction types and record structures */ +#define TX_COMMIT 0 /* Commit marker (no on-disk state) */ #define TX_CREATE 1 /* Create file */ #define TX_MKDIR 2 /* Make directory */ #define TX_MKXATTR 3 /* Make XATTR directory */ #define TX_SYMLINK 4 /* Create symbolic link to a file */ #define TX_REMOVE 5 /* Remove file */ #define TX_RMDIR 6 /* Remove directory */ #define TX_LINK 7 /* Create hard link to a file */ #define TX_RENAME 8 /* Rename a file */ #define TX_WRITE 9 /* File write */ #define TX_TRUNCATE 10 /* Truncate a file */ #define TX_SETATTR 11 /* Set file attributes */ #define TX_ACL_V0 12 /* Set old formatted ACL */ #define TX_ACL 13 /* Set ACL */ #define TX_CREATE_ACL 14 /* create with ACL */ #define TX_CREATE_ATTR 15 /* create + attrs */ #define TX_CREATE_ACL_ATTR 16 /* create with ACL + attrs */ #define TX_MKDIR_ACL 17 /* mkdir with ACL */ #define TX_MKDIR_ATTR 18 /* mkdir with attr */ #define TX_MKDIR_ACL_ATTR 19 /* mkdir with ACL + attrs */ #define TX_WRITE2 20 /* dmu_sync EALREADY write */ #define TX_MAX_TYPE 21 /* Max transaction type */ /* * The transactions for mkdir, symlink, remove, rmdir, link, and rename * may have the following bit set, indicating the original request * specified case-insensitive handling of names. */ #define TX_CI ((uint64_t)0x1 << 63) /* case-insensitive behavior requested */ /* * Transactions for write, truncate, setattr, acl_v0, and acl can be logged * out of order. For convenience in the code, all such records must have * lr_foid at the same offset. */ #define TX_OOO(txtype) \ ((txtype) == TX_WRITE || \ (txtype) == TX_TRUNCATE || \ (txtype) == TX_SETATTR || \ (txtype) == TX_ACL_V0 || \ (txtype) == TX_ACL || \ (txtype) == TX_WRITE2) /* * The number of dnode slots consumed by the object is stored in the 8 * unused upper bits of the object ID. We subtract 1 from the value * stored on disk for compatibility with implementations that don't * support large dnodes. The slot count for a single-slot dnode will * contain 0 for those bits to preserve the log record format for * "small" dnodes. */ #define LR_FOID_GET_SLOTS(oid) (BF64_GET((oid), 56, 8) + 1) #define LR_FOID_SET_SLOTS(oid, x) BF64_SET((oid), 56, 8, (x) - 1) #define LR_FOID_GET_OBJ(oid) BF64_GET((oid), 0, DN_MAX_OBJECT_SHIFT) #define LR_FOID_SET_OBJ(oid, x) BF64_SET((oid), 0, DN_MAX_OBJECT_SHIFT, (x)) /* * Format of log records. * The fields are carefully defined to allow them to be aligned * and sized the same on sparc & intel architectures. * Each log record has a common structure at the beginning. * * The log record on disk (lrc_seq) holds the sequence number of all log * records which is used to ensure we don't replay the same record. */ typedef struct { /* common log record header */ uint64_t lrc_txtype; /* intent log transaction type */ uint64_t lrc_reclen; /* transaction record length */ uint64_t lrc_txg; /* dmu transaction group number */ uint64_t lrc_seq; /* see comment above */ } lr_t; /* * Common start of all out-of-order record types (TX_OOO() above). */ typedef struct { lr_t lr_common; /* common portion of log record */ uint64_t lr_foid; /* object id */ } lr_ooo_t; /* * Handle option extended vattr attributes. * * Whenever new attributes are added the version number * will need to be updated as will code in * zfs_log.c and zfs_replay.c */ typedef struct { uint32_t lr_attr_masksize; /* number of elements in array */ uint32_t lr_attr_bitmap; /* First entry of array */ /* remainder of array and any additional fields */ } lr_attr_t; /* * log record for creates without optional ACL. * This log record does support optional xvattr_t attributes. */ typedef struct { lr_t lr_common; /* common portion of log record */ uint64_t lr_doid; /* object id of directory */ uint64_t lr_foid; /* object id of created file object */ uint64_t lr_mode; /* mode of object */ uint64_t lr_uid; /* uid of object */ uint64_t lr_gid; /* gid of object */ uint64_t lr_gen; /* generation (txg of creation) */ uint64_t lr_crtime[2]; /* creation time */ uint64_t lr_rdev; /* rdev of object to create */ /* name of object to create follows this */ /* for symlinks, link content follows name */ /* for creates with xvattr data, the name follows the xvattr info */ } lr_create_t; /* * FUID ACL record will be an array of ACEs from the original ACL. * If this array includes ephemeral IDs, the record will also include * an array of log-specific FUIDs to replace the ephemeral IDs. * Only one copy of each unique domain will be present, so the log-specific * FUIDs will use an index into a compressed domain table. On replay this * information will be used to construct real FUIDs (and bypass idmap, * since it may not be available). */ /* * Log record for creates with optional ACL * This log record is also used for recording any FUID * information needed for replaying the create. If the * file doesn't have any actual ACEs then the lr_aclcnt * would be zero. * * After lr_acl_flags, there are a lr_acl_bytes number of variable sized ace's. * If create is also setting xvattr's, then acl data follows xvattr. * If ACE FUIDs are needed then they will follow the xvattr_t. Following * the FUIDs will be the domain table information. The FUIDs for the owner * and group will be in lr_create. Name follows ACL data. */ typedef struct { lr_create_t lr_create; /* common create portion */ uint64_t lr_aclcnt; /* number of ACEs in ACL */ uint64_t lr_domcnt; /* number of unique domains */ uint64_t lr_fuidcnt; /* number of real fuids */ uint64_t lr_acl_bytes; /* number of bytes in ACL */ uint64_t lr_acl_flags; /* ACL flags */ } lr_acl_create_t; typedef struct { lr_t lr_common; /* common portion of log record */ uint64_t lr_doid; /* obj id of directory */ /* name of object to remove follows this */ } lr_remove_t; typedef struct { lr_t lr_common; /* common portion of log record */ uint64_t lr_doid; /* obj id of directory */ uint64_t lr_link_obj; /* obj id of link */ /* name of object to link follows this */ } lr_link_t; typedef struct { lr_t lr_common; /* common portion of log record */ uint64_t lr_sdoid; /* obj id of source directory */ uint64_t lr_tdoid; /* obj id of target directory */ /* 2 strings: names of source and destination follow this */ } lr_rename_t; typedef struct { lr_t lr_common; /* common portion of log record */ uint64_t lr_foid; /* file object to write */ uint64_t lr_offset; /* offset to write to */ uint64_t lr_length; /* user data length to write */ uint64_t lr_blkoff; /* no longer used */ blkptr_t lr_blkptr; /* spa block pointer for replay */ /* write data will follow for small writes */ } lr_write_t; typedef struct { lr_t lr_common; /* common portion of log record */ uint64_t lr_foid; /* object id of file to truncate */ uint64_t lr_offset; /* offset to truncate from */ uint64_t lr_length; /* length to truncate */ } lr_truncate_t; typedef struct { lr_t lr_common; /* common portion of log record */ uint64_t lr_foid; /* file object to change attributes */ uint64_t lr_mask; /* mask of attributes to set */ uint64_t lr_mode; /* mode to set */ uint64_t lr_uid; /* uid to set */ uint64_t lr_gid; /* gid to set */ uint64_t lr_size; /* size to set */ uint64_t lr_atime[2]; /* access time */ uint64_t lr_mtime[2]; /* modification time */ /* optional attribute lr_attr_t may be here */ } lr_setattr_t; typedef struct { lr_t lr_common; /* common portion of log record */ uint64_t lr_foid; /* obj id of file */ uint64_t lr_aclcnt; /* number of acl entries */ /* lr_aclcnt number of ace_t entries follow this */ } lr_acl_v0_t; typedef struct { lr_t lr_common; /* common portion of log record */ uint64_t lr_foid; /* obj id of file */ uint64_t lr_aclcnt; /* number of ACEs in ACL */ uint64_t lr_domcnt; /* number of unique domains */ uint64_t lr_fuidcnt; /* number of real fuids */ uint64_t lr_acl_bytes; /* number of bytes in ACL */ uint64_t lr_acl_flags; /* ACL flags */ /* lr_acl_bytes number of variable sized ace's follows */ } lr_acl_t; /* * ZIL structure definitions, interface function prototype and globals. */ /* * Writes are handled in three different ways: * * WR_INDIRECT: * In this mode, if we need to commit the write later, then the block * is immediately written into the file system (using dmu_sync), * and a pointer to the block is put into the log record. * When the txg commits the block is linked in. * This saves additionally writing the data into the log record. * There are a few requirements for this to occur: * - write is greater than zfs/zvol_immediate_write_sz * - not using slogs (as slogs are assumed to always be faster * than writing into the main pool) * - the write occupies only one block * WR_COPIED: * If we know we'll immediately be committing the * transaction (FSYNC or FDSYNC), then we allocate a larger * log record here for the data and copy the data in. * WR_NEED_COPY: * Otherwise we don't allocate a buffer, and *if* we need to * flush the write later then a buffer is allocated and * we retrieve the data using the dmu. */ typedef enum { WR_INDIRECT, /* indirect - a large write (dmu_sync() data */ /* and put blkptr in log, rather than actual data) */ WR_COPIED, /* immediate - data is copied into lr_write_t */ WR_NEED_COPY, /* immediate - data needs to be copied if pushed */ WR_NUM_STATES /* number of states */ } itx_wr_state_t; typedef void (*zil_callback_t)(void *data); typedef struct itx { list_node_t itx_node; /* linkage on zl_itx_list */ void *itx_private; /* type-specific opaque data */ itx_wr_state_t itx_wr_state; /* write state */ uint8_t itx_sync; /* synchronous transaction */ zil_callback_t itx_callback; /* Called when the itx is persistent */ void *itx_callback_data; /* User data for the callback */ size_t itx_size; /* allocated itx structure size */ uint64_t itx_oid; /* object id */ lr_t itx_lr; /* common part of log record */ /* followed by type-specific part of lr_xx_t and its immediate data */ } itx_t; /* * Used for zil kstat. */ typedef struct zil_stats { /* * Number of times a ZIL commit (e.g. fsync) has been requested. */ kstat_named_t zil_commit_count; /* * Number of times the ZIL has been flushed to stable storage. * This is less than zil_commit_count when commits are "merged" * (see the documentation above zil_commit()). */ kstat_named_t zil_commit_writer_count; /* * Number of transactions (reads, writes, renames, etc.) * that have been commited. */ kstat_named_t zil_itx_count; /* * See the documentation for itx_wr_state_t above. * Note that "bytes" accumulates the length of the transactions * (i.e. data), not the actual log record sizes. */ kstat_named_t zil_itx_indirect_count; kstat_named_t zil_itx_indirect_bytes; kstat_named_t zil_itx_copied_count; kstat_named_t zil_itx_copied_bytes; kstat_named_t zil_itx_needcopy_count; kstat_named_t zil_itx_needcopy_bytes; /* * Transactions which have been allocated to the "normal" * (i.e. not slog) storage pool. Note that "bytes" accumulate * the actual log record sizes - which do not include the actual * data in case of indirect writes. */ kstat_named_t zil_itx_metaslab_normal_count; kstat_named_t zil_itx_metaslab_normal_bytes; /* * Transactions which have been allocated to the "slog" storage pool. * If there are no separate log devices, this is the same as the * "normal" pool. */ kstat_named_t zil_itx_metaslab_slog_count; kstat_named_t zil_itx_metaslab_slog_bytes; } zil_stats_t; extern zil_stats_t zil_stats; #define ZIL_STAT_INCR(stat, val) \ atomic_add_64(&zil_stats.stat.value.ui64, (val)); #define ZIL_STAT_BUMP(stat) \ ZIL_STAT_INCR(stat, 1); typedef int zil_parse_blk_func_t(zilog_t *zilog, blkptr_t *bp, void *arg, uint64_t txg); typedef int zil_parse_lr_func_t(zilog_t *zilog, lr_t *lr, void *arg, uint64_t txg); typedef int zil_replay_func_t(void *arg1, void *arg2, boolean_t byteswap); -typedef int zil_get_data_t(void *arg, lr_write_t *lr, char *dbuf, zio_t *zio); +typedef int zil_get_data_t(void *arg, lr_write_t *lr, char *dbuf, + struct lwb *lwb, zio_t *zio); extern int zil_parse(zilog_t *zilog, zil_parse_blk_func_t *parse_blk_func, zil_parse_lr_func_t *parse_lr_func, void *arg, uint64_t txg, boolean_t decrypt); extern void zil_init(void); extern void zil_fini(void); extern zilog_t *zil_alloc(objset_t *os, zil_header_t *zh_phys); extern void zil_free(zilog_t *zilog); extern zilog_t *zil_open(objset_t *os, zil_get_data_t *get_data); extern void zil_close(zilog_t *zilog); extern void zil_replay(objset_t *os, void *arg, zil_replay_func_t *replay_func[TX_MAX_TYPE]); extern boolean_t zil_replaying(zilog_t *zilog, dmu_tx_t *tx); extern void zil_destroy(zilog_t *zilog, boolean_t keep_first); extern void zil_destroy_sync(zilog_t *zilog, dmu_tx_t *tx); extern itx_t *zil_itx_create(uint64_t txtype, size_t lrsize); extern void zil_itx_destroy(itx_t *itx); extern void zil_itx_assign(zilog_t *zilog, itx_t *itx, dmu_tx_t *tx); extern void zil_commit(zilog_t *zilog, uint64_t oid); extern int zil_vdev_offline(const char *osname, void *txarg); extern int zil_claim(struct dsl_pool *dp, struct dsl_dataset *ds, void *txarg); extern int zil_check_log_chain(struct dsl_pool *dp, struct dsl_dataset *ds, void *tx); extern void zil_sync(zilog_t *zilog, dmu_tx_t *tx); extern void zil_clean(zilog_t *zilog, uint64_t synced_txg); extern int zil_suspend(const char *osname, void **cookiep); extern void zil_resume(void *cookie); -extern void zil_add_block(zilog_t *zilog, const blkptr_t *bp); +extern void zil_lwb_add_block(struct lwb *lwb, const blkptr_t *bp); +extern void zil_lwb_add_txg(struct lwb *lwb, uint64_t txg); extern int zil_bp_tree_add(zilog_t *zilog, const blkptr_t *bp); extern void zil_set_sync(zilog_t *zilog, uint64_t syncval); extern void zil_set_logbias(zilog_t *zilog, uint64_t slogval); extern int zil_replay_disable; #ifdef __cplusplus } #endif #endif /* _SYS_ZIL_H */ diff --git a/include/sys/zil_impl.h b/include/sys/zil_impl.h index dd5304b79a91..3f69b326549b 100644 --- a/include/sys/zil_impl.h +++ b/include/sys/zil_impl.h @@ -1,165 +1,216 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. - * Copyright (c) 2012 by Delphix. All rights reserved. + * Copyright (c) 2012, 2017 by Delphix. All rights reserved. */ /* Portions Copyright 2010 Robert Milkowski */ #ifndef _SYS_ZIL_IMPL_H #define _SYS_ZIL_IMPL_H #include #include #ifdef __cplusplus extern "C" { #endif /* - * Log write buffer. + * Possbile states for a given lwb structure. An lwb will start out in + * the "closed" state, and then transition to the "opened" state via a + * call to zil_lwb_write_open(). After the lwb is "open", it can + * transition into the "issued" state via zil_lwb_write_issue(). After + * the lwb's zio completes, and the vdev's are flushed, the lwb will + * transition into the "done" state via zil_lwb_write_done(), and the + * structure eventually freed. + */ +typedef enum { + LWB_STATE_CLOSED, + LWB_STATE_OPENED, + LWB_STATE_ISSUED, + LWB_STATE_DONE, + LWB_NUM_STATES +} lwb_state_t; + +/* + * Log write block (lwb) + * + * Prior to an lwb being issued to disk via zil_lwb_write_issue(), it + * will be protected by the zilog's "zl_writer_lock". Basically, prior + * to it being issued, it will only be accessed by the thread that's + * holding the "zl_writer_lock". After the lwb is issued, the zilog's + * "zl_lock" is used to protect the lwb against concurrent access. */ typedef struct lwb { zilog_t *lwb_zilog; /* back pointer to log struct */ blkptr_t lwb_blk; /* on disk address of this log blk */ boolean_t lwb_fastwrite; /* is blk marked for fastwrite? */ boolean_t lwb_slog; /* lwb_blk is on SLOG device */ int lwb_nused; /* # used bytes in buffer */ int lwb_sz; /* size of block and buffer */ + lwb_state_t lwb_state; /* the state of this lwb */ char *lwb_buf; /* log write buffer */ - zio_t *lwb_zio; /* zio for this buffer */ + zio_t *lwb_write_zio; /* zio for the lwb buffer */ + zio_t *lwb_root_zio; /* root zio for lwb write and flushes */ dmu_tx_t *lwb_tx; /* tx for log block allocation */ uint64_t lwb_max_txg; /* highest txg in this lwb */ list_node_t lwb_node; /* zilog->zl_lwb_list linkage */ + list_t lwb_itxs; /* list of itx's */ + list_t lwb_waiters; /* list of zil_commit_waiter's */ + avl_tree_t lwb_vdev_tree; /* vdevs to flush after lwb write */ + kmutex_t lwb_vdev_lock; /* protects lwb_vdev_tree */ + hrtime_t lwb_issued_timestamp; /* when was the lwb issued? */ } lwb_t; +/* + * ZIL commit waiter. + * + * This structure is allocated each time zil_commit() is called, and is + * used by zil_commit() to communicate with other parts of the ZIL, such + * that zil_commit() can know when it safe for it return. For more + * details, see the comment above zil_commit(). + * + * The "zcw_lock" field is used to protect the commit waiter against + * concurrent access. This lock is often acquired while already holding + * the zilog's "zl_writer_lock" or "zl_lock"; see the functions + * zil_process_commit_list() and zil_lwb_flush_vdevs_done() as examples + * of this. Thus, one must be careful not to acquire the + * "zl_writer_lock" or "zl_lock" when already holding the "zcw_lock"; + * e.g. see the zil_commit_waiter_timeout() function. + */ +typedef struct zil_commit_waiter { + kcondvar_t zcw_cv; /* signalled when "done" */ + kmutex_t zcw_lock; /* protects fields of this struct */ + list_node_t zcw_node; /* linkage in lwb_t:lwb_waiter list */ + lwb_t *zcw_lwb; /* back pointer to lwb when linked */ + boolean_t zcw_done; /* B_TRUE when "done", else B_FALSE */ + int zcw_zio_error; /* contains the zio io_error value */ +} zil_commit_waiter_t; + /* * Intent log transaction lists */ typedef struct itxs { list_t i_sync_list; /* list of synchronous itxs */ avl_tree_t i_async_tree; /* tree of foids for async itxs */ } itxs_t; typedef struct itxg { kmutex_t itxg_lock; /* lock for this structure */ uint64_t itxg_txg; /* txg for this chain */ itxs_t *itxg_itxs; /* sync and async itxs */ } itxg_t; /* for async nodes we build up an AVL tree of lists of async itxs per file */ typedef struct itx_async_node { uint64_t ia_foid; /* file object id */ list_t ia_list; /* list of async itxs for this foid */ avl_node_t ia_node; /* AVL tree linkage */ } itx_async_node_t; /* * Vdev flushing: during a zil_commit(), we build up an AVL tree of the vdevs * we've touched so we know which ones need a write cache flush at the end. */ typedef struct zil_vdev_node { uint64_t zv_vdev; /* vdev to be flushed */ avl_node_t zv_node; /* AVL tree linkage */ } zil_vdev_node_t; #define ZIL_PREV_BLKS 16 /* * Stable storage intent log management structure. One per dataset. */ struct zilog { kmutex_t zl_lock; /* protects most zilog_t fields */ struct dsl_pool *zl_dmu_pool; /* DSL pool */ spa_t *zl_spa; /* handle for read/write log */ const zil_header_t *zl_header; /* log header buffer */ objset_t *zl_os; /* object set we're logging */ zil_get_data_t *zl_get_data; /* callback to get object content */ - zio_t *zl_root_zio; /* log writer root zio */ + lwb_t *zl_last_lwb_opened; /* most recent lwb opened */ + hrtime_t zl_last_lwb_latency; /* zio latency of last lwb done */ uint64_t zl_lr_seq; /* on-disk log record sequence number */ uint64_t zl_commit_lr_seq; /* last committed on-disk lr seq */ uint64_t zl_destroy_txg; /* txg of last zil_destroy() */ uint64_t zl_replayed_seq[TXG_SIZE]; /* last replayed rec seq */ uint64_t zl_replaying_seq; /* current replay seq number */ uint32_t zl_suspend; /* log suspend count */ - kcondvar_t zl_cv_writer; /* log writer thread completion */ kcondvar_t zl_cv_suspend; /* log suspend completion */ uint8_t zl_suspending; /* log is currently suspending */ uint8_t zl_keep_first; /* keep first log block in destroy */ uint8_t zl_replay; /* replaying records while set */ uint8_t zl_stop_sync; /* for debugging */ - uint8_t zl_writer; /* boolean: write setup in progress */ + kmutex_t zl_writer_lock; /* single writer, per ZIL, at a time */ uint8_t zl_logbias; /* latency or throughput */ uint8_t zl_sync; /* synchronous or asynchronous */ int zl_parse_error; /* last zil_parse() error */ uint64_t zl_parse_blk_seq; /* highest blk seq on last parse */ uint64_t zl_parse_lr_seq; /* highest lr seq on last parse */ uint64_t zl_parse_blk_count; /* number of blocks parsed */ uint64_t zl_parse_lr_count; /* number of log records parsed */ - uint64_t zl_next_batch; /* next batch number */ - uint64_t zl_com_batch; /* committed batch number */ - kcondvar_t zl_cv_batch[2]; /* batch condition variables */ itxg_t zl_itxg[TXG_SIZE]; /* intent log txg chains */ list_t zl_itx_commit_list; /* itx list to be committed */ uint64_t zl_cur_used; /* current commit log size used */ list_t zl_lwb_list; /* in-flight log write list */ - kmutex_t zl_vdev_lock; /* protects zl_vdev_tree */ - avl_tree_t zl_vdev_tree; /* vdevs to flush in zil_commit() */ avl_tree_t zl_bp_tree; /* track bps during log parse */ clock_t zl_replay_time; /* lbolt of when replay started */ uint64_t zl_replay_blks; /* number of log blocks replayed */ zil_header_t zl_old_header; /* debugging aid */ uint_t zl_prev_blks[ZIL_PREV_BLKS]; /* size - sector rounded */ uint_t zl_prev_rotor; /* rotor for zl_prev[] */ txg_node_t zl_dirty_link; /* protected by dp_dirty_zilogs list */ + uint64_t zl_dirty_max_txg; /* highest txg used to dirty zilog */ }; typedef struct zil_bp_node { dva_t zn_dva; avl_node_t zn_node; } zil_bp_node_t; /* * Maximum amount of write data that can be put into single log block. */ #define ZIL_MAX_LOG_DATA (SPA_OLD_MAXBLOCKSIZE - sizeof (zil_chain_t) - \ sizeof (lr_write_t)) /* * Maximum amount of log space we agree to waste to reduce number of * WR_NEED_COPY chunks to reduce zl_get_data() overhead (~12%). */ #define ZIL_MAX_WASTE_SPACE (ZIL_MAX_LOG_DATA / 8) /* * Maximum amount of write data for WR_COPIED. Fall back to WR_NEED_COPY * as more space efficient if we can't fit at least two log records into * maximum sized log block. */ #define ZIL_MAX_COPIED_DATA ((SPA_OLD_MAXBLOCKSIZE - \ sizeof (zil_chain_t)) / 2 - sizeof (lr_write_t)) #ifdef __cplusplus } #endif #endif /* _SYS_ZIL_IMPL_H */ diff --git a/include/sys/zio.h b/include/sys/zio.h index 30c4ee3ce4c3..1f9f48268463 100644 --- a/include/sys/zio.h +++ b/include/sys/zio.h @@ -1,657 +1,658 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2012, 2017 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 */ #ifndef _ZIO_H #define _ZIO_H #include #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. */ #define SPA_GANGBLOCKSIZE SPA_MINBLOCKSIZE #define SPA_GBH_NBLKPTRS ((SPA_GANGBLOCKSIZE - \ sizeof (zio_eck_t)) / sizeof (blkptr_t)) #define SPA_GBH_FILLER ((SPA_GANGBLOCKSIZE - \ sizeof (zio_eck_t) - \ (SPA_GBH_NBLKPTRS * sizeof (blkptr_t))) /\ sizeof (uint64_t)) typedef struct zio_gbh { blkptr_t zg_blkptr[SPA_GBH_NBLKPTRS]; uint64_t zg_filler[SPA_GBH_FILLER]; zio_eck_t zg_tail; } zio_gbh_phys_t; 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_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 (1 << 8) #define ZIO_DEDUPCHECKSUM ZIO_CHECKSUM_SHA256 #define ZIO_DEDUPDITTO_MIN 100 /* 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_CCM #define ZIO_CRYPT_DEFAULT ZIO_CRYPT_OFF /* 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_OFF #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_ON || \ (compress) == ZIO_COMPRESS_OFF) /* * Default Linux timeout for a sd device. */ #define ZIO_DELAY_MAX (30 * MILLISEC) #define ZIO_FAILURE_MODE_WAIT 0 #define ZIO_FAILURE_MODE_CONTINUE 1 #define ZIO_FAILURE_MODE_PANIC 2 enum zio_flag { /* * Flags inherited by gang, ddt, and vdev children, * and that must be equal for two zios to aggregate */ ZIO_FLAG_DONT_AGGREGATE = 1 << 0, ZIO_FLAG_IO_REPAIR = 1 << 1, ZIO_FLAG_SELF_HEAL = 1 << 2, ZIO_FLAG_RESILVER = 1 << 3, ZIO_FLAG_SCRUB = 1 << 4, ZIO_FLAG_SCAN_THREAD = 1 << 5, ZIO_FLAG_PHYSICAL = 1 << 6, #define ZIO_FLAG_AGG_INHERIT (ZIO_FLAG_CANFAIL - 1) /* * Flags inherited by ddt, gang, and vdev children. */ ZIO_FLAG_CANFAIL = 1 << 7, /* must be first for INHERIT */ ZIO_FLAG_SPECULATIVE = 1 << 8, ZIO_FLAG_CONFIG_WRITER = 1 << 9, ZIO_FLAG_DONT_RETRY = 1 << 10, ZIO_FLAG_DONT_CACHE = 1 << 11, ZIO_FLAG_NODATA = 1 << 12, ZIO_FLAG_INDUCE_DAMAGE = 1 << 13, ZIO_FLAG_IO_ALLOCATING = 1 << 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. */ ZIO_FLAG_IO_RETRY = 1 << 15, /* must be first for INHERIT */ ZIO_FLAG_PROBE = 1 << 16, ZIO_FLAG_TRYHARD = 1 << 17, ZIO_FLAG_OPTIONAL = 1 << 18, #define ZIO_FLAG_VDEV_INHERIT (ZIO_FLAG_DONT_QUEUE - 1) /* * Flags not inherited by any children. */ ZIO_FLAG_DONT_QUEUE = 1 << 19, /* must be first for INHERIT */ ZIO_FLAG_DONT_PROPAGATE = 1 << 20, ZIO_FLAG_IO_BYPASS = 1 << 21, ZIO_FLAG_IO_REWRITE = 1 << 22, ZIO_FLAG_RAW_COMPRESS = 1 << 23, ZIO_FLAG_RAW_ENCRYPT = 1 << 24, ZIO_FLAG_GANG_CHILD = 1 << 25, ZIO_FLAG_DDT_CHILD = 1 << 26, ZIO_FLAG_GODFATHER = 1 << 27, ZIO_FLAG_NOPWRITE = 1 << 28, ZIO_FLAG_REEXECUTED = 1 << 29, ZIO_FLAG_DELEGATED = 1 << 30, ZIO_FLAG_FASTWRITE = 1 << 31, }; #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_CANFAIL) enum zio_child { ZIO_CHILD_VDEV = 0, ZIO_CHILD_GANG, ZIO_CHILD_DDT, ZIO_CHILD_LOGICAL, ZIO_CHILD_TYPES }; enum zio_wait_type { ZIO_WAIT_READY = 0, ZIO_WAIT_DONE, ZIO_WAIT_TYPES }; /* * We'll take the unused errnos, 'EBADE' and 'EBADR' (from the Convergent * graveyard) to indicate checksum errors and fragmentation. */ #define ECKSUM EBADE #define EFRAGS EBADR typedef void zio_done_func_t(zio_t *zio); extern int zio_dva_throttle_enabled; extern const char *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). */ struct zbookmark_phys { uint64_t zb_objset; uint64_t zb_object; int64_t zb_level; uint64_t zb_blkid; }; #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; dmu_object_type_t zp_type; uint8_t zp_level; uint8_t zp_copies; boolean_t zp_dedup; boolean_t zp_dedup_verify; boolean_t zp_nopwrite; boolean_t zp_encrypt; boolean_t zp_byteorder; uint8_t zp_salt[ZIO_DATA_SALT_LEN]; uint8_t zp_iv[ZIO_DATA_IV_LEN]; uint8_t zp_mac[ZIO_DATA_MAC_LEN]; } 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_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 void zio_vsd_cksum_report_f(zio_t *zio, zio_cksum_report_t *zcr, void *arg); zio_vsd_cksum_report_f zio_vsd_default_cksum_report; typedef struct zio_vsd_ops { zio_done_func_t *vsd_free; zio_vsd_cksum_report_f *vsd_cksum_report; } zio_vsd_ops_t; typedef struct zio_gang_node { zio_gbh_phys_t *gn_gbh; struct zio_gang_node *gn_child[SPA_GBH_NBLKPTRS]; } 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 int zio_pipe_stage_t(zio_t *zio); /* * The io_reexecute flags 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_REEXECUTE_NOW 0x01 #define ZIO_REEXECUTE_SUSPEND 0x02 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; 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; int io_cmd; zio_priority_t io_priority; uint8_t io_reexecute; 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_physdone; 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; 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). */ avl_node_t io_queue_node; avl_node_t io_offset_node; avl_node_t io_alloc_node; zio_alloc_list_t io_alloc_list; /* Internal pipeline state */ enum zio_flag io_flags; enum zio_stage io_stage; enum zio_stage io_pipeline; enum zio_flag 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_child_count; uint64_t io_phys_children; uint64_t io_parent_count; uint64_t *io_stall; zio_t *io_gang_leader; zio_gang_node_t *io_gang_tree; void *io_executor; void *io_waiter; kmutex_t io_lock; kcondvar_t io_cv; /* FMA state */ zio_cksum_report_t *io_cksum_report; uint64_t io_ena; /* Taskq dispatching state */ taskq_ent_t io_tqent; }; 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 *private, enum zio_flag flags); extern zio_t *zio_root(spa_t *spa, zio_done_func_t *done, void *private, enum zio_flag flags); 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 *private, zio_priority_t priority, enum zio_flag 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 *physdone, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag 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 *private, zio_priority_t priority, enum zio_flag flags, zbookmark_phys_t *zb); extern void zio_write_override(zio_t *zio, blkptr_t *bp, int copies, boolean_t nopwrite); 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 *private, enum zio_flag flags); extern zio_t *zio_ioctl(zio_t *pio, spa_t *spa, vdev_t *vd, int cmd, zio_done_func_t *done, void *private, enum zio_flag flags); 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 *private, zio_priority_t priority, enum zio_flag 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 *private, zio_priority_t priority, enum zio_flag flags, boolean_t labels); extern zio_t *zio_free_sync(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, enum zio_flag 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_free_zil(spa_t *spa, uint64_t txg, blkptr_t *bp); extern void zio_flush(zio_t *zio, vdev_t *vd); extern void zio_shrink(zio_t *zio, uint64_t size); extern int zio_wait(zio_t *zio); extern void zio_nowait(zio_t *zio); extern void zio_execute(zio_t *zio); extern void zio_interrupt(zio_t *zio); extern void zio_delay_init(zio_t *zio); extern void zio_delay_interrupt(zio_t *zio); 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, enum zio_flag flags, zio_done_func_t *done, void *private); extern zio_t *zio_vdev_delegated_io(vdev_t *vd, uint64_t offset, struct abd *data, uint64_t size, int type, zio_priority_t priority, enum zio_flag flags, zio_done_func_t *done, void *private); 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_checksum_verified(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 void zio_cancel(zio_t *zio); extern void zio_suspend(spa_t *spa, zio_t *zio); extern int zio_resume(spa_t *spa); extern void zio_resume_wait(spa_t *spa); /* * 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, char *tag, uint64_t type); 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); /* * Checksum ereport functions */ extern void zfs_ereport_start_checksum(spa_t *spa, vdev_t *vd, zbookmark_phys_t *zb, struct zio *zio, uint64_t offset, uint64_t length, void *arg, 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 void zfs_ereport_post_checksum(spa_t *spa, vdev_t *vd, 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); /* 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); 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/lib/libspl/include/assert.h b/lib/libspl/include/assert.h index 026d6fef9df1..c080bc8aa5bd 100644 --- a/lib/libspl/include/assert.h +++ b/lib/libspl/include/assert.h @@ -1,125 +1,128 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License, Version 1.0 only * (the "License"). You may not use this file except in compliance * with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2007 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #include_next #ifndef _LIBSPL_ASSERT_H #define _LIBSPL_ASSERT_H #include #include #include static inline int libspl_assert(const char *buf, const char *file, const char *func, int line) { fprintf(stderr, "%s\n", buf); fprintf(stderr, "ASSERT at %s:%d:%s()", file, line, func); abort(); } /* printf version of libspl_assert */ static inline void libspl_assertf(const char *file, const char *func, int line, char *format, ...) { va_list args; va_start(args, format); vfprintf(stderr, format, args); fprintf(stderr, "\n"); fprintf(stderr, "ASSERT at %s:%d:%s()", file, line, func); va_end(args); abort(); } #ifdef verify #undef verify #endif #define VERIFY(cond) \ (void) ((!(cond)) && \ libspl_assert(#cond, __FILE__, __FUNCTION__, __LINE__)) #define verify(cond) \ (void) ((!(cond)) && \ libspl_assert(#cond, __FILE__, __FUNCTION__, __LINE__)) #define VERIFY3_IMPL(LEFT, OP, RIGHT, TYPE) \ do { \ const TYPE __left = (TYPE)(LEFT); \ const TYPE __right = (TYPE)(RIGHT); \ if (!(__left OP __right)) \ libspl_assertf(__FILE__, __FUNCTION__, __LINE__, \ "%s %s %s (0x%llx %s 0x%llx)", #LEFT, #OP, #RIGHT, \ (u_longlong_t)__left, #OP, (u_longlong_t)__right); \ } while (0) +#define VERIFY3B(x, y, z) VERIFY3_IMPL(x, y, z, boolean_t) #define VERIFY3S(x, y, z) VERIFY3_IMPL(x, y, z, int64_t) #define VERIFY3U(x, y, z) VERIFY3_IMPL(x, y, z, uint64_t) #define VERIFY3P(x, y, z) VERIFY3_IMPL(x, y, z, uintptr_t) #define VERIFY0(x) VERIFY3_IMPL(x, ==, 0, uint64_t) #ifdef assert #undef assert #endif /* Compile time assert */ #define CTASSERT_GLOBAL(x) _CTASSERT(x, __LINE__) #define CTASSERT(x) { _CTASSERT(x, __LINE__); } #define _CTASSERT(x, y) __CTASSERT(x, y) #define __CTASSERT(x, y) \ typedef char __attribute__((unused)) \ __compile_time_assertion__ ## y[(x) ? 1 : -1] #ifdef NDEBUG +#define ASSERT3B(x, y, z) ((void)0) #define ASSERT3S(x, y, z) ((void)0) #define ASSERT3U(x, y, z) ((void)0) #define ASSERT3P(x, y, z) ((void)0) #define ASSERT0(x) ((void)0) #define ASSERT(x) ((void)0) #define assert(x) ((void)0) #define ASSERTV(x) #define IMPLY(A, B) ((void)0) #define EQUIV(A, B) ((void)0) #else +#define ASSERT3B(x, y, z) VERIFY3B(x, y, z) #define ASSERT3S(x, y, z) VERIFY3S(x, y, z) #define ASSERT3U(x, y, z) VERIFY3U(x, y, z) #define ASSERT3P(x, y, z) VERIFY3P(x, y, z) #define ASSERT0(x) VERIFY0(x) #define ASSERT(x) VERIFY(x) #define assert(x) VERIFY(x) #define ASSERTV(x) x #define IMPLY(A, B) \ ((void)(((!(A)) || (B)) || \ libspl_assert("(" #A ") implies (" #B ")", \ __FILE__, __FUNCTION__, __LINE__))) #define EQUIV(A, B) \ ((void)((!!(A) == !!(B)) || \ libspl_assert("(" #A ") is equivalent to (" #B ")", \ __FILE__, __FUNCTION__, __LINE__))) #endif /* NDEBUG */ #endif /* _LIBSPL_ASSERT_H */ diff --git a/lib/libspl/include/sys/time.h b/lib/libspl/include/sys/time.h index f05fcaa1c663..b3dd74dd5715 100644 --- a/lib/libspl/include/sys/time.h +++ b/lib/libspl/include/sys/time.h @@ -1,78 +1,86 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License, Version 1.0 only * (the "License"). You may not use this file except in compliance * with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2006 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #ifndef _LIBSPL_SYS_TIME_H #define _LIBSPL_SYS_TIME_H #include_next #include #ifndef SEC #define SEC 1 #endif #ifndef MILLISEC #define MILLISEC 1000 #endif #ifndef MICROSEC #define MICROSEC 1000000 #endif #ifndef NANOSEC #define NANOSEC 1000000000 #endif #ifndef NSEC_PER_USEC #define NSEC_PER_USEC 1000L #endif #ifndef MSEC2NSEC -#define MSEC2NSEC(m) ((hrtime_t)(m) * (NANOSEC / MILLISEC)) +#define MSEC2NSEC(m) ((hrtime_t)(m) * (NANOSEC / MILLISEC)) #endif #ifndef NSEC2MSEC -#define NSEC2MSEC(n) ((n) / (NANOSEC / MILLISEC)) +#define NSEC2MSEC(n) ((n) / (NANOSEC / MILLISEC)) #endif -#ifndef NSEC2SEC +#ifndef USEC2NSEC +#define USEC2NSEC(m) ((hrtime_t)(m) * (NANOSEC / MICROSEC)) +#endif + +#ifndef NSEC2USEC +#define NSEC2USEC(n) ((n) / (NANOSEC / MICROSEC)) +#endif + +#ifndef NSEC2SEC #define NSEC2SEC(n) ((n) / (NANOSEC / SEC)) #endif #ifndef SEC2NSEC #define SEC2NSEC(m) ((hrtime_t)(m) * (NANOSEC / SEC)) #endif typedef long long hrtime_t; typedef struct timespec timestruc_t; typedef struct timespec timespec_t; extern hrtime_t gethrtime(void); extern void gethrestime(timestruc_t *); #endif /* _LIBSPL_SYS_TIME_H */ diff --git a/module/zfs/dmu.c b/module/zfs/dmu.c index 108dfe15736d..56740ae375a8 100644 --- a/module/zfs/dmu.c +++ b/module/zfs/dmu.c @@ -1,2512 +1,2519 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2017 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. */ #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. */ int zfs_nopwrite_enabled = 1; /* * Tunable to control percentage of dirtied blocks from frees in one TXG. * After this threshold is crossed, additional dirty blocks from frees * wait until the next TXG. * A value of zero will disable this throttle. */ unsigned long zfs_per_txg_dirty_frees_percent = 30; /* * Enable/disable forcing txg sync when dirty in dmu_offset_next. */ int zfs_dmu_offset_next_sync = 0; const dmu_object_type_info_t dmu_ot[DMU_OT_NUMTYPES] = { { DMU_BSWAP_UINT8, TRUE, FALSE, "unallocated" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "object directory" }, { DMU_BSWAP_UINT64, TRUE, FALSE, "object array" }, { DMU_BSWAP_UINT8, TRUE, FALSE, "packed nvlist" }, { DMU_BSWAP_UINT64, TRUE, FALSE, "packed nvlist size" }, { DMU_BSWAP_UINT64, TRUE, FALSE, "bpobj" }, { DMU_BSWAP_UINT64, TRUE, FALSE, "bpobj header" }, { DMU_BSWAP_UINT64, TRUE, FALSE, "SPA space map header" }, { DMU_BSWAP_UINT64, TRUE, FALSE, "SPA space map" }, { DMU_BSWAP_UINT64, TRUE, TRUE, "ZIL intent log" }, { DMU_BSWAP_DNODE, TRUE, TRUE, "DMU dnode" }, { DMU_BSWAP_OBJSET, TRUE, FALSE, "DMU objset" }, { DMU_BSWAP_UINT64, TRUE, FALSE, "DSL directory" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "DSL directory child map"}, { DMU_BSWAP_ZAP, TRUE, FALSE, "DSL dataset snap map" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "DSL props" }, { DMU_BSWAP_UINT64, TRUE, FALSE, "DSL dataset" }, { DMU_BSWAP_ZNODE, TRUE, FALSE, "ZFS znode" }, { DMU_BSWAP_OLDACL, TRUE, TRUE, "ZFS V0 ACL" }, { DMU_BSWAP_UINT8, FALSE, TRUE, "ZFS plain file" }, { DMU_BSWAP_ZAP, TRUE, TRUE, "ZFS directory" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "ZFS master node" }, { DMU_BSWAP_ZAP, TRUE, TRUE, "ZFS delete queue" }, { DMU_BSWAP_UINT8, FALSE, TRUE, "zvol object" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "zvol prop" }, { DMU_BSWAP_UINT8, FALSE, TRUE, "other uint8[]" }, { DMU_BSWAP_UINT64, FALSE, TRUE, "other uint64[]" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "other ZAP" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "persistent error log" }, { DMU_BSWAP_UINT8, TRUE, FALSE, "SPA history" }, { DMU_BSWAP_UINT64, TRUE, FALSE, "SPA history offsets" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "Pool properties" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "DSL permissions" }, { DMU_BSWAP_ACL, TRUE, TRUE, "ZFS ACL" }, { DMU_BSWAP_UINT8, TRUE, TRUE, "ZFS SYSACL" }, { DMU_BSWAP_UINT8, TRUE, TRUE, "FUID table" }, { DMU_BSWAP_UINT64, TRUE, FALSE, "FUID table size" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "DSL dataset next clones"}, { DMU_BSWAP_ZAP, TRUE, FALSE, "scan work queue" }, { DMU_BSWAP_ZAP, TRUE, TRUE, "ZFS user/group used" }, { DMU_BSWAP_ZAP, TRUE, TRUE, "ZFS user/group quota" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "snapshot refcount tags"}, { DMU_BSWAP_ZAP, TRUE, FALSE, "DDT ZAP algorithm" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "DDT statistics" }, { DMU_BSWAP_UINT8, TRUE, TRUE, "System attributes" }, { DMU_BSWAP_ZAP, TRUE, TRUE, "SA master node" }, { DMU_BSWAP_ZAP, TRUE, TRUE, "SA attr registration" }, { DMU_BSWAP_ZAP, TRUE, TRUE, "SA attr layouts" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "scan translations" }, { DMU_BSWAP_UINT8, FALSE, TRUE, "deduplicated block" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "DSL deadlist map" }, { DMU_BSWAP_UINT64, TRUE, FALSE, "DSL deadlist map hdr" }, { DMU_BSWAP_ZAP, TRUE, FALSE, "DSL dir clones" }, { DMU_BSWAP_UINT64, TRUE, FALSE, "bpobj subobj" } }; const 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, void *tag, dmu_buf_t **dbp) { uint64_t blkid; dmu_buf_impl_t *db; blkid = dbuf_whichblock(dn, 0, offset); rw_enter(&dn->dn_struct_rwlock, RW_READER); 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, 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); blkid = dbuf_whichblock(dn, 0, offset); rw_enter(&dn->dn_struct_rwlock, RW_READER); 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, void *tag, dmu_buf_t **dbp, int flags) { int err; int db_flags = DB_RF_CANFAIL; if (flags & DMU_READ_NO_PREFETCH) db_flags |= DB_RF_NOPREFETCH; if (flags & DMU_READ_NO_DECRYPT) db_flags |= DB_RF_NO_DECRYPT; 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, db_flags); if (err != 0) { dbuf_rele(db, tag); *dbp = NULL; } } return (err); } int dmu_buf_hold(objset_t *os, uint64_t object, uint64_t offset, void *tag, dmu_buf_t **dbp, int flags) { int err; int db_flags = DB_RF_CANFAIL; if (flags & DMU_READ_NO_PREFETCH) db_flags |= DB_RF_NOPREFETCH; if (flags & DMU_READ_NO_DECRYPT) db_flags |= DB_RF_NO_DECRYPT; 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, db_flags); 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; DB_DNODE_ENTER(db); dn = DB_DNODE(db); if (dn->dn_bonus != db) { error = SET_ERROR(EINVAL); } else if (newsize < 0 || newsize > db_fake->db_size) { 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; DB_DNODE_ENTER(db); dn = DB_DNODE(db); if (!DMU_OT_IS_VALID(type)) { error = SET_ERROR(EINVAL); } else 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; dnode_t *dn; dmu_object_type_t type; DB_DNODE_ENTER(db); dn = DB_DNODE(db); type = dn->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); } /* * returns ENOENT, EIO, or 0. */ int dmu_bonus_hold_impl(objset_t *os, uint64_t object, void *tag, uint32_t flags, dmu_buf_t **dbp) { dnode_t *dn; dmu_buf_impl_t *db; int error; uint32_t db_flags = DB_RF_MUST_SUCCEED; if (flags & DMU_READ_NO_PREFETCH) db_flags |= DB_RF_NOPREFETCH; if (flags & DMU_READ_NO_DECRYPT) db_flags |= DB_RF_NO_DECRYPT; error = dnode_hold(os, object, FTAG, &dn); if (error) return (error); rw_enter(&dn->dn_struct_rwlock, RW_READER); if (dn->dn_bonus == NULL) { 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 (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); dnode_rele(dn, FTAG); error = dbuf_read(db, NULL, db_flags); 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 obj, void *tag, dmu_buf_t **dbp) { return (dmu_bonus_hold_impl(os, obj, tag, DMU_READ_NO_PREFETCH, dbp)); } /* * 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, uint32_t flags, 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, 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, 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); err = dmu_spill_hold_by_dnode(dn, DB_RF_CANFAIL, 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. */ static int dmu_buf_hold_array_by_dnode(dnode_t *dn, uint64_t offset, uint64_t length, boolean_t read, void *tag, int *numbufsp, dmu_buf_t ***dbpp, uint32_t flags) { dmu_buf_t **dbp; uint64_t blkid, nblks, i; uint32_t dbuf_flags; int err; zio_t *zio; ASSERT(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 = DB_RF_CANFAIL | DB_RF_NEVERWAIT | DB_RF_HAVESTRUCT | DB_RF_NOPREFETCH; rw_enter(&dn->dn_struct_rwlock, RW_READER); if (dn->dn_datablkshift) { int blkshift = dn->dn_datablkshift; nblks = (P2ROUNDUP(offset + length, 1ULL << blkshift) - P2ALIGN(offset, 1ULL << blkshift)) >> 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); zio = zio_root(dn->dn_objset->os_spa, NULL, NULL, ZIO_FLAG_CANFAIL); blkid = dbuf_whichblock(dn, 0, offset); for (i = 0; i < nblks; i++) { dmu_buf_impl_t *db = dbuf_hold(dn, blkid + i, tag); if (db == NULL) { rw_exit(&dn->dn_struct_rwlock); dmu_buf_rele_array(dbp, nblks, tag); zio_nowait(zio); return (SET_ERROR(EIO)); } /* initiate async i/o */ if (read) (void) dbuf_read(db, zio, dbuf_flags); dbp[i] = &db->db; } if ((flags & DMU_READ_NO_PREFETCH) == 0 && DNODE_META_IS_CACHEABLE(dn) && length <= zfetch_array_rd_sz) { dmu_zfetch(&dn->dn_zfetch, blkid, nblks, read && DNODE_IS_CACHEABLE(dn)); } rw_exit(&dn->dn_struct_rwlock); /* wait for async i/o */ err = zio_wait(zio); if (err) { dmu_buf_rele_array(dbp, nblks, tag); return (err); } /* wait for other io to complete */ if (read) { 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); } static int dmu_buf_hold_array(objset_t *os, uint64_t object, uint64_t offset, uint64_t length, int read, 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, void *tag, int *numbufsp, dmu_buf_t ***dbpp) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dnode_t *dn; int err; DB_DNODE_ENTER(db); dn = DB_DNODE(db); err = dmu_buf_hold_array_by_dnode(dn, 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, 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 prefeteched will be those that point to the blocks containing * the data starting at offset, and continuing to offset + len. * * Note that if the indirect blocks above the blocks being prefetched are not * in cache, they will be asychronously read in. */ 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; uint64_t blkid; int nblks, err; if (len == 0) { /* they're interested in the bonus buffer */ dn = DMU_META_DNODE(os); if (object == 0 || object >= DN_MAX_OBJECT) return; rw_enter(&dn->dn_struct_rwlock, RW_READER); blkid = dbuf_whichblock(dn, level, object * sizeof (dnode_phys_t)); dbuf_prefetch(dn, level, blkid, pri, 0); rw_exit(&dn->dn_struct_rwlock); return; } /* * XXX - Note, if the dnode for the requested object is not * already cached, we will do a *synchronous* read in the * dnode_hold() call. The same is true for any indirects. */ err = dnode_hold(os, object, FTAG, &dn); if (err != 0) return; rw_enter(&dn->dn_struct_rwlock, RW_READER); /* * offset + len - 1 is the last byte we want to prefetch for, and offset * is the first. Then dbuf_whichblk(dn, level, off + len - 1) is the * last block we want to prefetch, and dbuf_whichblock(dn, level, * offset) is the first. Then the number we need to prefetch is the * last - first + 1. */ if (level > 0 || dn->dn_datablkshift != 0) { nblks = dbuf_whichblock(dn, level, offset + len - 1) - dbuf_whichblock(dn, level, offset) + 1; } else { nblks = (offset < dn->dn_datablksz); } if (nblks != 0) { blkid = dbuf_whichblock(dn, level, offset); for (int i = 0; i < nblks; i++) dbuf_prefetch(dn, level, blkid + i, pri, 0); } rw_exit(&dn->dn_struct_rwlock); dnode_rele(dn, FTAG); } /* * 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 crashes 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. */ static int get_next_chunk(dnode_t *dn, uint64_t *start, uint64_t minimum) { uint64_t maxblks = DMU_MAX_ACCESS >> (dn->dn_indblkshift + 1); /* bytes of data covered by a level-1 indirect block */ uint64_t iblkrange = dn->dn_datablksz * EPB(dn->dn_indblkshift, SPA_BLKPTRSHIFT); ASSERT3U(minimum, <=, *start); if (*start - minimum <= iblkrange * maxblks) { *start = minimum; return (0); } ASSERT(ISP2(iblkrange)); for (uint64_t 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) { return (err); } /* set start to the beginning of this L1 indirect */ *start = P2ALIGN(*start, iblkrange); } if (*start < minimum) *start = minimum; 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 */ /*ARGSUSED*/ 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)); #endif return (B_FALSE); } static int dmu_free_long_range_impl(objset_t *os, dnode_t *dn, uint64_t offset, uint64_t length, boolean_t raw) { 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 / 4; 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 long_free_dirty_all_txgs = 0; 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); if (err) return (err); ASSERT3U(chunk_begin, >=, offset); ASSERT3U(chunk_begin, <=, chunk_end); chunk_len = chunk_end - chunk_begin; mutex_enter(&dp->dp_lock); for (int t = 0; t < TXG_SIZE; t++) { long_free_dirty_all_txgs += dp->dp_long_free_dirty_pertxg[t]; } 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_all_txgs >= dirty_frees_threshold) { txg_wait_open(dp, 0); continue; } 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, TXG_WAIT); if (err) { dmu_tx_abort(tx); return (err); } mutex_enter(&dp->dp_lock); dp->dp_long_free_dirty_pertxg[dmu_tx_get_txg(tx) & TXG_MASK] += chunk_len; mutex_exit(&dp->dp_lock); DTRACE_PROBE3(free__long__range, uint64_t, long_free_dirty_all_txgs, uint64_t, chunk_len, uint64_t, dmu_tx_get_txg(tx)); dnode_free_range(dn, chunk_begin, chunk_len, tx); /* if this is a raw free, mark the dirty record as such */ if (raw) { dbuf_dirty_record_t *dr = dn->dn_dbuf->db_last_dirty; while (dr != NULL && dr->dr_txg > tx->tx_txg) dr = dr->dr_next; if (dr != NULL && dr->dr_txg == tx->tx_txg) dr->dt.dl.dr_raw = B_TRUE; } 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, B_FALSE); /* * 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); } /* * This function is equivalent to dmu_free_long_range(), but also * marks the new dirty record as a raw write. */ int dmu_free_long_range_raw(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, B_TRUE); /* * 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); } static int dmu_free_long_object_impl(objset_t *os, uint64_t object, boolean_t raw) { 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, TXG_WAIT); if (err == 0) { if (raw) err = dmu_object_dirty_raw(os, object, tx); if (err == 0) err = dmu_object_free(os, object, tx); dmu_tx_commit(tx); } else { dmu_tx_abort(tx); } return (err); } int dmu_free_long_object(objset_t *os, uint64_t object) { return (dmu_free_long_object_impl(os, object, B_FALSE)); } int dmu_free_long_object_raw(objset_t *os, uint64_t object) { return (dmu_free_long_object_impl(os, object, B_TRUE)); } 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, uint32_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); bzero((char *)buf + newsz, size - newsz); size = newsz; } 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); (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, uint32_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, uint32_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) { 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(db, tx); else dmu_buf_will_dirty(db, tx); (void) memcpy((char *)db->db_data + bufoff, buf, tocpy); if (tocpy == db->db_size) dmu_buf_fill_done(db, tx); 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_buf_rele_array(dbp, numbufs, FTAG); } void dmu_write_by_dnode(dnode_t *dn, 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_by_dnode(dn, offset, size, FALSE, FTAG, &numbufs, &dbp, DMU_READ_PREFETCH)); dmu_write_impl(dbp, numbufs, offset, size, buf, tx); dmu_buf_rele_array(dbp, numbufs, FTAG); } 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); } /* * DMU support for xuio */ kstat_t *xuio_ksp = NULL; typedef struct xuio_stats { /* loaned yet not returned arc_buf */ kstat_named_t xuiostat_onloan_rbuf; kstat_named_t xuiostat_onloan_wbuf; /* whether a copy is made when loaning out a read buffer */ kstat_named_t xuiostat_rbuf_copied; kstat_named_t xuiostat_rbuf_nocopy; /* whether a copy is made when assigning a write buffer */ kstat_named_t xuiostat_wbuf_copied; kstat_named_t xuiostat_wbuf_nocopy; } xuio_stats_t; static xuio_stats_t xuio_stats = { { "onloan_read_buf", KSTAT_DATA_UINT64 }, { "onloan_write_buf", KSTAT_DATA_UINT64 }, { "read_buf_copied", KSTAT_DATA_UINT64 }, { "read_buf_nocopy", KSTAT_DATA_UINT64 }, { "write_buf_copied", KSTAT_DATA_UINT64 }, { "write_buf_nocopy", KSTAT_DATA_UINT64 } }; #define XUIOSTAT_INCR(stat, val) \ atomic_add_64(&xuio_stats.stat.value.ui64, (val)) #define XUIOSTAT_BUMP(stat) XUIOSTAT_INCR(stat, 1) #ifdef HAVE_UIO_ZEROCOPY int dmu_xuio_init(xuio_t *xuio, int nblk) { dmu_xuio_t *priv; uio_t *uio = &xuio->xu_uio; uio->uio_iovcnt = nblk; uio->uio_iov = kmem_zalloc(nblk * sizeof (iovec_t), KM_SLEEP); priv = kmem_zalloc(sizeof (dmu_xuio_t), KM_SLEEP); priv->cnt = nblk; priv->bufs = kmem_zalloc(nblk * sizeof (arc_buf_t *), KM_SLEEP); priv->iovp = (iovec_t *)uio->uio_iov; XUIO_XUZC_PRIV(xuio) = priv; if (XUIO_XUZC_RW(xuio) == UIO_READ) XUIOSTAT_INCR(xuiostat_onloan_rbuf, nblk); else XUIOSTAT_INCR(xuiostat_onloan_wbuf, nblk); return (0); } void dmu_xuio_fini(xuio_t *xuio) { dmu_xuio_t *priv = XUIO_XUZC_PRIV(xuio); int nblk = priv->cnt; kmem_free(priv->iovp, nblk * sizeof (iovec_t)); kmem_free(priv->bufs, nblk * sizeof (arc_buf_t *)); kmem_free(priv, sizeof (dmu_xuio_t)); if (XUIO_XUZC_RW(xuio) == UIO_READ) XUIOSTAT_INCR(xuiostat_onloan_rbuf, -nblk); else XUIOSTAT_INCR(xuiostat_onloan_wbuf, -nblk); } /* * Initialize iov[priv->next] and priv->bufs[priv->next] with { off, n, abuf } * and increase priv->next by 1. */ int dmu_xuio_add(xuio_t *xuio, arc_buf_t *abuf, offset_t off, size_t n) { struct iovec *iov; uio_t *uio = &xuio->xu_uio; dmu_xuio_t *priv = XUIO_XUZC_PRIV(xuio); int i = priv->next++; ASSERT(i < priv->cnt); ASSERT(off + n <= arc_buf_lsize(abuf)); iov = (iovec_t *)uio->uio_iov + i; iov->iov_base = (char *)abuf->b_data + off; iov->iov_len = n; priv->bufs[i] = abuf; return (0); } int dmu_xuio_cnt(xuio_t *xuio) { dmu_xuio_t *priv = XUIO_XUZC_PRIV(xuio); return (priv->cnt); } arc_buf_t * dmu_xuio_arcbuf(xuio_t *xuio, int i) { dmu_xuio_t *priv = XUIO_XUZC_PRIV(xuio); ASSERT(i < priv->cnt); return (priv->bufs[i]); } void dmu_xuio_clear(xuio_t *xuio, int i) { dmu_xuio_t *priv = XUIO_XUZC_PRIV(xuio); ASSERT(i < priv->cnt); priv->bufs[i] = NULL; } #endif /* HAVE_UIO_ZEROCOPY */ static void xuio_stat_init(void) { xuio_ksp = kstat_create("zfs", 0, "xuio_stats", "misc", KSTAT_TYPE_NAMED, sizeof (xuio_stats) / sizeof (kstat_named_t), KSTAT_FLAG_VIRTUAL); if (xuio_ksp != NULL) { xuio_ksp->ks_data = &xuio_stats; kstat_install(xuio_ksp); } } static void xuio_stat_fini(void) { if (xuio_ksp != NULL) { kstat_delete(xuio_ksp); xuio_ksp = NULL; } } void xuio_stat_wbuf_copied(void) { XUIOSTAT_BUMP(xuiostat_wbuf_copied); } void xuio_stat_wbuf_nocopy(void) { XUIOSTAT_BUMP(xuiostat_wbuf_nocopy); } #ifdef _KERNEL int dmu_read_uio_dnode(dnode_t *dn, uio_t *uio, uint64_t size) { dmu_buf_t **dbp; int numbufs, i, err; #ifdef HAVE_UIO_ZEROCOPY xuio_t *xuio = NULL; #endif /* * 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, uio->uio_loffset, size, TRUE, FTAG, &numbufs, &dbp, 0); 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 = uio->uio_loffset - db->db_offset; tocpy = MIN(db->db_size - bufoff, size); #ifdef HAVE_UIO_ZEROCOPY if (xuio) { dmu_buf_impl_t *dbi = (dmu_buf_impl_t *)db; arc_buf_t *dbuf_abuf = dbi->db_buf; arc_buf_t *abuf = dbuf_loan_arcbuf(dbi); err = dmu_xuio_add(xuio, abuf, bufoff, tocpy); if (!err) { uio->uio_resid -= tocpy; uio->uio_loffset += tocpy; } if (abuf == dbuf_abuf) XUIOSTAT_BUMP(xuiostat_rbuf_nocopy); else XUIOSTAT_BUMP(xuiostat_rbuf_copied); } else #endif err = uiomove((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 offset uio->uio_loffset. * * 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, uio_t *uio, uint64_t size) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)zdb; dnode_t *dn; int err; if (size == 0) return (0); DB_DNODE_ENTER(db); dn = DB_DNODE(db); err = dmu_read_uio_dnode(dn, uio, size); DB_DNODE_EXIT(db); return (err); } /* * Read 'size' bytes into the uio buffer. * From the specified object * Starting at offset uio->uio_loffset. */ int dmu_read_uio(objset_t *os, uint64_t object, uio_t *uio, uint64_t size) { 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); dnode_rele(dn, FTAG); return (err); } int dmu_write_uio_dnode(dnode_t *dn, uio_t *uio, uint64_t size, dmu_tx_t *tx) { dmu_buf_t **dbp; int numbufs; int err = 0; int i; err = dmu_buf_hold_array_by_dnode(dn, uio->uio_loffset, size, FALSE, FTAG, &numbufs, &dbp, DMU_READ_PREFETCH); 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 = uio->uio_loffset - 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(db, tx); else dmu_buf_will_dirty(db, tx); /* * XXX uiomove could block forever (eg.nfs-backed * pages). There needs to be a uiolockdown() function * to lock the pages in memory, so that uiomove won't * block. */ err = uiomove((char *)db->db_data + bufoff, tocpy, UIO_WRITE, uio); if (tocpy == db->db_size) dmu_buf_fill_done(db, tx); if (err) break; size -= tocpy; } dmu_buf_rele_array(dbp, numbufs, FTAG); return (err); } /* * Write 'size' bytes from the uio buffer. * To object zdb->db_object. * Starting at offset uio->uio_loffset. * * 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, uio_t *uio, uint64_t size, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)zdb; dnode_t *dn; int err; if (size == 0) return (0); DB_DNODE_ENTER(db); dn = DB_DNODE(db); err = dmu_write_uio_dnode(dn, uio, size, tx); DB_DNODE_EXIT(db); return (err); } /* * Write 'size' bytes from the uio buffer. * To the specified object. * Starting at offset uio->uio_loffset. */ int dmu_write_uio(objset_t *os, uint64_t object, uio_t *uio, uint64_t size, dmu_tx_t *tx) { 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); dnode_rele(dn, FTAG); return (err); } #endif /* _KERNEL */ /* * 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); } void dmu_convert_to_raw(dmu_buf_t *handle, boolean_t byteorder, const uint8_t *salt, const uint8_t *iv, const uint8_t *mac, dmu_tx_t *tx) { dmu_object_type_t type; dmu_buf_impl_t *db = (dmu_buf_impl_t *)handle; uint64_t dsobj = dmu_objset_id(db->db_objset); ASSERT3P(db->db_buf, !=, NULL); ASSERT3U(dsobj, !=, 0); dmu_buf_will_change_crypt_params(handle, tx); DB_DNODE_ENTER(db); type = DB_DNODE(db)->dn_type; DB_DNODE_EXIT(db); /* * This technically violates the assumption the dmu code makes * that dnode blocks are only released in syncing context. */ (void) arc_release(db->db_buf, db); arc_convert_to_raw(db->db_buf, dsobj, byteorder, type, salt, iv, mac); } void dmu_copy_from_buf(objset_t *os, uint64_t object, uint64_t offset, dmu_buf_t *handle, dmu_tx_t *tx) { dmu_buf_t *dst_handle; dmu_buf_impl_t *dstdb; dmu_buf_impl_t *srcdb = (dmu_buf_impl_t *)handle; arc_buf_t *abuf; uint64_t datalen; boolean_t byteorder; uint8_t salt[ZIO_DATA_SALT_LEN]; uint8_t iv[ZIO_DATA_IV_LEN]; uint8_t mac[ZIO_DATA_MAC_LEN]; ASSERT3P(srcdb->db_buf, !=, NULL); /* hold the db that we want to write to */ VERIFY0(dmu_buf_hold(os, object, offset, FTAG, &dst_handle, DMU_READ_NO_DECRYPT)); dstdb = (dmu_buf_impl_t *)dst_handle; datalen = arc_buf_size(srcdb->db_buf); /* allocated an arc buffer that matches the type of srcdb->db_buf */ if (arc_is_encrypted(srcdb->db_buf)) { arc_get_raw_params(srcdb->db_buf, &byteorder, salt, iv, mac); abuf = arc_loan_raw_buf(os->os_spa, dmu_objset_id(os), byteorder, salt, iv, mac, DB_DNODE(dstdb)->dn_type, datalen, arc_buf_lsize(srcdb->db_buf), arc_get_compression(srcdb->db_buf)); } else { /* we won't get a compressed db back from dmu_buf_hold() */ ASSERT3U(arc_get_compression(srcdb->db_buf), ==, ZIO_COMPRESS_OFF); abuf = arc_loan_buf(os->os_spa, DMU_OT_IS_METADATA(DB_DNODE(dstdb)->dn_type), datalen); } ASSERT3U(datalen, ==, arc_buf_size(abuf)); /* copy the data to the new buffer and assign it to the dstdb */ bcopy(srcdb->db_buf->b_data, abuf->b_data, datalen); dbuf_assign_arcbuf(dstdb, abuf, tx); dmu_buf_rele(dst_handle, FTAG); } /* * 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(). */ void dmu_assign_arcbuf_by_dnode(dnode_t *dn, uint64_t offset, arc_buf_t *buf, dmu_tx_t *tx) { dmu_buf_impl_t *db; objset_t *os = dn->dn_objset; uint64_t object = dn->dn_object; 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); VERIFY((db = dbuf_hold(dn, blkid, FTAG)) != NULL); rw_exit(&dn->dn_struct_rwlock); /* * We can only assign if the offset is aligned, the arc buf is the * same size as the dbuf, and the dbuf is not metadata. */ if (offset == db->db.db_offset && blksz == db->db.db_size) { dbuf_assign_arcbuf(db, buf, tx); 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(os, object, offset, blksz, buf->b_data, tx); dmu_return_arcbuf(buf); XUIOSTAT_BUMP(xuiostat_wbuf_copied); } } void dmu_assign_arcbuf_by_dbuf(dmu_buf_t *handle, uint64_t offset, arc_buf_t *buf, dmu_tx_t *tx) { dmu_buf_impl_t *dbuf = (dmu_buf_impl_t *)handle; DB_DNODE_ENTER(dbuf); dmu_assign_arcbuf_by_dnode(DB_DNODE(dbuf), offset, buf, tx); DB_DNODE_EXIT(dbuf); } typedef struct { dbuf_dirty_record_t *dsa_dr; dmu_sync_cb_t *dsa_done; zgd_t *dsa_zgd; dmu_tx_t *dsa_tx; } dmu_sync_arg_t; /* ARGSUSED */ static void dmu_sync_ready(zio_t *zio, arc_buf_t *buf, void *varg) { dmu_sync_arg_t *dsa = varg; dmu_buf_t *db = dsa->dsa_zgd->zgd_db; blkptr_t *bp = zio->io_bp; if (zio->io_error == 0) { if (BP_IS_HOLE(bp)) { /* * 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); } /* ARGSUSED */ static void dmu_sync_done(zio_t *zio, arc_buf_t *buf, void *varg) { dmu_sync_arg_t *dsa = varg; dbuf_dirty_record_t *dr = dsa->dsa_dr; dmu_buf_impl_t *db = dr->dr_dbuf; mutex_enter(&db->db_mtx); ASSERT(dr->dt.dl.dr_override_state == DR_IN_DMU_SYNC); if (zio->io_error == 0) { 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; /* * 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) && dr->dt.dl.dr_overridden_by.blk_birth == 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); 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; ASSERTV(blkptr_t *bp_orig = &zio->io_bp_orig); if (zio->io_error == 0 && !BP_IS_HOLE(bp)) { ASSERT(!(zio->io_flags & ZIO_FLAG_NOPWRITE)); ASSERT(BP_IS_HOLE(bp_orig) || !BP_EQUAL(bp, bp_orig)); ASSERT(zio->io_bp->blk_birth == 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_put(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; tx = dmu_tx_create(os); dmu_tx_hold_space(tx, zgd->zgd_db->db_size); if (dmu_tx_assign(tx, TXG_WAIT) != 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, 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; dmu_sync_arg_t *dsa; zbookmark_phys_t zb; zio_prop_t zp; dnode_t *dn; ASSERT(pio != NULL); ASSERT(txg != 0); /* dbuf is within the locked range */ ASSERT3U(db->db.db_offset, >=, zgd->zgd_rl->r_off); ASSERT3U(db->db.db_offset + db->db.db_size, <=, zgd->zgd_rl->r_off + zgd->zgd_rl->r_len); SET_BOOKMARK(&zb, ds->ds_object, db->db.db_object, db->db_level, db->db_blkid); DB_DNODE_ENTER(db); dn = DB_DNODE(db); dmu_write_policy(os, dn, 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 = db->db_last_dirty; while (dr && dr->dr_txg != txg) dr = dr->dr_next; 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)); } ASSERT(dr->dr_next == NULL || dr->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. */ DB_DNODE_ENTER(db); dn = DB_DNODE(db); if (dr->dr_next != NULL || dnode_block_freed(dn, 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)); } 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_L2CACHEABLE(db), &zp, dmu_sync_ready, NULL, 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); } 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); } /* * Dirty an object and set the dirty record's raw flag. This is used * when writing raw data to an object that will not effect the * encryption parameters, specifically during raw receives. */ int dmu_object_dirty_raw(objset_t *os, uint64_t object, dmu_tx_t *tx) { dnode_t *dn; int err; err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); dmu_buf_will_change_crypt_params((dmu_buf_t *)dn->dn_dbuf, tx); dnode_rele(dn, FTAG); return (err); } int zfs_mdcomp_disable = 0; /* * When the "redundant_metadata" property is set to "most", only indirect * blocks of this level and higher will have an additional ditto block. */ 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; 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; /* * 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) { if (zfs_mdcomp_disable) { compress = ZIO_COMPRESS_EMPTY; } else { /* * 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; if (os->os_redundant_metadata == ZFS_REDUNDANT_METADATA_ALL || (os->os_redundant_metadata == ZFS_REDUNDANT_METADATA_MOST && (level >= zfs_redundant_metadata_most_ditto_level || DMU_OT_IS_METADATA(type) || (wp & WP_SPILL)))) 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); 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 checkum. 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); } /* * 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); nopwrite = B_FALSE; } else { dedup = B_FALSE; } if (type == DMU_OT_DNODE || type == DMU_OT_OBJSET) compress = ZIO_COMPRESS_EMPTY; } zp->zp_compress = compress; 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_dedup = dedup; zp->zp_dedup_verify = dedup && dedup_verify; zp->zp_nopwrite = nopwrite; zp->zp_encrypt = encrypt; zp->zp_byteorder = ZFS_HOST_BYTEORDER; bzero(zp->zp_salt, ZIO_DATA_SALT_LEN); bzero(zp->zp_iv, ZIO_DATA_IV_LEN); bzero(zp->zp_mac, ZIO_DATA_MAC_LEN); ASSERT3U(zp->zp_compress, !=, ZIO_COMPRESS_INHERIT); } /* * This function is only called from zfs_holey_common() for zpl_llseek() * in order to determine the location of holes. 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 which is always a safe thing to do. */ int dmu_offset_next(objset_t *os, uint64_t object, boolean_t hole, uint64_t *off) { dnode_t *dn; int i, err; boolean_t clean = B_TRUE; err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); /* * Check if dnode is dirty */ for (i = 0; i < TXG_SIZE; i++) { if (list_link_active(&dn->dn_dirty_link[i])) { clean = B_FALSE; break; } } /* * If compatibility option is on, sync any current changes before * we go trundling through the block pointers. */ if (!clean && zfs_dmu_offset_next_sync) { clean = B_TRUE; dnode_rele(dn, FTAG); txg_wait_synced(dmu_objset_pool(os), 0); err = dnode_hold(os, object, FTAG, &dn); if (err) return (err); } if (clean) err = dnode_next_offset(dn, (hole ? DNODE_FIND_HOLE : 0), off, 1, 1, 0); else err = SET_ERROR(EBUSY); dnode_rele(dn, FTAG); return (err); } 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. * This is specifically optimized for zfs_getattr(). */ 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; dnode_t *dn; DB_DNODE_ENTER(db); dn = DB_DNODE(db); *dnsize = dn->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]); } /* ARGSUSED */ void byteswap_uint8_array(void *vbuf, size_t size) { } void dmu_init(void) { abd_init(); zfs_dbgmsg_init(); sa_cache_init(); xuio_stat_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(); xuio_stat_fini(); sa_cache_fini(); zfs_dbgmsg_fini(); abd_fini(); } #if defined(_KERNEL) && defined(HAVE_SPL) EXPORT_SYMBOL(dmu_bonus_hold); EXPORT_SYMBOL(dmu_buf_hold_array_by_bonus); EXPORT_SYMBOL(dmu_buf_rele_array); EXPORT_SYMBOL(dmu_prefetch); EXPORT_SYMBOL(dmu_free_range); EXPORT_SYMBOL(dmu_free_long_range); EXPORT_SYMBOL(dmu_free_long_range_raw); EXPORT_SYMBOL(dmu_free_long_object); EXPORT_SYMBOL(dmu_free_long_object_raw); EXPORT_SYMBOL(dmu_read); EXPORT_SYMBOL(dmu_read_by_dnode); EXPORT_SYMBOL(dmu_write); EXPORT_SYMBOL(dmu_write_by_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_checksum); EXPORT_SYMBOL(dmu_object_set_compress); 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); /* BEGIN CSTYLED */ module_param(zfs_mdcomp_disable, int, 0644); MODULE_PARM_DESC(zfs_mdcomp_disable, "Disable meta data compression"); module_param(zfs_nopwrite_enabled, int, 0644); MODULE_PARM_DESC(zfs_nopwrite_enabled, "Enable NOP writes"); module_param(zfs_per_txg_dirty_frees_percent, ulong, 0644); MODULE_PARM_DESC(zfs_per_txg_dirty_frees_percent, "percentage of dirtied blocks from frees in one TXG"); module_param(zfs_dmu_offset_next_sync, int, 0644); MODULE_PARM_DESC(zfs_dmu_offset_next_sync, "Enable forcing txg sync to find holes"); /* END CSTYLED */ #endif diff --git a/module/zfs/txg.c b/module/zfs/txg.c index 078839a3203b..bf8544507b60 100644 --- a/module/zfs/txg.c +++ b/module/zfs/txg.c @@ -1,953 +1,953 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Portions Copyright 2011 Martin Matuska * Copyright (c) 2012, 2017 by Delphix. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include /* * ZFS Transaction Groups * ---------------------- * * ZFS transaction groups are, as the name implies, groups of transactions * that act on persistent state. ZFS asserts consistency at the granularity of * these transaction groups. Each successive transaction group (txg) is * assigned a 64-bit consecutive identifier. There are three active * transaction group states: open, quiescing, or syncing. At any given time, * there may be an active txg associated with each state; each active txg may * either be processing, or blocked waiting to enter the next state. There may * be up to three active txgs, and there is always a txg in the open state * (though it may be blocked waiting to enter the quiescing state). In broad * strokes, transactions -- operations that change in-memory structures -- are * accepted into the txg in the open state, and are completed while the txg is * in the open or quiescing states. The accumulated changes are written to * disk in the syncing state. * * Open * * When a new txg becomes active, it first enters the open state. New * transactions -- updates to in-memory structures -- are assigned to the * currently open txg. There is always a txg in the open state so that ZFS can * accept new changes (though the txg may refuse new changes if it has hit * some limit). ZFS advances the open txg to the next state for a variety of * reasons such as it hitting a time or size threshold, or the execution of an * administrative action that must be completed in the syncing state. * * Quiescing * * After a txg exits the open state, it enters the quiescing state. The * quiescing state is intended to provide a buffer between accepting new * transactions in the open state and writing them out to stable storage in * the syncing state. While quiescing, transactions can continue their * operation without delaying either of the other states. Typically, a txg is * in the quiescing state very briefly since the operations are bounded by * software latencies rather than, say, slower I/O latencies. After all * transactions complete, the txg is ready to enter the next state. * * Syncing * * In the syncing state, the in-memory state built up during the open and (to * a lesser degree) the quiescing states is written to stable storage. The * process of writing out modified data can, in turn modify more data. For * example when we write new blocks, we need to allocate space for them; those * allocations modify metadata (space maps)... which themselves must be * written to stable storage. During the sync state, ZFS iterates, writing out * data until it converges and all in-memory changes have been written out. * The first such pass is the largest as it encompasses all the modified user * data (as opposed to filesystem metadata). Subsequent passes typically have * far less data to write as they consist exclusively of filesystem metadata. * * To ensure convergence, after a certain number of passes ZFS begins * overwriting locations on stable storage that had been allocated earlier in * the syncing state (and subsequently freed). ZFS usually allocates new * blocks to optimize for large, continuous, writes. For the syncing state to * converge however it must complete a pass where no new blocks are allocated * since each allocation requires a modification of persistent metadata. * Further, to hasten convergence, after a prescribed number of passes, ZFS * also defers frees, and stops compressing. * * In addition to writing out user data, we must also execute synctasks during * the syncing context. A synctask is the mechanism by which some * administrative activities work such as creating and destroying snapshots or * datasets. Note that when a synctask is initiated it enters the open txg, * and ZFS then pushes that txg as quickly as possible to completion of the * syncing state in order to reduce the latency of the administrative * activity. To complete the syncing state, ZFS writes out a new uberblock, * the root of the tree of blocks that comprise all state stored on the ZFS * pool. Finally, if there is a quiesced txg waiting, we signal that it can * now transition to the syncing state. */ static void txg_sync_thread(void *arg); static void txg_quiesce_thread(void *arg); int zfs_txg_timeout = 5; /* max seconds worth of delta per txg */ /* * Prepare the txg subsystem. */ void txg_init(dsl_pool_t *dp, uint64_t txg) { tx_state_t *tx = &dp->dp_tx; int c; bzero(tx, sizeof (tx_state_t)); tx->tx_cpu = vmem_zalloc(max_ncpus * sizeof (tx_cpu_t), KM_SLEEP); for (c = 0; c < max_ncpus; c++) { int i; mutex_init(&tx->tx_cpu[c].tc_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&tx->tx_cpu[c].tc_open_lock, NULL, MUTEX_NOLOCKDEP, NULL); for (i = 0; i < TXG_SIZE; i++) { cv_init(&tx->tx_cpu[c].tc_cv[i], NULL, CV_DEFAULT, NULL); list_create(&tx->tx_cpu[c].tc_callbacks[i], sizeof (dmu_tx_callback_t), offsetof(dmu_tx_callback_t, dcb_node)); } } mutex_init(&tx->tx_sync_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&tx->tx_sync_more_cv, NULL, CV_DEFAULT, NULL); cv_init(&tx->tx_sync_done_cv, NULL, CV_DEFAULT, NULL); cv_init(&tx->tx_quiesce_more_cv, NULL, CV_DEFAULT, NULL); cv_init(&tx->tx_quiesce_done_cv, NULL, CV_DEFAULT, NULL); cv_init(&tx->tx_exit_cv, NULL, CV_DEFAULT, NULL); tx->tx_open_txg = txg; } /* * Close down the txg subsystem. */ void txg_fini(dsl_pool_t *dp) { tx_state_t *tx = &dp->dp_tx; int c; - ASSERT(tx->tx_threads == 0); + ASSERT0(tx->tx_threads); mutex_destroy(&tx->tx_sync_lock); cv_destroy(&tx->tx_sync_more_cv); cv_destroy(&tx->tx_sync_done_cv); cv_destroy(&tx->tx_quiesce_more_cv); cv_destroy(&tx->tx_quiesce_done_cv); cv_destroy(&tx->tx_exit_cv); for (c = 0; c < max_ncpus; c++) { int i; mutex_destroy(&tx->tx_cpu[c].tc_open_lock); mutex_destroy(&tx->tx_cpu[c].tc_lock); for (i = 0; i < TXG_SIZE; i++) { cv_destroy(&tx->tx_cpu[c].tc_cv[i]); list_destroy(&tx->tx_cpu[c].tc_callbacks[i]); } } if (tx->tx_commit_cb_taskq != NULL) taskq_destroy(tx->tx_commit_cb_taskq); vmem_free(tx->tx_cpu, max_ncpus * sizeof (tx_cpu_t)); bzero(tx, sizeof (tx_state_t)); } /* * Start syncing transaction groups. */ void txg_sync_start(dsl_pool_t *dp) { tx_state_t *tx = &dp->dp_tx; mutex_enter(&tx->tx_sync_lock); dprintf("pool %p\n", dp); - ASSERT(tx->tx_threads == 0); + ASSERT0(tx->tx_threads); tx->tx_threads = 2; tx->tx_quiesce_thread = thread_create(NULL, 0, txg_quiesce_thread, dp, 0, &p0, TS_RUN, defclsyspri); /* * The sync thread can need a larger-than-default stack size on * 32-bit x86. This is due in part to nested pools and * scrub_visitbp() recursion. */ tx->tx_sync_thread = thread_create(NULL, 0, txg_sync_thread, dp, 0, &p0, TS_RUN, defclsyspri); mutex_exit(&tx->tx_sync_lock); } static void txg_thread_enter(tx_state_t *tx, callb_cpr_t *cpr) { CALLB_CPR_INIT(cpr, &tx->tx_sync_lock, callb_generic_cpr, FTAG); mutex_enter(&tx->tx_sync_lock); } static void txg_thread_exit(tx_state_t *tx, callb_cpr_t *cpr, kthread_t **tpp) { ASSERT(*tpp != NULL); *tpp = NULL; tx->tx_threads--; cv_broadcast(&tx->tx_exit_cv); CALLB_CPR_EXIT(cpr); /* drops &tx->tx_sync_lock */ thread_exit(); } static void txg_thread_wait(tx_state_t *tx, callb_cpr_t *cpr, kcondvar_t *cv, clock_t time) { CALLB_CPR_SAFE_BEGIN(cpr); if (time) (void) cv_timedwait_sig(cv, &tx->tx_sync_lock, ddi_get_lbolt() + time); else cv_wait_sig(cv, &tx->tx_sync_lock); CALLB_CPR_SAFE_END(cpr, &tx->tx_sync_lock); } /* * Stop syncing transaction groups. */ void txg_sync_stop(dsl_pool_t *dp) { tx_state_t *tx = &dp->dp_tx; dprintf("pool %p\n", dp); /* * Finish off any work in progress. */ - ASSERT(tx->tx_threads == 2); + ASSERT3U(tx->tx_threads, ==, 2); /* * We need to ensure that we've vacated the deferred space_maps. */ txg_wait_synced(dp, tx->tx_open_txg + TXG_DEFER_SIZE); /* * Wake all sync threads and wait for them to die. */ mutex_enter(&tx->tx_sync_lock); - ASSERT(tx->tx_threads == 2); + ASSERT3U(tx->tx_threads, ==, 2); tx->tx_exiting = 1; cv_broadcast(&tx->tx_quiesce_more_cv); cv_broadcast(&tx->tx_quiesce_done_cv); cv_broadcast(&tx->tx_sync_more_cv); while (tx->tx_threads != 0) cv_wait(&tx->tx_exit_cv, &tx->tx_sync_lock); tx->tx_exiting = 0; mutex_exit(&tx->tx_sync_lock); } uint64_t txg_hold_open(dsl_pool_t *dp, txg_handle_t *th) { tx_state_t *tx = &dp->dp_tx; tx_cpu_t *tc; uint64_t txg; /* * It appears the processor id is simply used as a "random" * number to index into the array, and there isn't any other * significance to the chosen tx_cpu. Because.. Why not use * the current cpu to index into the array? */ kpreempt_disable(); tc = &tx->tx_cpu[CPU_SEQID]; kpreempt_enable(); mutex_enter(&tc->tc_open_lock); txg = tx->tx_open_txg; mutex_enter(&tc->tc_lock); tc->tc_count[txg & TXG_MASK]++; mutex_exit(&tc->tc_lock); th->th_cpu = tc; th->th_txg = txg; return (txg); } void txg_rele_to_quiesce(txg_handle_t *th) { tx_cpu_t *tc = th->th_cpu; ASSERT(!MUTEX_HELD(&tc->tc_lock)); mutex_exit(&tc->tc_open_lock); } void txg_register_callbacks(txg_handle_t *th, list_t *tx_callbacks) { tx_cpu_t *tc = th->th_cpu; int g = th->th_txg & TXG_MASK; mutex_enter(&tc->tc_lock); list_move_tail(&tc->tc_callbacks[g], tx_callbacks); mutex_exit(&tc->tc_lock); } void txg_rele_to_sync(txg_handle_t *th) { tx_cpu_t *tc = th->th_cpu; int g = th->th_txg & TXG_MASK; mutex_enter(&tc->tc_lock); ASSERT(tc->tc_count[g] != 0); if (--tc->tc_count[g] == 0) cv_broadcast(&tc->tc_cv[g]); mutex_exit(&tc->tc_lock); th->th_cpu = NULL; /* defensive */ } /* * Blocks until all transactions in the group are committed. * * On return, the transaction group has reached a stable state in which it can * then be passed off to the syncing context. */ static void txg_quiesce(dsl_pool_t *dp, uint64_t txg) { tx_state_t *tx = &dp->dp_tx; uint64_t tx_open_time; int g = txg & TXG_MASK; int c; /* * Grab all tc_open_locks so nobody else can get into this txg. */ for (c = 0; c < max_ncpus; c++) mutex_enter(&tx->tx_cpu[c].tc_open_lock); ASSERT(txg == tx->tx_open_txg); tx->tx_open_txg++; tx->tx_open_time = tx_open_time = gethrtime(); DTRACE_PROBE2(txg__quiescing, dsl_pool_t *, dp, uint64_t, txg); DTRACE_PROBE2(txg__opened, dsl_pool_t *, dp, uint64_t, tx->tx_open_txg); /* * Now that we've incremented tx_open_txg, we can let threads * enter the next transaction group. */ for (c = 0; c < max_ncpus; c++) mutex_exit(&tx->tx_cpu[c].tc_open_lock); spa_txg_history_set(dp->dp_spa, txg, TXG_STATE_OPEN, tx_open_time); spa_txg_history_add(dp->dp_spa, txg + 1, tx_open_time); /* * Quiesce the transaction group by waiting for everyone to txg_exit(). */ for (c = 0; c < max_ncpus; c++) { tx_cpu_t *tc = &tx->tx_cpu[c]; mutex_enter(&tc->tc_lock); while (tc->tc_count[g] != 0) cv_wait(&tc->tc_cv[g], &tc->tc_lock); mutex_exit(&tc->tc_lock); } spa_txg_history_set(dp->dp_spa, txg, TXG_STATE_QUIESCED, gethrtime()); } static void txg_do_callbacks(list_t *cb_list) { dmu_tx_do_callbacks(cb_list, 0); list_destroy(cb_list); kmem_free(cb_list, sizeof (list_t)); } /* * Dispatch the commit callbacks registered on this txg to worker threads. * * If no callbacks are registered for a given TXG, nothing happens. * This function creates a taskq for the associated pool, if needed. */ static void txg_dispatch_callbacks(dsl_pool_t *dp, uint64_t txg) { int c; tx_state_t *tx = &dp->dp_tx; list_t *cb_list; for (c = 0; c < max_ncpus; c++) { tx_cpu_t *tc = &tx->tx_cpu[c]; /* * No need to lock tx_cpu_t at this point, since this can * only be called once a txg has been synced. */ int g = txg & TXG_MASK; if (list_is_empty(&tc->tc_callbacks[g])) continue; if (tx->tx_commit_cb_taskq == NULL) { /* * Commit callback taskq hasn't been created yet. */ tx->tx_commit_cb_taskq = taskq_create("tx_commit_cb", max_ncpus, defclsyspri, max_ncpus, max_ncpus * 2, TASKQ_PREPOPULATE | TASKQ_DYNAMIC); } cb_list = kmem_alloc(sizeof (list_t), KM_SLEEP); list_create(cb_list, sizeof (dmu_tx_callback_t), offsetof(dmu_tx_callback_t, dcb_node)); list_move_tail(cb_list, &tc->tc_callbacks[g]); (void) taskq_dispatch(tx->tx_commit_cb_taskq, (task_func_t *) txg_do_callbacks, cb_list, TQ_SLEEP); } } /* * Wait for pending commit callbacks of already-synced transactions to finish * processing. * Calling this function from within a commit callback will deadlock. */ void txg_wait_callbacks(dsl_pool_t *dp) { tx_state_t *tx = &dp->dp_tx; if (tx->tx_commit_cb_taskq != NULL) taskq_wait_outstanding(tx->tx_commit_cb_taskq, 0); } static void txg_sync_thread(void *arg) { dsl_pool_t *dp = arg; spa_t *spa = dp->dp_spa; tx_state_t *tx = &dp->dp_tx; callb_cpr_t cpr; clock_t start, delta; (void) spl_fstrans_mark(); txg_thread_enter(tx, &cpr); start = delta = 0; for (;;) { clock_t timeout = zfs_txg_timeout * hz; clock_t timer; uint64_t txg; txg_stat_t *ts; /* * We sync when we're scanning, there's someone waiting * on us, or the quiesce thread has handed off a txg to * us, or we have reached our timeout. */ timer = (delta >= timeout ? 0 : timeout - delta); while (!dsl_scan_active(dp->dp_scan) && !tx->tx_exiting && timer > 0 && tx->tx_synced_txg >= tx->tx_sync_txg_waiting && tx->tx_quiesced_txg == 0 && dp->dp_dirty_total < zfs_dirty_data_sync) { dprintf("waiting; tx_synced=%llu waiting=%llu dp=%p\n", tx->tx_synced_txg, tx->tx_sync_txg_waiting, dp); txg_thread_wait(tx, &cpr, &tx->tx_sync_more_cv, timer); delta = ddi_get_lbolt() - start; timer = (delta > timeout ? 0 : timeout - delta); } /* * Wait until the quiesce thread hands off a txg to us, * prompting it to do so if necessary. */ while (!tx->tx_exiting && tx->tx_quiesced_txg == 0) { if (tx->tx_quiesce_txg_waiting < tx->tx_open_txg+1) tx->tx_quiesce_txg_waiting = tx->tx_open_txg+1; cv_broadcast(&tx->tx_quiesce_more_cv); txg_thread_wait(tx, &cpr, &tx->tx_quiesce_done_cv, 0); } if (tx->tx_exiting) txg_thread_exit(tx, &cpr, &tx->tx_sync_thread); /* * Consume the quiesced txg which has been handed off to * us. This may cause the quiescing thread to now be * able to quiesce another txg, so we must signal it. */ txg = tx->tx_quiesced_txg; tx->tx_quiesced_txg = 0; tx->tx_syncing_txg = txg; DTRACE_PROBE2(txg__syncing, dsl_pool_t *, dp, uint64_t, txg); ts = spa_txg_history_init_io(spa, txg, dp); cv_broadcast(&tx->tx_quiesce_more_cv); dprintf("txg=%llu quiesce_txg=%llu sync_txg=%llu\n", txg, tx->tx_quiesce_txg_waiting, tx->tx_sync_txg_waiting); mutex_exit(&tx->tx_sync_lock); start = ddi_get_lbolt(); spa_sync(spa, txg); delta = ddi_get_lbolt() - start; mutex_enter(&tx->tx_sync_lock); tx->tx_synced_txg = txg; tx->tx_syncing_txg = 0; DTRACE_PROBE2(txg__synced, dsl_pool_t *, dp, uint64_t, txg); spa_txg_history_fini_io(spa, ts); cv_broadcast(&tx->tx_sync_done_cv); /* * Dispatch commit callbacks to worker threads. */ txg_dispatch_callbacks(dp, txg); } } static void txg_quiesce_thread(void *arg) { dsl_pool_t *dp = arg; tx_state_t *tx = &dp->dp_tx; callb_cpr_t cpr; txg_thread_enter(tx, &cpr); for (;;) { uint64_t txg; /* * We quiesce when there's someone waiting on us. * However, we can only have one txg in "quiescing" or * "quiesced, waiting to sync" state. So we wait until * the "quiesced, waiting to sync" txg has been consumed * by the sync thread. */ while (!tx->tx_exiting && (tx->tx_open_txg >= tx->tx_quiesce_txg_waiting || tx->tx_quiesced_txg != 0)) txg_thread_wait(tx, &cpr, &tx->tx_quiesce_more_cv, 0); if (tx->tx_exiting) txg_thread_exit(tx, &cpr, &tx->tx_quiesce_thread); txg = tx->tx_open_txg; dprintf("txg=%llu quiesce_txg=%llu sync_txg=%llu\n", txg, tx->tx_quiesce_txg_waiting, tx->tx_sync_txg_waiting); mutex_exit(&tx->tx_sync_lock); txg_quiesce(dp, txg); mutex_enter(&tx->tx_sync_lock); /* * Hand this txg off to the sync thread. */ dprintf("quiesce done, handing off txg %llu\n", txg); tx->tx_quiesced_txg = txg; DTRACE_PROBE2(txg__quiesced, dsl_pool_t *, dp, uint64_t, txg); cv_broadcast(&tx->tx_sync_more_cv); cv_broadcast(&tx->tx_quiesce_done_cv); } } /* * Delay this thread by delay nanoseconds if we are still in the open * transaction group and there is already a waiting txg quiesing or quiesced. * Abort the delay if this txg stalls or enters the quiesing state. */ void txg_delay(dsl_pool_t *dp, uint64_t txg, hrtime_t delay, hrtime_t resolution) { tx_state_t *tx = &dp->dp_tx; hrtime_t start = gethrtime(); /* don't delay if this txg could transition to quiescing immediately */ if (tx->tx_open_txg > txg || tx->tx_syncing_txg == txg-1 || tx->tx_synced_txg == txg-1) return; mutex_enter(&tx->tx_sync_lock); if (tx->tx_open_txg > txg || tx->tx_synced_txg == txg-1) { mutex_exit(&tx->tx_sync_lock); return; } while (gethrtime() - start < delay && tx->tx_syncing_txg < txg-1 && !txg_stalled(dp)) { (void) cv_timedwait_hires(&tx->tx_quiesce_more_cv, &tx->tx_sync_lock, delay, resolution, 0); } DMU_TX_STAT_BUMP(dmu_tx_delay); mutex_exit(&tx->tx_sync_lock); } void txg_wait_synced(dsl_pool_t *dp, uint64_t txg) { tx_state_t *tx = &dp->dp_tx; ASSERT(!dsl_pool_config_held(dp)); mutex_enter(&tx->tx_sync_lock); - ASSERT(tx->tx_threads == 2); + ASSERT3U(tx->tx_threads, ==, 2); if (txg == 0) txg = tx->tx_open_txg + TXG_DEFER_SIZE; if (tx->tx_sync_txg_waiting < txg) tx->tx_sync_txg_waiting = txg; dprintf("txg=%llu quiesce_txg=%llu sync_txg=%llu\n", txg, tx->tx_quiesce_txg_waiting, tx->tx_sync_txg_waiting); while (tx->tx_synced_txg < txg) { dprintf("broadcasting sync more " "tx_synced=%llu waiting=%llu dp=%p\n", tx->tx_synced_txg, tx->tx_sync_txg_waiting, dp); cv_broadcast(&tx->tx_sync_more_cv); cv_wait(&tx->tx_sync_done_cv, &tx->tx_sync_lock); } mutex_exit(&tx->tx_sync_lock); } void txg_wait_open(dsl_pool_t *dp, uint64_t txg) { tx_state_t *tx = &dp->dp_tx; ASSERT(!dsl_pool_config_held(dp)); mutex_enter(&tx->tx_sync_lock); - ASSERT(tx->tx_threads == 2); + ASSERT3U(tx->tx_threads, ==, 2); if (txg == 0) txg = tx->tx_open_txg + 1; if (tx->tx_quiesce_txg_waiting < txg) tx->tx_quiesce_txg_waiting = txg; dprintf("txg=%llu quiesce_txg=%llu sync_txg=%llu\n", txg, tx->tx_quiesce_txg_waiting, tx->tx_sync_txg_waiting); while (tx->tx_open_txg < txg) { cv_broadcast(&tx->tx_quiesce_more_cv); cv_wait(&tx->tx_quiesce_done_cv, &tx->tx_sync_lock); } mutex_exit(&tx->tx_sync_lock); } /* * If there isn't a txg syncing or in the pipeline, push another txg through * the pipeline by queiscing the open txg. */ void txg_kick(dsl_pool_t *dp) { tx_state_t *tx = &dp->dp_tx; ASSERT(!dsl_pool_config_held(dp)); mutex_enter(&tx->tx_sync_lock); if (tx->tx_syncing_txg == 0 && tx->tx_quiesce_txg_waiting <= tx->tx_open_txg && tx->tx_sync_txg_waiting <= tx->tx_synced_txg && tx->tx_quiesced_txg <= tx->tx_synced_txg) { tx->tx_quiesce_txg_waiting = tx->tx_open_txg + 1; cv_broadcast(&tx->tx_quiesce_more_cv); } mutex_exit(&tx->tx_sync_lock); } boolean_t txg_stalled(dsl_pool_t *dp) { tx_state_t *tx = &dp->dp_tx; return (tx->tx_quiesce_txg_waiting > tx->tx_open_txg); } boolean_t txg_sync_waiting(dsl_pool_t *dp) { tx_state_t *tx = &dp->dp_tx; return (tx->tx_syncing_txg <= tx->tx_sync_txg_waiting || tx->tx_quiesced_txg != 0); } /* * Verify that this txg is active (open, quiescing, syncing). Non-active * txg's should not be manipulated. */ void txg_verify(spa_t *spa, uint64_t txg) { ASSERTV(dsl_pool_t *dp = spa_get_dsl(spa)); if (txg <= TXG_INITIAL || txg == ZILTEST_TXG) return; ASSERT3U(txg, <=, dp->dp_tx.tx_open_txg); ASSERT3U(txg, >=, dp->dp_tx.tx_synced_txg); ASSERT3U(txg, >=, dp->dp_tx.tx_open_txg - TXG_CONCURRENT_STATES); } /* * Per-txg object lists. */ void txg_list_create(txg_list_t *tl, spa_t *spa, size_t offset) { int t; mutex_init(&tl->tl_lock, NULL, MUTEX_DEFAULT, NULL); tl->tl_offset = offset; tl->tl_spa = spa; for (t = 0; t < TXG_SIZE; t++) tl->tl_head[t] = NULL; } void txg_list_destroy(txg_list_t *tl) { int t; for (t = 0; t < TXG_SIZE; t++) ASSERT(txg_list_empty(tl, t)); mutex_destroy(&tl->tl_lock); } boolean_t txg_list_empty(txg_list_t *tl, uint64_t txg) { txg_verify(tl->tl_spa, txg); return (tl->tl_head[txg & TXG_MASK] == NULL); } /* * Returns true if all txg lists are empty. * * Warning: this is inherently racy (an item could be added immediately * after this function returns). We don't bother with the lock because * it wouldn't change the semantics. */ boolean_t txg_all_lists_empty(txg_list_t *tl) { for (int i = 0; i < TXG_SIZE; i++) { if (!txg_list_empty(tl, i)) { return (B_FALSE); } } return (B_TRUE); } /* * Add an entry to the list (unless it's already on the list). * Returns B_TRUE if it was actually added. */ boolean_t txg_list_add(txg_list_t *tl, void *p, uint64_t txg) { int t = txg & TXG_MASK; txg_node_t *tn = (txg_node_t *)((char *)p + tl->tl_offset); boolean_t add; txg_verify(tl->tl_spa, txg); mutex_enter(&tl->tl_lock); add = (tn->tn_member[t] == 0); if (add) { tn->tn_member[t] = 1; tn->tn_next[t] = tl->tl_head[t]; tl->tl_head[t] = tn; } mutex_exit(&tl->tl_lock); return (add); } /* * Add an entry to the end of the list, unless it's already on the list. * (walks list to find end) * Returns B_TRUE if it was actually added. */ boolean_t txg_list_add_tail(txg_list_t *tl, void *p, uint64_t txg) { int t = txg & TXG_MASK; txg_node_t *tn = (txg_node_t *)((char *)p + tl->tl_offset); boolean_t add; txg_verify(tl->tl_spa, txg); mutex_enter(&tl->tl_lock); add = (tn->tn_member[t] == 0); if (add) { txg_node_t **tp; for (tp = &tl->tl_head[t]; *tp != NULL; tp = &(*tp)->tn_next[t]) continue; tn->tn_member[t] = 1; tn->tn_next[t] = NULL; *tp = tn; } mutex_exit(&tl->tl_lock); return (add); } /* * Remove the head of the list and return it. */ void * txg_list_remove(txg_list_t *tl, uint64_t txg) { int t = txg & TXG_MASK; txg_node_t *tn; void *p = NULL; txg_verify(tl->tl_spa, txg); mutex_enter(&tl->tl_lock); if ((tn = tl->tl_head[t]) != NULL) { p = (char *)tn - tl->tl_offset; tl->tl_head[t] = tn->tn_next[t]; tn->tn_next[t] = NULL; tn->tn_member[t] = 0; } mutex_exit(&tl->tl_lock); return (p); } /* * Remove a specific item from the list and return it. */ void * txg_list_remove_this(txg_list_t *tl, void *p, uint64_t txg) { int t = txg & TXG_MASK; txg_node_t *tn, **tp; txg_verify(tl->tl_spa, txg); mutex_enter(&tl->tl_lock); for (tp = &tl->tl_head[t]; (tn = *tp) != NULL; tp = &tn->tn_next[t]) { if ((char *)tn - tl->tl_offset == p) { *tp = tn->tn_next[t]; tn->tn_next[t] = NULL; tn->tn_member[t] = 0; mutex_exit(&tl->tl_lock); return (p); } } mutex_exit(&tl->tl_lock); return (NULL); } boolean_t txg_list_member(txg_list_t *tl, void *p, uint64_t txg) { int t = txg & TXG_MASK; txg_node_t *tn = (txg_node_t *)((char *)p + tl->tl_offset); txg_verify(tl->tl_spa, txg); return (tn->tn_member[t] != 0); } /* * Walk a txg list -- only safe if you know it's not changing. */ void * txg_list_head(txg_list_t *tl, uint64_t txg) { int t = txg & TXG_MASK; txg_node_t *tn = tl->tl_head[t]; txg_verify(tl->tl_spa, txg); return (tn == NULL ? NULL : (char *)tn - tl->tl_offset); } void * txg_list_next(txg_list_t *tl, void *p, uint64_t txg) { int t = txg & TXG_MASK; txg_node_t *tn = (txg_node_t *)((char *)p + tl->tl_offset); txg_verify(tl->tl_spa, txg); tn = tn->tn_next[t]; return (tn == NULL ? NULL : (char *)tn - tl->tl_offset); } #if defined(_KERNEL) && defined(HAVE_SPL) EXPORT_SYMBOL(txg_init); EXPORT_SYMBOL(txg_fini); EXPORT_SYMBOL(txg_sync_start); EXPORT_SYMBOL(txg_sync_stop); EXPORT_SYMBOL(txg_hold_open); EXPORT_SYMBOL(txg_rele_to_quiesce); EXPORT_SYMBOL(txg_rele_to_sync); EXPORT_SYMBOL(txg_register_callbacks); EXPORT_SYMBOL(txg_delay); EXPORT_SYMBOL(txg_wait_synced); EXPORT_SYMBOL(txg_wait_open); EXPORT_SYMBOL(txg_wait_callbacks); EXPORT_SYMBOL(txg_stalled); EXPORT_SYMBOL(txg_sync_waiting); module_param(zfs_txg_timeout, int, 0644); MODULE_PARM_DESC(zfs_txg_timeout, "Max seconds worth of delta per txg"); #endif diff --git a/module/zfs/zfs_vnops.c b/module/zfs/zfs_vnops.c index a60cd619838a..977035fd94f1 100644 --- a/module/zfs/zfs_vnops.c +++ b/module/zfs/zfs_vnops.c @@ -1,4951 +1,4953 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2012, 2015 by Delphix. All rights reserved. * Copyright (c) 2015 by Chunwei Chen. All rights reserved. * Copyright 2017 Nexenta Systems, Inc. */ /* 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 #include #include #include #include #include #include #include #include #include #include "fs/fs_subr.h" #include #include #include #include #include #include #include #include #include #include #include +#include /* * Programming rules. * * Each vnode op performs some logical unit of work. To do this, the ZPL must * properly lock its in-core state, create a DMU transaction, do the work, * record this work in the intent log (ZIL), commit the DMU transaction, * and wait for the intent log to commit if it is a synchronous operation. * Moreover, the vnode ops must work in both normal and log replay context. * The ordering of events is important to avoid deadlocks and references * to freed memory. The example below illustrates the following Big Rules: * * (1) A check must be made in each zfs thread for a mounted file system. * This is done avoiding races using ZFS_ENTER(zfsvfs). * A ZFS_EXIT(zfsvfs) is needed before all returns. Any znodes * must be checked with ZFS_VERIFY_ZP(zp). Both of these macros * can return EIO from the calling function. * * (2) iput() should always be the last thing except for zil_commit() * (if necessary) and ZFS_EXIT(). This is for 3 reasons: * First, if it's the last reference, the vnode/znode * can be freed, so the zp may point to freed memory. Second, the last * reference will call zfs_zinactive(), which may induce a lot of work -- * pushing cached pages (which acquires range locks) and syncing out * cached atime changes. Third, zfs_zinactive() may require a new tx, * which could deadlock the system if you were already holding one. * If you must call iput() within a tx then use zfs_iput_async(). * * (3) All range locks must be grabbed before calling dmu_tx_assign(), * as they can span dmu_tx_assign() calls. * * (4) If ZPL locks are held, pass TXG_NOWAIT as the second argument to * dmu_tx_assign(). This is critical because we don't want to block * while holding locks. * * If no ZPL locks are held (aside from ZFS_ENTER()), use TXG_WAIT. This * reduces lock contention and CPU usage when we must wait (note that if * throughput is constrained by the storage, nearly every transaction * must wait). * * Note, in particular, that if a lock is sometimes acquired before * the tx assigns, and sometimes after (e.g. z_lock), then failing * to use a non-blocking assign can deadlock the system. The scenario: * * Thread A has grabbed a lock before calling dmu_tx_assign(). * Thread B is in an already-assigned tx, and blocks for this lock. * Thread A calls dmu_tx_assign(TXG_WAIT) and blocks in txg_wait_open() * forever, because the previous txg can't quiesce until B's tx commits. * * If dmu_tx_assign() returns ERESTART and zfsvfs->z_assign is TXG_NOWAIT, * then drop all locks, call dmu_tx_wait(), and try again. On subsequent * calls to dmu_tx_assign(), pass TXG_WAITED rather than TXG_NOWAIT, * to indicate that this operation has already called dmu_tx_wait(). * This will ensure that we don't retry forever, waiting a short bit * each time. * * (5) If the operation succeeded, generate the intent log entry for it * before dropping locks. This ensures that the ordering of events * in the intent log matches the order in which they actually occurred. * During ZIL replay the zfs_log_* functions will update the sequence * number to indicate the zil transaction has replayed. * * (6) At the end of each vnode op, the DMU tx must always commit, * regardless of whether there were any errors. * * (7) After dropping all locks, invoke zil_commit(zilog, foid) * to ensure that synchronous semantics are provided when necessary. * * In general, this is how things should be ordered in each vnode op: * * ZFS_ENTER(zfsvfs); // exit if unmounted * top: * zfs_dirent_lock(&dl, ...) // lock directory entry (may igrab()) * rw_enter(...); // grab any other locks you need * tx = dmu_tx_create(...); // get DMU tx * dmu_tx_hold_*(); // hold each object you might modify * error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); * if (error) { * rw_exit(...); // drop locks * zfs_dirent_unlock(dl); // unlock directory entry * iput(...); // release held vnodes * if (error == ERESTART) { * waited = B_TRUE; * dmu_tx_wait(tx); * dmu_tx_abort(tx); * goto top; * } * dmu_tx_abort(tx); // abort DMU tx * ZFS_EXIT(zfsvfs); // finished in zfs * return (error); // really out of space * } * error = do_real_work(); // do whatever this VOP does * if (error == 0) * zfs_log_*(...); // on success, make ZIL entry * dmu_tx_commit(tx); // commit DMU tx -- error or not * rw_exit(...); // drop locks * zfs_dirent_unlock(dl); // unlock directory entry * iput(...); // release held vnodes * zil_commit(zilog, foid); // synchronous when necessary * ZFS_EXIT(zfsvfs); // finished in zfs * return (error); // done, report error */ /* * Virus scanning is unsupported. It would be possible to add a hook * here to performance the required virus scan. This could be done * entirely in the kernel or potentially as an update to invoke a * scanning utility. */ static int zfs_vscan(struct inode *ip, cred_t *cr, int async) { return (0); } /* ARGSUSED */ int zfs_open(struct inode *ip, int mode, int flag, cred_t *cr) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); /* Honor ZFS_APPENDONLY file attribute */ if ((mode & FMODE_WRITE) && (zp->z_pflags & ZFS_APPENDONLY) && ((flag & O_APPEND) == 0)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } /* Virus scan eligible files on open */ if (!zfs_has_ctldir(zp) && zfsvfs->z_vscan && S_ISREG(ip->i_mode) && !(zp->z_pflags & ZFS_AV_QUARANTINED) && zp->z_size > 0) { if (zfs_vscan(ip, cr, 0) != 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EACCES)); } } /* Keep a count of the synchronous opens in the znode */ if (flag & O_SYNC) atomic_inc_32(&zp->z_sync_cnt); ZFS_EXIT(zfsvfs); return (0); } /* ARGSUSED */ int zfs_close(struct inode *ip, int flag, cred_t *cr) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); /* Decrement the synchronous opens in the znode */ if (flag & O_SYNC) atomic_dec_32(&zp->z_sync_cnt); if (!zfs_has_ctldir(zp) && zfsvfs->z_vscan && S_ISREG(ip->i_mode) && !(zp->z_pflags & ZFS_AV_QUARANTINED) && zp->z_size > 0) VERIFY(zfs_vscan(ip, cr, 1) == 0); ZFS_EXIT(zfsvfs); return (0); } #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(struct inode *ip, int cmd, loff_t *off) { znode_t *zp = ITOZ(ip); 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 == SEEK_HOLE) hole = B_TRUE; else hole = B_FALSE; error = dmu_offset_next(ZTOZSB(zp)->z_os, zp->z_id, hole, &noff); 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(struct inode *ip, int cmd, loff_t *off) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); int error; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); error = zfs_holey_common(ip, cmd, off); ZFS_EXIT(zfsvfs); return (error); } #endif /* SEEK_HOLE && SEEK_DATA */ #if defined(_KERNEL) /* * When a file is memory mapped, we must keep the IO data synchronized * between the DMU cache and the memory mapped pages. What this means: * * On Write: If we find a memory mapped page, we write to *both* * the page and the dmu buffer. */ static void update_pages(struct inode *ip, int64_t start, int len, objset_t *os, uint64_t oid) { struct address_space *mp = ip->i_mapping; struct page *pp; uint64_t nbytes; int64_t off; void *pb; off = start & (PAGE_SIZE-1); for (start &= PAGE_MASK; len > 0; start += PAGE_SIZE) { nbytes = MIN(PAGE_SIZE - off, len); pp = find_lock_page(mp, start >> PAGE_SHIFT); if (pp) { if (mapping_writably_mapped(mp)) flush_dcache_page(pp); pb = kmap(pp); (void) dmu_read(os, oid, start+off, nbytes, pb+off, DMU_READ_PREFETCH); kunmap(pp); if (mapping_writably_mapped(mp)) flush_dcache_page(pp); mark_page_accessed(pp); SetPageUptodate(pp); ClearPageError(pp); unlock_page(pp); put_page(pp); } len -= nbytes; off = 0; } } /* * When a file is memory mapped, we must keep the IO data synchronized * between the DMU cache and the memory mapped pages. What this means: * * On Read: We "read" preferentially from memory mapped pages, * else we default from the dmu buffer. * * NOTE: We will always "break up" the IO into PAGESIZE uiomoves when * the file is memory mapped. */ static int mappedread(struct inode *ip, int nbytes, uio_t *uio) { struct address_space *mp = ip->i_mapping; struct page *pp; znode_t *zp = ITOZ(ip); int64_t start, off; uint64_t bytes; int len = nbytes; int error = 0; void *pb; start = uio->uio_loffset; off = start & (PAGE_SIZE-1); for (start &= PAGE_MASK; len > 0; start += PAGE_SIZE) { bytes = MIN(PAGE_SIZE - off, len); pp = find_lock_page(mp, start >> PAGE_SHIFT); if (pp) { ASSERT(PageUptodate(pp)); pb = kmap(pp); error = uiomove(pb + off, bytes, UIO_READ, uio); kunmap(pp); if (mapping_writably_mapped(mp)) flush_dcache_page(pp); mark_page_accessed(pp); unlock_page(pp); put_page(pp); } else { error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl), uio, bytes); } len -= bytes; off = 0; if (error) break; } return (error); } #endif /* _KERNEL */ unsigned long zfs_read_chunk_size = 1024 * 1024; /* Tunable */ unsigned long zfs_delete_blocks = DMU_MAX_DELETEBLKCNT; /* * Read bytes from specified file into supplied buffer. * * IN: ip - inode of file to be read from. * uio - structure supplying read location, range info, * and return buffer. * ioflag - FSYNC 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 */ /* ARGSUSED */ int zfs_read(struct inode *ip, uio_t *uio, int ioflag, cred_t *cr) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); ssize_t n, nbytes; int error = 0; rl_t *rl; #ifdef HAVE_UIO_ZEROCOPY xuio_t *xuio = NULL; #endif /* HAVE_UIO_ZEROCOPY */ ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); if (zp->z_pflags & ZFS_AV_QUARANTINED) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EACCES)); } /* * Validate file offset */ if (uio->uio_loffset < (offset_t)0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } /* * Fasttrack empty reads */ if (uio->uio_resid == 0) { ZFS_EXIT(zfsvfs); return (0); } /* * If we're in FRSYNC mode, sync out this znode before reading it. * Only do this for non-snapshots. */ if (zfsvfs->z_log && (ioflag & FRSYNC || zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)) zil_commit(zfsvfs->z_log, zp->z_id); /* * Lock the range against changes. */ rl = zfs_range_lock(&zp->z_range_lock, uio->uio_loffset, uio->uio_resid, 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 (uio->uio_loffset >= zp->z_size) { error = 0; goto out; } ASSERT(uio->uio_loffset < zp->z_size); n = MIN(uio->uio_resid, zp->z_size - uio->uio_loffset); #ifdef HAVE_UIO_ZEROCOPY if ((uio->uio_extflg == UIO_XUIO) && (((xuio_t *)uio)->xu_type == UIOTYPE_ZEROCOPY)) { int nblk; int blksz = zp->z_blksz; uint64_t offset = uio->uio_loffset; xuio = (xuio_t *)uio; if ((ISP2(blksz))) { nblk = (P2ROUNDUP(offset + n, blksz) - P2ALIGN(offset, blksz)) / blksz; } else { ASSERT(offset + n <= blksz); nblk = 1; } (void) dmu_xuio_init(xuio, nblk); if (vn_has_cached_data(ip)) { /* * For simplicity, we always allocate a full buffer * even if we only expect to read a portion of a block. */ while (--nblk >= 0) { (void) dmu_xuio_add(xuio, dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl), blksz), 0, blksz); } } } #endif /* HAVE_UIO_ZEROCOPY */ while (n > 0) { nbytes = MIN(n, zfs_read_chunk_size - P2PHASE(uio->uio_loffset, zfs_read_chunk_size)); if (zp->z_is_mapped && !(ioflag & O_DIRECT)) { error = mappedread(ip, nbytes, uio); } else { error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl), uio, nbytes); } if (error) { /* convert checksum errors into IO errors */ if (error == ECKSUM) error = SET_ERROR(EIO); break; } n -= nbytes; } out: zfs_range_unlock(rl); ZFS_EXIT(zfsvfs); return (error); } /* * Write the bytes to a file. * * IN: ip - inode of file to be written to. * uio - structure supplying write location, range info, * and data buffer. * ioflag - FAPPEND 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 */ /* ARGSUSED */ int zfs_write(struct inode *ip, uio_t *uio, int ioflag, cred_t *cr) { znode_t *zp = ITOZ(ip); rlim64_t limit = uio->uio_limit; ssize_t start_resid = uio->uio_resid; ssize_t tx_bytes; uint64_t end_size; dmu_tx_t *tx; zfsvfs_t *zfsvfs = ZTOZSB(zp); zilog_t *zilog; offset_t woff; ssize_t n, nbytes; rl_t *rl; int max_blksz = zfsvfs->z_max_blksz; int error = 0; arc_buf_t *abuf; const iovec_t *aiov = NULL; xuio_t *xuio = NULL; int write_eof; int count = 0; sa_bulk_attr_t bulk[4]; uint64_t mtime[2], ctime[2]; uint32_t uid; #ifdef HAVE_UIO_ZEROCOPY int i_iov = 0; const iovec_t *iovp = uio->uio_iov; ASSERTV(int iovcnt = uio->uio_iovcnt); #endif /* * Fasttrack empty write */ n = start_resid; if (n == 0) return (0); if (limit == RLIM64_INFINITY || limit > MAXOFFSET_T) limit = MAXOFFSET_T; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); 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); return (SET_ERROR(EROFS)); } /* * If immutable or not appending then return EPERM */ if ((zp->z_pflags & (ZFS_IMMUTABLE | ZFS_READONLY)) || ((zp->z_pflags & ZFS_APPENDONLY) && !(ioflag & FAPPEND) && (uio->uio_loffset < zp->z_size))) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } zilog = zfsvfs->z_log; /* * Validate file offset */ woff = ioflag & FAPPEND ? zp->z_size : uio->uio_loffset; if (woff < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } /* * Pre-fault the pages to ensure slow (eg NFS) pages * don't hold up txg. * Skip this if uio contains loaned arc_buf. */ #ifdef HAVE_UIO_ZEROCOPY if ((uio->uio_extflg == UIO_XUIO) && (((xuio_t *)uio)->xu_type == UIOTYPE_ZEROCOPY)) xuio = (xuio_t *)uio; else #endif uio_prefaultpages(MIN(n, max_blksz), uio); /* * If in append mode, set the io offset pointer to eof. */ if (ioflag & FAPPEND) { /* * Obtain an appending range lock to guarantee file append * semantics. We reset the write offset once we have the lock. */ rl = zfs_range_lock(&zp->z_range_lock, 0, n, RL_APPEND); woff = rl->r_off; if (rl->r_len == 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; } uio->uio_loffset = 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. */ rl = zfs_range_lock(&zp->z_range_lock, woff, n, RL_WRITER); } if (woff >= limit) { zfs_range_unlock(rl); ZFS_EXIT(zfsvfs); return (SET_ERROR(EFBIG)); } if ((woff + n) > limit || woff > (limit - n)) n = limit - woff; /* Will this write extend the file length? */ write_eof = (woff + n > zp->z_size); end_size = MAX(zp->z_size, woff + n); /* * 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) { abuf = NULL; woff = uio->uio_loffset; if (zfs_owner_overquota(zfsvfs, zp, B_FALSE) || zfs_owner_overquota(zfsvfs, zp, B_TRUE)) { if (abuf != NULL) dmu_return_arcbuf(abuf); error = SET_ERROR(EDQUOT); break; } if (xuio && abuf == NULL) { #ifdef HAVE_UIO_ZEROCOPY ASSERT(i_iov < iovcnt); ASSERT3U(uio->uio_segflg, !=, UIO_BVEC); aiov = &iovp[i_iov]; abuf = dmu_xuio_arcbuf(xuio, i_iov); dmu_xuio_clear(xuio, i_iov); ASSERT((aiov->iov_base == abuf->b_data) || ((char *)aiov->iov_base - (char *)abuf->b_data + aiov->iov_len == arc_buf_size(abuf))); i_iov++; #endif } else if (abuf == NULL && n >= max_blksz && woff >= zp->z_size && P2PHASE(woff, max_blksz) == 0 && zp->z_blksz == max_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). */ size_t cbytes; abuf = dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl), max_blksz); ASSERT(abuf != NULL); ASSERT(arc_buf_size(abuf) == max_blksz); if ((error = uiocopy(abuf->b_data, max_blksz, UIO_WRITE, uio, &cbytes))) { dmu_return_arcbuf(abuf); break; } ASSERT(cbytes == max_blksz); } /* * Start a transaction. */ tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); dmu_tx_hold_write(tx, zp->z_id, woff, MIN(n, max_blksz)); zfs_sa_upgrade_txholds(tx, zp); error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); if (abuf != NULL) dmu_return_arcbuf(abuf); break; } /* * If zfs_range_lock() over-locked we grow the blocksize * and then reduce the lock range. This will only happen * on the first iteration since zfs_range_reduce() will * shrink down r_len to the appropriate size. */ if (rl->r_len == UINT64_MAX) { uint64_t new_blksz; if (zp->z_blksz > max_blksz) { /* * File's blocksize is already larger than the * "recordsize" property. Only let it grow to * the next power of 2. */ ASSERT(!ISP2(zp->z_blksz)); new_blksz = MIN(end_size, 1 << highbit64(zp->z_blksz)); } else { new_blksz = MIN(end_size, max_blksz); } zfs_grow_blocksize(zp, new_blksz, tx); zfs_range_reduce(rl, woff, n); } /* * XXX - should we really limit each write to z_max_blksz? * Perhaps we should use SPA_MAXBLOCKSIZE chunks? */ nbytes = MIN(n, max_blksz - P2PHASE(woff, max_blksz)); if (abuf == NULL) { tx_bytes = uio->uio_resid; error = dmu_write_uio_dbuf(sa_get_db(zp->z_sa_hdl), uio, nbytes, tx); tx_bytes -= uio->uio_resid; } else { tx_bytes = nbytes; ASSERT(xuio == NULL || tx_bytes == aiov->iov_len); /* * If this is not a full block write, but we are * extending the file past EOF and this data starts * block-aligned, use assign_arcbuf(). Otherwise, * write via dmu_write(). */ if (tx_bytes < max_blksz && (!write_eof || aiov->iov_base != abuf->b_data)) { ASSERT(xuio); dmu_write(zfsvfs->z_os, zp->z_id, woff, /* cppcheck-suppress nullPointer */ aiov->iov_len, aiov->iov_base, tx); dmu_return_arcbuf(abuf); xuio_stat_wbuf_copied(); } else { ASSERT(xuio || tx_bytes == max_blksz); dmu_assign_arcbuf_by_dbuf( sa_get_db(zp->z_sa_hdl), woff, abuf, tx); } ASSERT(tx_bytes <= uio->uio_resid); uioskip(uio, tx_bytes); } if (tx_bytes && zp->z_is_mapped && !(ioflag & O_DIRECT)) { update_pages(ip, woff, tx_bytes, zfsvfs->z_os, zp->z_id); } /* * 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; } /* * 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 to 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); uid = KUID_TO_SUID(ip->i_uid); 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(cr, ((zp->z_mode & S_ISUID) != 0 && uid == 0)) != 0) { uint64_t newmode; zp->z_mode &= ~(S_ISUID | S_ISGID); ip->i_mode = 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); 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) < uio->uio_loffset) { (void) atomic_cas_64(&zp->z_size, end_size, uio->uio_loffset); ASSERT(error == 0); } /* * 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; error = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx); zfs_log_write(zilog, tx, TX_WRITE, zp, woff, tx_bytes, ioflag, NULL, NULL); dmu_tx_commit(tx); if (error != 0) break; ASSERT(tx_bytes == nbytes); n -= nbytes; if (!xuio && n > 0) uio_prefaultpages(MIN(n, max_blksz), uio); } zfs_inode_update(zp); zfs_range_unlock(rl); /* * If we're in replay mode, or we made no progress, return error. * Otherwise, it's at least a partial write, so it's successful. */ if (zfsvfs->z_replay || uio->uio_resid == start_resid) { ZFS_EXIT(zfsvfs); return (error); } if (ioflag & (FSYNC | FDSYNC) || zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, zp->z_id); ZFS_EXIT(zfsvfs); return (0); } /* * Drop a reference on the passed inode asynchronously. This ensures * that the caller will never drop the last reference on an inode in * the current context. Doing so while holding open a tx could result * in a deadlock if iput_final() re-enters the filesystem code. */ void zfs_iput_async(struct inode *ip) { objset_t *os = ITOZSB(ip)->z_os; ASSERT(atomic_read(&ip->i_count) > 0); ASSERT(os != NULL); if (atomic_read(&ip->i_count) == 1) VERIFY(taskq_dispatch(dsl_pool_iput_taskq(dmu_objset_pool(os)), (task_func_t *)iput, ip, TQ_SLEEP) != TASKQID_INVALID); else iput(ip); } void zfs_get_done(zgd_t *zgd, int error) { znode_t *zp = zgd->zgd_private; if (zgd->zgd_db) dmu_buf_rele(zgd->zgd_db, zgd); zfs_range_unlock(zgd->zgd_rl); /* * Release the vnode asynchronously as we currently have the * txg stopped from syncing. */ zfs_iput_async(ZTOI(zp)); if (error == 0 && zgd->zgd_bp) - zil_add_block(zgd->zgd_zilog, zgd->zgd_bp); + zil_lwb_add_block(zgd->zgd_lwb, zgd->zgd_bp); kmem_free(zgd, sizeof (zgd_t)); } #ifdef DEBUG static int zil_fault_io = 0; #endif /* * Get data to generate a TX_WRITE intent log record. */ int -zfs_get_data(void *arg, lr_write_t *lr, char *buf, zio_t *zio) +zfs_get_data(void *arg, 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; dmu_buf_t *db; zgd_t *zgd; int error = 0; - ASSERT(zio != NULL); - ASSERT(size != 0); + ASSERT3P(lwb, !=, NULL); + ASSERT3P(zio, !=, 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_iput_async(ZTOI(zp)); return (SET_ERROR(ENOENT)); } zgd = (zgd_t *)kmem_zalloc(sizeof (zgd_t), KM_SLEEP); - zgd->zgd_zilog = zfsvfs->z_log; + 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_rl = zfs_range_lock(&zp->z_range_lock, 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); } ASSERT(error == 0 || error == ENOENT); } else { /* indirect write */ /* * 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_rl = zfs_range_lock(&zp->z_range_lock, offset, size, RL_READER); if (zp->z_blksz == size) break; offset += blkoff; zfs_range_unlock(zgd->zgd_rl); } /* test for truncation needs to be done while range locked */ if (lr->lr_offset >= zp->z_size) error = SET_ERROR(ENOENT); #ifdef DEBUG if (zil_fault_io) { error = SET_ERROR(EIO); zil_fault_io = 0; } #endif if (error == 0) error = dmu_buf_hold(os, object, offset, zgd, &db, DMU_READ_NO_PREFETCH); if (error == 0) { blkptr_t *bp = &lr->lr_blkptr; zgd->zgd_db = db; zgd->zgd_bp = bp; ASSERT(db->db_offset == offset); ASSERT(db->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; error = 0; } } } zfs_get_done(zgd, error); return (error); } /*ARGSUSED*/ int zfs_access(struct inode *ip, int mode, int flag, cred_t *cr) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); int error; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); if (flag & V_ACE_MASK) error = zfs_zaccess(zp, mode, flag, B_FALSE, cr); else error = zfs_zaccess_rwx(zp, mode, flag, cr); ZFS_EXIT(zfsvfs); return (error); } /* * Lookup an entry in a directory, or an extended attribute directory. * If it exists, return a held inode reference for it. * * IN: dip - inode of directory to search. * nm - name of entry to lookup. * flags - LOOKUP_XATTR set if looking for an attribute. * cr - credentials of caller. * direntflags - directory lookup flags * realpnp - returned pathname. * * OUT: ipp - inode of located entry, NULL if not found. * * RETURN: 0 on success, error code on failure. * * Timestamps: * NA */ /* ARGSUSED */ int zfs_lookup(struct inode *dip, char *nm, struct inode **ipp, int flags, cred_t *cr, int *direntflags, pathname_t *realpnp) { znode_t *zdp = ITOZ(dip); zfsvfs_t *zfsvfs = ITOZSB(dip); int error = 0; /* * Fast path lookup, however we must skip DNLC lookup * for case folding or normalizing lookups because the * DNLC code only stores the passed in name. This means * creating 'a' and removing 'A' on a case insensitive * file system would work, but DNLC still thinks 'a' * exists and won't let you create it again on the next * pass through fast path. */ if (!(flags & (LOOKUP_XATTR | FIGNORECASE))) { if (!S_ISDIR(dip->i_mode)) { return (SET_ERROR(ENOTDIR)); } else if (zdp->z_sa_hdl == NULL) { return (SET_ERROR(EIO)); } if (nm[0] == 0 || (nm[0] == '.' && nm[1] == '\0')) { error = zfs_fastaccesschk_execute(zdp, cr); if (!error) { *ipp = dip; igrab(*ipp); return (0); } return (error); #ifdef HAVE_DNLC } else if (!zdp->z_zfsvfs->z_norm && (zdp->z_zfsvfs->z_case == ZFS_CASE_SENSITIVE)) { vnode_t *tvp = dnlc_lookup(dvp, nm); if (tvp) { error = zfs_fastaccesschk_execute(zdp, cr); if (error) { iput(tvp); return (error); } if (tvp == DNLC_NO_VNODE) { iput(tvp); return (SET_ERROR(ENOENT)); } else { *vpp = tvp; return (specvp_check(vpp, cr)); } } #endif /* HAVE_DNLC */ } } ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zdp); *ipp = NULL; if (flags & LOOKUP_XATTR) { /* * We don't allow recursive attributes.. * Maybe someday we will. */ if (zdp->z_pflags & ZFS_XATTR) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } if ((error = zfs_get_xattrdir(zdp, ipp, cr, flags))) { ZFS_EXIT(zfsvfs); return (error); } /* * Do we have permission to get into attribute directory? */ if ((error = zfs_zaccess(ITOZ(*ipp), ACE_EXECUTE, 0, B_FALSE, cr))) { iput(*ipp); *ipp = NULL; } ZFS_EXIT(zfsvfs); return (error); } if (!S_ISDIR(dip->i_mode)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(ENOTDIR)); } /* * Check accessibility of directory. */ if ((error = zfs_zaccess(zdp, ACE_EXECUTE, 0, B_FALSE, cr))) { ZFS_EXIT(zfsvfs); return (error); } if (zfsvfs->z_utf8 && u8_validate(nm, strlen(nm), NULL, U8_VALIDATE_ENTIRE, &error) < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EILSEQ)); } error = zfs_dirlook(zdp, nm, ipp, flags, direntflags, realpnp); if ((error == 0) && (*ipp)) zfs_inode_update(ITOZ(*ipp)); ZFS_EXIT(zfsvfs); return (error); } /* * Attempt to create a new entry in a directory. If the entry * already exists, truncate the file if permissible, else return * an error. Return the ip of the created or trunc'd file. * * IN: dip - inode of directory to put new file entry in. * name - name of new file entry. * vap - attributes of new file. * excl - flag indicating exclusive or non-exclusive mode. * mode - mode to open file with. * cr - credentials of caller. * flag - large file flag [UNUSED]. * vsecp - ACL to be set * * OUT: ipp - inode of created or trunc'd entry. * * RETURN: 0 on success, error code on failure. * * Timestamps: * dip - ctime|mtime updated if new entry created * ip - ctime|mtime always, atime if new */ /* ARGSUSED */ int zfs_create(struct inode *dip, char *name, vattr_t *vap, int excl, int mode, struct inode **ipp, cred_t *cr, int flag, vsecattr_t *vsecp) { znode_t *zp, *dzp = ITOZ(dip); zfsvfs_t *zfsvfs = ITOZSB(dip); zilog_t *zilog; objset_t *os; zfs_dirlock_t *dl; dmu_tx_t *tx; int error; uid_t uid; gid_t gid; zfs_acl_ids_t acl_ids; boolean_t fuid_dirtied; boolean_t have_acl = B_FALSE; boolean_t waited = B_FALSE; /* * If we have an ephemeral id, ACL, or XVATTR then * make sure file system is at proper version */ gid = crgetgid(cr); uid = crgetuid(cr); if (zfsvfs->z_use_fuids == B_FALSE && (vsecp || IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid))) return (SET_ERROR(EINVAL)); if (name == NULL) return (SET_ERROR(EINVAL)); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(dzp); os = zfsvfs->z_os; zilog = zfsvfs->z_log; if (zfsvfs->z_utf8 && u8_validate(name, strlen(name), NULL, U8_VALIDATE_ENTIRE, &error) < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EILSEQ)); } if (vap->va_mask & ATTR_XVATTR) { if ((error = secpolicy_xvattr((xvattr_t *)vap, crgetuid(cr), cr, vap->va_mode)) != 0) { ZFS_EXIT(zfsvfs); return (error); } } top: *ipp = NULL; if (*name == '\0') { /* * Null component name refers to the directory itself. */ igrab(dip); zp = dzp; dl = NULL; error = 0; } else { /* possible igrab(zp) */ int zflg = 0; if (flag & FIGNORECASE) zflg |= ZCILOOK; error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg, NULL, NULL); if (error) { if (have_acl) zfs_acl_ids_free(&acl_ids); if (strcmp(name, "..") == 0) error = SET_ERROR(EISDIR); ZFS_EXIT(zfsvfs); return (error); } } if (zp == NULL) { uint64_t txtype; /* * Create a new file object and update the directory * to reference it. */ if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) { if (have_acl) zfs_acl_ids_free(&acl_ids); goto out; } /* * We only support the creation of regular files in * extended attribute directories. */ if ((dzp->z_pflags & ZFS_XATTR) && !S_ISREG(vap->va_mode)) { if (have_acl) zfs_acl_ids_free(&acl_ids); error = SET_ERROR(EINVAL); goto out; } if (!have_acl && (error = zfs_acl_ids_create(dzp, 0, vap, cr, vsecp, &acl_ids)) != 0) goto out; have_acl = B_TRUE; if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) { zfs_acl_ids_free(&acl_ids); error = SET_ERROR(EDQUOT); goto out; } tx = dmu_tx_create(os); dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes + ZFS_SA_BASE_ATTR_SIZE); fuid_dirtied = zfsvfs->z_fuid_dirty; if (fuid_dirtied) zfs_fuid_txhold(zfsvfs, tx); dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name); dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE); if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) { dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, acl_ids.z_aclp->z_acl_bytes); } error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { zfs_dirent_unlock(dl); if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); goto top; } zfs_acl_ids_free(&acl_ids); dmu_tx_abort(tx); ZFS_EXIT(zfsvfs); return (error); } zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids); if (fuid_dirtied) zfs_fuid_sync(zfsvfs, tx); (void) zfs_link_create(dl, zp, tx, ZNEW); txtype = zfs_log_create_txtype(Z_FILE, vsecp, vap); if (flag & FIGNORECASE) txtype |= TX_CI; zfs_log_create(zilog, tx, txtype, dzp, zp, name, vsecp, acl_ids.z_fuidp, vap); zfs_acl_ids_free(&acl_ids); dmu_tx_commit(tx); } else { int aflags = (flag & FAPPEND) ? V_APPEND : 0; if (have_acl) zfs_acl_ids_free(&acl_ids); have_acl = B_FALSE; /* * A directory entry already exists for this name. */ /* * Can't truncate an existing file if in exclusive mode. */ if (excl) { error = SET_ERROR(EEXIST); goto out; } /* * Can't open a directory for writing. */ if (S_ISDIR(ZTOI(zp)->i_mode)) { error = SET_ERROR(EISDIR); goto out; } /* * Verify requested access to file. */ if (mode && (error = zfs_zaccess_rwx(zp, mode, aflags, cr))) { goto out; } mutex_enter(&dzp->z_lock); dzp->z_seq++; mutex_exit(&dzp->z_lock); /* * Truncate regular files if requested. */ if (S_ISREG(ZTOI(zp)->i_mode) && (vap->va_mask & ATTR_SIZE) && (vap->va_size == 0)) { /* we can't hold any locks when calling zfs_freesp() */ if (dl) { zfs_dirent_unlock(dl); dl = NULL; } error = zfs_freesp(zp, 0, 0, mode, TRUE); } } out: if (dl) zfs_dirent_unlock(dl); if (error) { if (zp) iput(ZTOI(zp)); } else { zfs_inode_update(dzp); zfs_inode_update(zp); *ipp = ZTOI(zp); } if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); ZFS_EXIT(zfsvfs); return (error); } /* ARGSUSED */ int zfs_tmpfile(struct inode *dip, vattr_t *vap, int excl, int mode, struct inode **ipp, cred_t *cr, int flag, vsecattr_t *vsecp) { znode_t *zp = NULL, *dzp = ITOZ(dip); zfsvfs_t *zfsvfs = ITOZSB(dip); objset_t *os; dmu_tx_t *tx; int error; uid_t uid; gid_t gid; zfs_acl_ids_t acl_ids; boolean_t fuid_dirtied; boolean_t have_acl = B_FALSE; boolean_t waited = B_FALSE; /* * If we have an ephemeral id, ACL, or XVATTR then * make sure file system is at proper version */ gid = crgetgid(cr); uid = crgetuid(cr); if (zfsvfs->z_use_fuids == B_FALSE && (vsecp || IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid))) return (SET_ERROR(EINVAL)); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(dzp); os = zfsvfs->z_os; if (vap->va_mask & ATTR_XVATTR) { if ((error = secpolicy_xvattr((xvattr_t *)vap, crgetuid(cr), cr, vap->va_mode)) != 0) { ZFS_EXIT(zfsvfs); return (error); } } top: *ipp = NULL; /* * Create a new file object and update the directory * to reference it. */ if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) { if (have_acl) zfs_acl_ids_free(&acl_ids); goto out; } if (!have_acl && (error = zfs_acl_ids_create(dzp, 0, vap, cr, vsecp, &acl_ids)) != 0) goto out; have_acl = B_TRUE; if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) { zfs_acl_ids_free(&acl_ids); error = SET_ERROR(EDQUOT); goto out; } tx = dmu_tx_create(os); dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes + ZFS_SA_BASE_ATTR_SIZE); dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL); fuid_dirtied = zfsvfs->z_fuid_dirty; if (fuid_dirtied) zfs_fuid_txhold(zfsvfs, tx); if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) { dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, acl_ids.z_aclp->z_acl_bytes); } error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); goto top; } zfs_acl_ids_free(&acl_ids); dmu_tx_abort(tx); ZFS_EXIT(zfsvfs); return (error); } zfs_mknode(dzp, vap, tx, cr, IS_TMPFILE, &zp, &acl_ids); if (fuid_dirtied) zfs_fuid_sync(zfsvfs, tx); /* Add to unlinked set */ zp->z_unlinked = 1; zfs_unlinked_add(zp, tx); zfs_acl_ids_free(&acl_ids); dmu_tx_commit(tx); out: if (error) { if (zp) iput(ZTOI(zp)); } else { zfs_inode_update(dzp); zfs_inode_update(zp); *ipp = ZTOI(zp); } ZFS_EXIT(zfsvfs); return (error); } /* * Remove an entry from a directory. * * IN: dip - inode of directory to remove entry from. * name - name of entry to remove. * cr - credentials of caller. * * RETURN: 0 if success * error code if failure * * Timestamps: * dip - ctime|mtime * ip - ctime (if nlink > 0) */ uint64_t null_xattr = 0; /*ARGSUSED*/ int zfs_remove(struct inode *dip, char *name, cred_t *cr, int flags) { znode_t *zp, *dzp = ITOZ(dip); znode_t *xzp; struct inode *ip; zfsvfs_t *zfsvfs = ITOZSB(dip); zilog_t *zilog; uint64_t acl_obj, xattr_obj; uint64_t xattr_obj_unlinked = 0; uint64_t obj = 0; uint64_t links; zfs_dirlock_t *dl; dmu_tx_t *tx; boolean_t may_delete_now, delete_now = FALSE; boolean_t unlinked, toobig = FALSE; uint64_t txtype; pathname_t *realnmp = NULL; pathname_t realnm; int error; int zflg = ZEXISTS; boolean_t waited = B_FALSE; if (name == NULL) return (SET_ERROR(EINVAL)); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(dzp); zilog = zfsvfs->z_log; if (flags & FIGNORECASE) { zflg |= ZCILOOK; pn_alloc(&realnm); realnmp = &realnm; } top: xattr_obj = 0; xzp = NULL; /* * Attempt to lock directory; fail if entry doesn't exist. */ if ((error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg, NULL, realnmp))) { if (realnmp) pn_free(realnmp); ZFS_EXIT(zfsvfs); return (error); } ip = ZTOI(zp); if ((error = zfs_zaccess_delete(dzp, zp, cr))) { goto out; } /* * Need to use rmdir for removing directories. */ if (S_ISDIR(ip->i_mode)) { error = SET_ERROR(EPERM); goto out; } #ifdef HAVE_DNLC if (realnmp) dnlc_remove(dvp, realnmp->pn_buf); else dnlc_remove(dvp, name); #endif /* HAVE_DNLC */ mutex_enter(&zp->z_lock); may_delete_now = atomic_read(&ip->i_count) == 1 && !(zp->z_is_mapped); mutex_exit(&zp->z_lock); /* * We may delete the znode now, or we may put it in the unlinked set; * it depends on whether we're the last link, and on whether there are * other holds on the inode. So we dmu_tx_hold() the right things to * allow for either case. */ obj = zp->z_id; tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_zap(tx, dzp->z_id, FALSE, name); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, zp); zfs_sa_upgrade_txholds(tx, dzp); if (may_delete_now) { toobig = zp->z_size > zp->z_blksz * zfs_delete_blocks; /* if the file is too big, only hold_free a token amount */ dmu_tx_hold_free(tx, zp->z_id, 0, (toobig ? DMU_MAX_ACCESS : DMU_OBJECT_END)); } /* are there any extended attributes? */ error = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), &xattr_obj, sizeof (xattr_obj)); if (error == 0 && xattr_obj) { error = zfs_zget(zfsvfs, xattr_obj, &xzp); ASSERT0(error); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE); dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE); } mutex_enter(&zp->z_lock); if ((acl_obj = zfs_external_acl(zp)) != 0 && may_delete_now) dmu_tx_hold_free(tx, acl_obj, 0, DMU_OBJECT_END); mutex_exit(&zp->z_lock); /* charge as an update -- would be nice not to charge at all */ dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL); /* * Mark this transaction as typically resulting in a net free of space */ dmu_tx_mark_netfree(tx); error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { zfs_dirent_unlock(dl); if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); iput(ip); if (xzp) iput(ZTOI(xzp)); goto top; } if (realnmp) pn_free(realnmp); dmu_tx_abort(tx); iput(ip); if (xzp) iput(ZTOI(xzp)); ZFS_EXIT(zfsvfs); return (error); } /* * Remove the directory entry. */ error = zfs_link_destroy(dl, zp, tx, zflg, &unlinked); if (error) { dmu_tx_commit(tx); goto out; } if (unlinked) { /* * Hold z_lock so that we can make sure that the ACL obj * hasn't changed. Could have been deleted due to * zfs_sa_upgrade(). */ mutex_enter(&zp->z_lock); (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), &xattr_obj_unlinked, sizeof (xattr_obj_unlinked)); delete_now = may_delete_now && !toobig && atomic_read(&ip->i_count) == 1 && !(zp->z_is_mapped) && xattr_obj == xattr_obj_unlinked && zfs_external_acl(zp) == acl_obj; } if (delete_now) { if (xattr_obj_unlinked) { ASSERT3U(ZTOI(xzp)->i_nlink, ==, 2); mutex_enter(&xzp->z_lock); xzp->z_unlinked = 1; clear_nlink(ZTOI(xzp)); links = 0; error = sa_update(xzp->z_sa_hdl, SA_ZPL_LINKS(zfsvfs), &links, sizeof (links), tx); ASSERT3U(error, ==, 0); mutex_exit(&xzp->z_lock); zfs_unlinked_add(xzp, tx); if (zp->z_is_sa) error = sa_remove(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), tx); else error = sa_update(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), &null_xattr, sizeof (uint64_t), tx); ASSERT0(error); } /* * Add to the unlinked set because a new reference could be * taken concurrently resulting in a deferred destruction. */ zfs_unlinked_add(zp, tx); mutex_exit(&zp->z_lock); } else if (unlinked) { mutex_exit(&zp->z_lock); zfs_unlinked_add(zp, tx); } txtype = TX_REMOVE; if (flags & FIGNORECASE) txtype |= TX_CI; zfs_log_remove(zilog, tx, txtype, dzp, name, obj); dmu_tx_commit(tx); out: if (realnmp) pn_free(realnmp); zfs_dirent_unlock(dl); zfs_inode_update(dzp); zfs_inode_update(zp); if (delete_now) iput(ip); else zfs_iput_async(ip); if (xzp) { zfs_inode_update(xzp); zfs_iput_async(ZTOI(xzp)); } if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); ZFS_EXIT(zfsvfs); return (error); } /* * Create a new directory and insert it into dip using the name * provided. Return a pointer to the inserted directory. * * IN: dip - inode of directory to add subdir to. * dirname - name of new directory. * vap - attributes of new directory. * cr - credentials of caller. * vsecp - ACL to be set * * OUT: ipp - inode of created directory. * * RETURN: 0 if success * error code if failure * * Timestamps: * dip - ctime|mtime updated * ipp - ctime|mtime|atime updated */ /*ARGSUSED*/ int zfs_mkdir(struct inode *dip, char *dirname, vattr_t *vap, struct inode **ipp, cred_t *cr, int flags, vsecattr_t *vsecp) { znode_t *zp, *dzp = ITOZ(dip); zfsvfs_t *zfsvfs = ITOZSB(dip); zilog_t *zilog; zfs_dirlock_t *dl; uint64_t txtype; dmu_tx_t *tx; int error; int zf = ZNEW; uid_t uid; gid_t gid = crgetgid(cr); zfs_acl_ids_t acl_ids; boolean_t fuid_dirtied; boolean_t waited = B_FALSE; ASSERT(S_ISDIR(vap->va_mode)); /* * If we have an ephemeral id, ACL, or XVATTR then * make sure file system is at proper version */ uid = crgetuid(cr); if (zfsvfs->z_use_fuids == B_FALSE && (vsecp || IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid))) return (SET_ERROR(EINVAL)); if (dirname == NULL) return (SET_ERROR(EINVAL)); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(dzp); zilog = zfsvfs->z_log; if (dzp->z_pflags & ZFS_XATTR) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } if (zfsvfs->z_utf8 && u8_validate(dirname, strlen(dirname), NULL, U8_VALIDATE_ENTIRE, &error) < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EILSEQ)); } if (flags & FIGNORECASE) zf |= ZCILOOK; if (vap->va_mask & ATTR_XVATTR) { if ((error = secpolicy_xvattr((xvattr_t *)vap, crgetuid(cr), cr, vap->va_mode)) != 0) { ZFS_EXIT(zfsvfs); return (error); } } if ((error = zfs_acl_ids_create(dzp, 0, vap, cr, vsecp, &acl_ids)) != 0) { ZFS_EXIT(zfsvfs); return (error); } /* * First make sure the new directory doesn't exist. * * Existence is checked first to make sure we don't return * EACCES instead of EEXIST which can cause some applications * to fail. */ top: *ipp = NULL; if ((error = zfs_dirent_lock(&dl, dzp, dirname, &zp, zf, NULL, NULL))) { zfs_acl_ids_free(&acl_ids); ZFS_EXIT(zfsvfs); return (error); } if ((error = zfs_zaccess(dzp, ACE_ADD_SUBDIRECTORY, 0, B_FALSE, cr))) { zfs_acl_ids_free(&acl_ids); zfs_dirent_unlock(dl); ZFS_EXIT(zfsvfs); return (error); } if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) { zfs_acl_ids_free(&acl_ids); zfs_dirent_unlock(dl); ZFS_EXIT(zfsvfs); return (SET_ERROR(EDQUOT)); } /* * Add a new entry to the directory. */ tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_zap(tx, dzp->z_id, TRUE, dirname); dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL); fuid_dirtied = zfsvfs->z_fuid_dirty; if (fuid_dirtied) zfs_fuid_txhold(zfsvfs, tx); if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) { dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, acl_ids.z_aclp->z_acl_bytes); } dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes + ZFS_SA_BASE_ATTR_SIZE); error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { zfs_dirent_unlock(dl); if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); goto top; } zfs_acl_ids_free(&acl_ids); dmu_tx_abort(tx); ZFS_EXIT(zfsvfs); return (error); } /* * Create new node. */ zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids); if (fuid_dirtied) zfs_fuid_sync(zfsvfs, tx); /* * Now put new name in parent dir. */ (void) zfs_link_create(dl, zp, tx, ZNEW); *ipp = ZTOI(zp); txtype = zfs_log_create_txtype(Z_DIR, vsecp, vap); if (flags & FIGNORECASE) txtype |= TX_CI; zfs_log_create(zilog, tx, txtype, dzp, zp, dirname, vsecp, acl_ids.z_fuidp, vap); zfs_acl_ids_free(&acl_ids); dmu_tx_commit(tx); zfs_dirent_unlock(dl); if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); zfs_inode_update(dzp); zfs_inode_update(zp); ZFS_EXIT(zfsvfs); return (0); } /* * Remove a directory subdir entry. If the current working * directory is the same as the subdir to be removed, the * remove will fail. * * IN: dip - inode of directory to remove from. * name - name of directory to be removed. * cwd - inode of current working directory. * cr - credentials of caller. * flags - case flags * * RETURN: 0 on success, error code on failure. * * Timestamps: * dip - ctime|mtime updated */ /*ARGSUSED*/ int zfs_rmdir(struct inode *dip, char *name, struct inode *cwd, cred_t *cr, int flags) { znode_t *dzp = ITOZ(dip); znode_t *zp; struct inode *ip; zfsvfs_t *zfsvfs = ITOZSB(dip); zilog_t *zilog; zfs_dirlock_t *dl; dmu_tx_t *tx; int error; int zflg = ZEXISTS; boolean_t waited = B_FALSE; if (name == NULL) return (SET_ERROR(EINVAL)); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(dzp); zilog = zfsvfs->z_log; if (flags & FIGNORECASE) zflg |= ZCILOOK; top: zp = NULL; /* * Attempt to lock directory; fail if entry doesn't exist. */ if ((error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg, NULL, NULL))) { ZFS_EXIT(zfsvfs); return (error); } ip = ZTOI(zp); if ((error = zfs_zaccess_delete(dzp, zp, cr))) { goto out; } if (!S_ISDIR(ip->i_mode)) { error = SET_ERROR(ENOTDIR); goto out; } if (ip == cwd) { error = SET_ERROR(EINVAL); goto out; } /* * Grab a lock on the directory to make sure that no one is * trying to add (or lookup) entries while we are removing it. */ rw_enter(&zp->z_name_lock, RW_WRITER); /* * Grab a lock on the parent pointer to make sure we play well * with the treewalk and directory rename code. */ rw_enter(&zp->z_parent_lock, RW_WRITER); tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_zap(tx, dzp->z_id, FALSE, name); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL); zfs_sa_upgrade_txholds(tx, zp); zfs_sa_upgrade_txholds(tx, dzp); dmu_tx_mark_netfree(tx); error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { rw_exit(&zp->z_parent_lock); rw_exit(&zp->z_name_lock); zfs_dirent_unlock(dl); if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); iput(ip); goto top; } dmu_tx_abort(tx); iput(ip); ZFS_EXIT(zfsvfs); return (error); } error = zfs_link_destroy(dl, zp, tx, zflg, NULL); if (error == 0) { uint64_t txtype = TX_RMDIR; if (flags & FIGNORECASE) txtype |= TX_CI; zfs_log_remove(zilog, tx, txtype, dzp, name, ZFS_NO_OBJECT); } dmu_tx_commit(tx); rw_exit(&zp->z_parent_lock); rw_exit(&zp->z_name_lock); out: zfs_dirent_unlock(dl); zfs_inode_update(dzp); zfs_inode_update(zp); iput(ip); if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); ZFS_EXIT(zfsvfs); return (error); } /* * Read as many directory entries as will fit into the provided * dirent buffer from the given directory cursor position. * * IN: ip - inode of directory to read. * dirent - buffer for directory entries. * * OUT: dirent - filler buffer of directory entries. * * RETURN: 0 if success * error code if failure * * Timestamps: * ip - atime updated * * Note that the low 4 bits of the cookie returned by zap is always zero. * This allows us to use the low range for "special" directory entries: * We use 0 for '.', and 1 for '..'. If this is the root of the filesystem, * we use the offset 2 for the '.zfs' directory. */ /* ARGSUSED */ int zfs_readdir(struct inode *ip, struct dir_context *ctx, cred_t *cr) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); objset_t *os; zap_cursor_t zc; zap_attribute_t zap; int error; uint8_t prefetch; uint8_t type; int done = 0; uint64_t parent; uint64_t offset; /* must be unsigned; checks for < 1 */ ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs), &parent, sizeof (parent))) != 0) goto out; /* * Quit if directory has been removed (posix) */ if (zp->z_unlinked) goto out; error = 0; os = zfsvfs->z_os; offset = ctx->pos; prefetch = zp->z_zn_prefetch; /* * Initialize the iterator cursor. */ if (offset <= 3) { /* * Start iteration from the beginning of the directory. */ zap_cursor_init(&zc, os, zp->z_id); } else { /* * The offset is a serialized cursor. */ zap_cursor_init_serialized(&zc, os, zp->z_id, offset); } /* * Transform to file-system independent format */ while (!done) { uint64_t objnum; /* * Special case `.', `..', and `.zfs'. */ if (offset == 0) { (void) strcpy(zap.za_name, "."); zap.za_normalization_conflict = 0; objnum = zp->z_id; type = DT_DIR; } else if (offset == 1) { (void) strcpy(zap.za_name, ".."); zap.za_normalization_conflict = 0; objnum = parent; type = DT_DIR; } else if (offset == 2 && zfs_show_ctldir(zp)) { (void) strcpy(zap.za_name, ZFS_CTLDIR_NAME); zap.za_normalization_conflict = 0; objnum = ZFSCTL_INO_ROOT; type = DT_DIR; } else { /* * Grab next entry. */ if ((error = zap_cursor_retrieve(&zc, &zap))) { if (error == ENOENT) break; else goto update; } /* * Allow multiple entries provided the first entry is * the object id. Non-zpl consumers may safely make * use of the additional space. * * XXX: This should be a feature flag for compatibility */ if (zap.za_integer_length != 8 || zap.za_num_integers == 0) { cmn_err(CE_WARN, "zap_readdir: bad directory " "entry, obj = %lld, offset = %lld, " "length = %d, num = %lld\n", (u_longlong_t)zp->z_id, (u_longlong_t)offset, zap.za_integer_length, (u_longlong_t)zap.za_num_integers); error = SET_ERROR(ENXIO); goto update; } objnum = ZFS_DIRENT_OBJ(zap.za_first_integer); type = ZFS_DIRENT_TYPE(zap.za_first_integer); } done = !dir_emit(ctx, zap.za_name, strlen(zap.za_name), objnum, type); if (done) break; /* Prefetch znode */ if (prefetch) { dmu_prefetch(os, objnum, 0, 0, 0, ZIO_PRIORITY_SYNC_READ); } /* * Move to the next entry, fill in the previous offset. */ if (offset > 2 || (offset == 2 && !zfs_show_ctldir(zp))) { zap_cursor_advance(&zc); offset = zap_cursor_serialize(&zc); } else { offset += 1; } ctx->pos = offset; } zp->z_zn_prefetch = B_FALSE; /* a lookup will re-enable pre-fetching */ update: zap_cursor_fini(&zc); if (error == ENOENT) error = 0; out: ZFS_EXIT(zfsvfs); return (error); } ulong_t zfs_fsync_sync_cnt = 4; int zfs_fsync(struct inode *ip, int syncflag, cred_t *cr) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); (void) tsd_set(zfs_fsyncer_key, (void *)zfs_fsync_sync_cnt); if (zfsvfs->z_os->os_sync != ZFS_SYNC_DISABLED) { ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); zil_commit(zfsvfs->z_log, zp->z_id); ZFS_EXIT(zfsvfs); } tsd_set(zfs_fsyncer_key, NULL); return (0); } /* * Get the requested file attributes and place them in the provided * vattr structure. * * IN: ip - inode of file. * vap - va_mask identifies requested attributes. * If ATTR_XVATTR set, then optional attrs are requested * flags - ATTR_NOACLCHECK (CIFS server context) * cr - credentials of caller. * * OUT: vap - attribute values. * * RETURN: 0 (always succeeds) */ /* ARGSUSED */ int zfs_getattr(struct inode *ip, vattr_t *vap, int flags, cred_t *cr) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); int error = 0; uint64_t links; uint64_t atime[2], mtime[2], ctime[2]; xvattr_t *xvap = (xvattr_t *)vap; /* vap may be an xvattr_t * */ xoptattr_t *xoap = NULL; boolean_t skipaclchk = (flags & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE; sa_bulk_attr_t bulk[3]; int count = 0; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); zfs_fuid_map_ids(zp, cr, &vap->va_uid, &vap->va_gid); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL, &atime, 16); 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); if ((error = sa_bulk_lookup(zp->z_sa_hdl, bulk, count)) != 0) { ZFS_EXIT(zfsvfs); return (error); } /* * If ACL is trivial don't bother looking for ACE_READ_ATTRIBUTES. * Also, if we are the owner don't bother, since owner should * always be allowed to read basic attributes of file. */ if (!(zp->z_pflags & ZFS_ACL_TRIVIAL) && (vap->va_uid != crgetuid(cr))) { if ((error = zfs_zaccess(zp, ACE_READ_ATTRIBUTES, 0, skipaclchk, cr))) { ZFS_EXIT(zfsvfs); return (error); } } /* * Return all attributes. It's cheaper to provide the answer * than to determine whether we were asked the question. */ mutex_enter(&zp->z_lock); vap->va_type = vn_mode_to_vtype(zp->z_mode); vap->va_mode = zp->z_mode; vap->va_fsid = ZTOI(zp)->i_sb->s_dev; vap->va_nodeid = zp->z_id; if ((zp->z_id == zfsvfs->z_root) && zfs_show_ctldir(zp)) links = ZTOI(zp)->i_nlink + 1; else links = ZTOI(zp)->i_nlink; vap->va_nlink = MIN(links, ZFS_LINK_MAX); vap->va_size = i_size_read(ip); vap->va_rdev = ip->i_rdev; vap->va_seq = ip->i_generation; /* * Add in any requested optional attributes and the create time. * Also set the corresponding bits in the returned attribute bitmap. */ if ((xoap = xva_getxoptattr(xvap)) != NULL && zfsvfs->z_use_fuids) { if (XVA_ISSET_REQ(xvap, XAT_ARCHIVE)) { xoap->xoa_archive = ((zp->z_pflags & ZFS_ARCHIVE) != 0); XVA_SET_RTN(xvap, XAT_ARCHIVE); } if (XVA_ISSET_REQ(xvap, XAT_READONLY)) { xoap->xoa_readonly = ((zp->z_pflags & ZFS_READONLY) != 0); XVA_SET_RTN(xvap, XAT_READONLY); } if (XVA_ISSET_REQ(xvap, XAT_SYSTEM)) { xoap->xoa_system = ((zp->z_pflags & ZFS_SYSTEM) != 0); XVA_SET_RTN(xvap, XAT_SYSTEM); } if (XVA_ISSET_REQ(xvap, XAT_HIDDEN)) { xoap->xoa_hidden = ((zp->z_pflags & ZFS_HIDDEN) != 0); XVA_SET_RTN(xvap, XAT_HIDDEN); } if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK)) { xoap->xoa_nounlink = ((zp->z_pflags & ZFS_NOUNLINK) != 0); XVA_SET_RTN(xvap, XAT_NOUNLINK); } if (XVA_ISSET_REQ(xvap, XAT_IMMUTABLE)) { xoap->xoa_immutable = ((zp->z_pflags & ZFS_IMMUTABLE) != 0); XVA_SET_RTN(xvap, XAT_IMMUTABLE); } if (XVA_ISSET_REQ(xvap, XAT_APPENDONLY)) { xoap->xoa_appendonly = ((zp->z_pflags & ZFS_APPENDONLY) != 0); XVA_SET_RTN(xvap, XAT_APPENDONLY); } if (XVA_ISSET_REQ(xvap, XAT_NODUMP)) { xoap->xoa_nodump = ((zp->z_pflags & ZFS_NODUMP) != 0); XVA_SET_RTN(xvap, XAT_NODUMP); } if (XVA_ISSET_REQ(xvap, XAT_OPAQUE)) { xoap->xoa_opaque = ((zp->z_pflags & ZFS_OPAQUE) != 0); XVA_SET_RTN(xvap, XAT_OPAQUE); } if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED)) { xoap->xoa_av_quarantined = ((zp->z_pflags & ZFS_AV_QUARANTINED) != 0); XVA_SET_RTN(xvap, XAT_AV_QUARANTINED); } if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED)) { xoap->xoa_av_modified = ((zp->z_pflags & ZFS_AV_MODIFIED) != 0); XVA_SET_RTN(xvap, XAT_AV_MODIFIED); } if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP) && S_ISREG(ip->i_mode)) { zfs_sa_get_scanstamp(zp, xvap); } if (XVA_ISSET_REQ(xvap, XAT_CREATETIME)) { uint64_t times[2]; (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_CRTIME(zfsvfs), times, sizeof (times)); ZFS_TIME_DECODE(&xoap->xoa_createtime, times); XVA_SET_RTN(xvap, XAT_CREATETIME); } if (XVA_ISSET_REQ(xvap, XAT_REPARSE)) { xoap->xoa_reparse = ((zp->z_pflags & ZFS_REPARSE) != 0); XVA_SET_RTN(xvap, XAT_REPARSE); } if (XVA_ISSET_REQ(xvap, XAT_GEN)) { xoap->xoa_generation = ip->i_generation; XVA_SET_RTN(xvap, XAT_GEN); } if (XVA_ISSET_REQ(xvap, XAT_OFFLINE)) { xoap->xoa_offline = ((zp->z_pflags & ZFS_OFFLINE) != 0); XVA_SET_RTN(xvap, XAT_OFFLINE); } if (XVA_ISSET_REQ(xvap, XAT_SPARSE)) { xoap->xoa_sparse = ((zp->z_pflags & ZFS_SPARSE) != 0); XVA_SET_RTN(xvap, XAT_SPARSE); } } ZFS_TIME_DECODE(&vap->va_atime, atime); ZFS_TIME_DECODE(&vap->va_mtime, mtime); ZFS_TIME_DECODE(&vap->va_ctime, ctime); mutex_exit(&zp->z_lock); sa_object_size(zp->z_sa_hdl, &vap->va_blksize, &vap->va_nblocks); if (zp->z_blksz == 0) { /* * Block size hasn't been set; suggest maximal I/O transfers. */ vap->va_blksize = zfsvfs->z_max_blksz; } ZFS_EXIT(zfsvfs); return (0); } /* * Get the basic file attributes and place them in the provided kstat * structure. The inode is assumed to be the authoritative source * for most of the attributes. However, the znode currently has the * authoritative atime, blksize, and block count. * * IN: ip - inode of file. * * OUT: sp - kstat values. * * RETURN: 0 (always succeeds) */ /* ARGSUSED */ int zfs_getattr_fast(struct inode *ip, struct kstat *sp) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); uint32_t blksize; u_longlong_t nblocks; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); mutex_enter(&zp->z_lock); generic_fillattr(ip, sp); sa_object_size(zp->z_sa_hdl, &blksize, &nblocks); sp->blksize = blksize; sp->blocks = nblocks; if (unlikely(zp->z_blksz == 0)) { /* * Block size hasn't been set; suggest maximal I/O transfers. */ sp->blksize = zfsvfs->z_max_blksz; } mutex_exit(&zp->z_lock); /* * Required to prevent NFS client from detecting different inode * numbers of snapshot root dentry before and after snapshot mount. */ if (zfsvfs->z_issnap) { if (ip->i_sb->s_root->d_inode == ip) sp->ino = ZFSCTL_INO_SNAPDIRS - dmu_objset_id(zfsvfs->z_os); } ZFS_EXIT(zfsvfs); return (0); } /* * Set the file attributes to the values contained in the * vattr structure. * * IN: ip - inode of file to be modified. * vap - new attribute values. * If ATTR_XVATTR set, then optional attrs are being set * flags - ATTR_UTIME set if non-default time values provided. * - ATTR_NOACLCHECK (CIFS context only). * cr - credentials of caller. * * RETURN: 0 if success * error code if failure * * Timestamps: * ip - ctime updated, mtime updated if size changed. */ /* ARGSUSED */ int zfs_setattr(struct inode *ip, vattr_t *vap, int flags, cred_t *cr) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); zilog_t *zilog; dmu_tx_t *tx; vattr_t oldva; xvattr_t *tmpxvattr; uint_t mask = vap->va_mask; uint_t saved_mask = 0; int trim_mask = 0; uint64_t new_mode; uint64_t new_kuid = 0, new_kgid = 0, new_uid, new_gid; uint64_t xattr_obj; uint64_t mtime[2], ctime[2], atime[2]; znode_t *attrzp; int need_policy = FALSE; int err, err2; zfs_fuid_info_t *fuidp = NULL; xvattr_t *xvap = (xvattr_t *)vap; /* vap may be an xvattr_t * */ xoptattr_t *xoap; zfs_acl_t *aclp; boolean_t skipaclchk = (flags & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE; boolean_t fuid_dirtied = B_FALSE; sa_bulk_attr_t *bulk, *xattr_bulk; int count = 0, xattr_count = 0; if (mask == 0) return (0); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); zilog = zfsvfs->z_log; /* * Make sure that if we have ephemeral uid/gid or xvattr specified * that file system is at proper version level */ if (zfsvfs->z_use_fuids == B_FALSE && (((mask & ATTR_UID) && IS_EPHEMERAL(vap->va_uid)) || ((mask & ATTR_GID) && IS_EPHEMERAL(vap->va_gid)) || (mask & ATTR_XVATTR))) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } if (mask & ATTR_SIZE && S_ISDIR(ip->i_mode)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EISDIR)); } if (mask & ATTR_SIZE && !S_ISREG(ip->i_mode) && !S_ISFIFO(ip->i_mode)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } /* * If this is an xvattr_t, then get a pointer to the structure of * optional attributes. If this is NULL, then we have a vattr_t. */ xoap = xva_getxoptattr(xvap); tmpxvattr = kmem_alloc(sizeof (xvattr_t), KM_SLEEP); xva_init(tmpxvattr); bulk = kmem_alloc(sizeof (sa_bulk_attr_t) * 7, KM_SLEEP); xattr_bulk = kmem_alloc(sizeof (sa_bulk_attr_t) * 7, KM_SLEEP); /* * Immutable files can only alter immutable bit and atime */ if ((zp->z_pflags & ZFS_IMMUTABLE) && ((mask & (ATTR_SIZE|ATTR_UID|ATTR_GID|ATTR_MTIME|ATTR_MODE)) || ((mask & ATTR_XVATTR) && XVA_ISSET_REQ(xvap, XAT_CREATETIME)))) { err = SET_ERROR(EPERM); goto out3; } if ((mask & ATTR_SIZE) && (zp->z_pflags & ZFS_READONLY)) { err = SET_ERROR(EPERM); goto out3; } /* * Verify timestamps doesn't overflow 32 bits. * ZFS can handle large timestamps, but 32bit syscalls can't * handle times greater than 2039. This check should be removed * once large timestamps are fully supported. */ if (mask & (ATTR_ATIME | ATTR_MTIME)) { if (((mask & ATTR_ATIME) && TIMESPEC_OVERFLOW(&vap->va_atime)) || ((mask & ATTR_MTIME) && TIMESPEC_OVERFLOW(&vap->va_mtime))) { err = SET_ERROR(EOVERFLOW); goto out3; } } top: attrzp = NULL; aclp = NULL; /* Can this be moved to before the top label? */ if (zfs_is_readonly(zfsvfs)) { err = SET_ERROR(EROFS); goto out3; } /* * First validate permissions */ if (mask & ATTR_SIZE) { err = zfs_zaccess(zp, ACE_WRITE_DATA, 0, skipaclchk, cr); if (err) goto out3; /* * XXX - Note, we are not providing any open * mode flags here (like FNDELAY), so we may * block if there are locks present... this * should be addressed in openat(). */ /* XXX - would it be OK to generate a log record here? */ err = zfs_freesp(zp, vap->va_size, 0, 0, FALSE); if (err) goto out3; } if (mask & (ATTR_ATIME|ATTR_MTIME) || ((mask & ATTR_XVATTR) && (XVA_ISSET_REQ(xvap, XAT_HIDDEN) || XVA_ISSET_REQ(xvap, XAT_READONLY) || XVA_ISSET_REQ(xvap, XAT_ARCHIVE) || XVA_ISSET_REQ(xvap, XAT_OFFLINE) || XVA_ISSET_REQ(xvap, XAT_SPARSE) || XVA_ISSET_REQ(xvap, XAT_CREATETIME) || XVA_ISSET_REQ(xvap, XAT_SYSTEM)))) { need_policy = zfs_zaccess(zp, ACE_WRITE_ATTRIBUTES, 0, skipaclchk, cr); } if (mask & (ATTR_UID|ATTR_GID)) { int idmask = (mask & (ATTR_UID|ATTR_GID)); int take_owner; int take_group; /* * NOTE: even if a new mode is being set, * we may clear S_ISUID/S_ISGID bits. */ if (!(mask & ATTR_MODE)) vap->va_mode = zp->z_mode; /* * Take ownership or chgrp to group we are a member of */ take_owner = (mask & ATTR_UID) && (vap->va_uid == crgetuid(cr)); take_group = (mask & ATTR_GID) && zfs_groupmember(zfsvfs, vap->va_gid, cr); /* * If both ATTR_UID and ATTR_GID are set then take_owner and * take_group must both be set in order to allow taking * ownership. * * Otherwise, send the check through secpolicy_vnode_setattr() * */ if (((idmask == (ATTR_UID|ATTR_GID)) && take_owner && take_group) || ((idmask == ATTR_UID) && take_owner) || ((idmask == ATTR_GID) && take_group)) { if (zfs_zaccess(zp, ACE_WRITE_OWNER, 0, skipaclchk, cr) == 0) { /* * Remove setuid/setgid for non-privileged users */ (void) secpolicy_setid_clear(vap, cr); trim_mask = (mask & (ATTR_UID|ATTR_GID)); } else { need_policy = TRUE; } } else { need_policy = TRUE; } } mutex_enter(&zp->z_lock); oldva.va_mode = zp->z_mode; zfs_fuid_map_ids(zp, cr, &oldva.va_uid, &oldva.va_gid); if (mask & ATTR_XVATTR) { /* * Update xvattr mask to include only those attributes * that are actually changing. * * the bits will be restored prior to actually setting * the attributes so the caller thinks they were set. */ if (XVA_ISSET_REQ(xvap, XAT_APPENDONLY)) { if (xoap->xoa_appendonly != ((zp->z_pflags & ZFS_APPENDONLY) != 0)) { need_policy = TRUE; } else { XVA_CLR_REQ(xvap, XAT_APPENDONLY); XVA_SET_REQ(tmpxvattr, XAT_APPENDONLY); } } if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK)) { if (xoap->xoa_nounlink != ((zp->z_pflags & ZFS_NOUNLINK) != 0)) { need_policy = TRUE; } else { XVA_CLR_REQ(xvap, XAT_NOUNLINK); XVA_SET_REQ(tmpxvattr, XAT_NOUNLINK); } } if (XVA_ISSET_REQ(xvap, XAT_IMMUTABLE)) { if (xoap->xoa_immutable != ((zp->z_pflags & ZFS_IMMUTABLE) != 0)) { need_policy = TRUE; } else { XVA_CLR_REQ(xvap, XAT_IMMUTABLE); XVA_SET_REQ(tmpxvattr, XAT_IMMUTABLE); } } if (XVA_ISSET_REQ(xvap, XAT_NODUMP)) { if (xoap->xoa_nodump != ((zp->z_pflags & ZFS_NODUMP) != 0)) { need_policy = TRUE; } else { XVA_CLR_REQ(xvap, XAT_NODUMP); XVA_SET_REQ(tmpxvattr, XAT_NODUMP); } } if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED)) { if (xoap->xoa_av_modified != ((zp->z_pflags & ZFS_AV_MODIFIED) != 0)) { need_policy = TRUE; } else { XVA_CLR_REQ(xvap, XAT_AV_MODIFIED); XVA_SET_REQ(tmpxvattr, XAT_AV_MODIFIED); } } if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED)) { if ((!S_ISREG(ip->i_mode) && xoap->xoa_av_quarantined) || xoap->xoa_av_quarantined != ((zp->z_pflags & ZFS_AV_QUARANTINED) != 0)) { need_policy = TRUE; } else { XVA_CLR_REQ(xvap, XAT_AV_QUARANTINED); XVA_SET_REQ(tmpxvattr, XAT_AV_QUARANTINED); } } if (XVA_ISSET_REQ(xvap, XAT_REPARSE)) { mutex_exit(&zp->z_lock); err = SET_ERROR(EPERM); goto out3; } if (need_policy == FALSE && (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP) || XVA_ISSET_REQ(xvap, XAT_OPAQUE))) { need_policy = TRUE; } } mutex_exit(&zp->z_lock); if (mask & ATTR_MODE) { if (zfs_zaccess(zp, ACE_WRITE_ACL, 0, skipaclchk, cr) == 0) { err = secpolicy_setid_setsticky_clear(ip, vap, &oldva, cr); if (err) goto out3; trim_mask |= ATTR_MODE; } else { need_policy = TRUE; } } if (need_policy) { /* * If trim_mask is set then take ownership * has been granted or write_acl is present and user * has the ability to modify mode. In that case remove * UID|GID and or MODE from mask so that * secpolicy_vnode_setattr() doesn't revoke it. */ if (trim_mask) { saved_mask = vap->va_mask; vap->va_mask &= ~trim_mask; } err = secpolicy_vnode_setattr(cr, ip, vap, &oldva, flags, (int (*)(void *, int, cred_t *))zfs_zaccess_unix, zp); if (err) goto out3; if (trim_mask) vap->va_mask |= saved_mask; } /* * secpolicy_vnode_setattr, or take ownership may have * changed va_mask */ mask = vap->va_mask; if ((mask & (ATTR_UID | ATTR_GID))) { err = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), &xattr_obj, sizeof (xattr_obj)); if (err == 0 && xattr_obj) { err = zfs_zget(ZTOZSB(zp), xattr_obj, &attrzp); if (err) goto out2; } if (mask & ATTR_UID) { new_kuid = zfs_fuid_create(zfsvfs, (uint64_t)vap->va_uid, cr, ZFS_OWNER, &fuidp); if (new_kuid != KUID_TO_SUID(ZTOI(zp)->i_uid) && zfs_fuid_overquota(zfsvfs, B_FALSE, new_kuid)) { if (attrzp) iput(ZTOI(attrzp)); err = SET_ERROR(EDQUOT); goto out2; } } if (mask & ATTR_GID) { new_kgid = zfs_fuid_create(zfsvfs, (uint64_t)vap->va_gid, cr, ZFS_GROUP, &fuidp); if (new_kgid != KGID_TO_SGID(ZTOI(zp)->i_gid) && zfs_fuid_overquota(zfsvfs, B_TRUE, new_kgid)) { if (attrzp) iput(ZTOI(attrzp)); err = SET_ERROR(EDQUOT); goto out2; } } } tx = dmu_tx_create(zfsvfs->z_os); if (mask & ATTR_MODE) { uint64_t pmode = zp->z_mode; uint64_t acl_obj; new_mode = (pmode & S_IFMT) | (vap->va_mode & ~S_IFMT); zfs_acl_chmod_setattr(zp, &aclp, new_mode); mutex_enter(&zp->z_lock); if (!zp->z_is_sa && ((acl_obj = zfs_external_acl(zp)) != 0)) { /* * Are we upgrading ACL from old V0 format * to V1 format? */ if (zfsvfs->z_version >= ZPL_VERSION_FUID && zfs_znode_acl_version(zp) == ZFS_ACL_VERSION_INITIAL) { dmu_tx_hold_free(tx, acl_obj, 0, DMU_OBJECT_END); dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, aclp->z_acl_bytes); } else { dmu_tx_hold_write(tx, acl_obj, 0, aclp->z_acl_bytes); } } else if (!zp->z_is_sa && aclp->z_acl_bytes > ZFS_ACE_SPACE) { dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, aclp->z_acl_bytes); } mutex_exit(&zp->z_lock); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE); } else { if ((mask & ATTR_XVATTR) && XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP)) dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE); else dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); } if (attrzp) { dmu_tx_hold_sa(tx, attrzp->z_sa_hdl, B_FALSE); } fuid_dirtied = zfsvfs->z_fuid_dirty; if (fuid_dirtied) zfs_fuid_txhold(zfsvfs, tx); zfs_sa_upgrade_txholds(tx, zp); err = dmu_tx_assign(tx, TXG_WAIT); if (err) goto out; count = 0; /* * Set each attribute requested. * We group settings according to the locks they need to acquire. * * Note: you cannot set ctime directly, although it will be * updated as a side-effect of calling this function. */ if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE)) mutex_enter(&zp->z_acl_lock); mutex_enter(&zp->z_lock); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL, &zp->z_pflags, sizeof (zp->z_pflags)); if (attrzp) { if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE)) mutex_enter(&attrzp->z_acl_lock); mutex_enter(&attrzp->z_lock); SA_ADD_BULK_ATTR(xattr_bulk, xattr_count, SA_ZPL_FLAGS(zfsvfs), NULL, &attrzp->z_pflags, sizeof (attrzp->z_pflags)); } if (mask & (ATTR_UID|ATTR_GID)) { if (mask & ATTR_UID) { ZTOI(zp)->i_uid = SUID_TO_KUID(new_kuid); new_uid = zfs_uid_read(ZTOI(zp)); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL, &new_uid, sizeof (new_uid)); if (attrzp) { SA_ADD_BULK_ATTR(xattr_bulk, xattr_count, SA_ZPL_UID(zfsvfs), NULL, &new_uid, sizeof (new_uid)); ZTOI(attrzp)->i_uid = SUID_TO_KUID(new_uid); } } if (mask & ATTR_GID) { ZTOI(zp)->i_gid = SGID_TO_KGID(new_kgid); new_gid = zfs_gid_read(ZTOI(zp)); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs), NULL, &new_gid, sizeof (new_gid)); if (attrzp) { SA_ADD_BULK_ATTR(xattr_bulk, xattr_count, SA_ZPL_GID(zfsvfs), NULL, &new_gid, sizeof (new_gid)); ZTOI(attrzp)->i_gid = SGID_TO_KGID(new_kgid); } } if (!(mask & ATTR_MODE)) { SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL, &new_mode, sizeof (new_mode)); new_mode = zp->z_mode; } err = zfs_acl_chown_setattr(zp); ASSERT(err == 0); if (attrzp) { err = zfs_acl_chown_setattr(attrzp); ASSERT(err == 0); } } if (mask & ATTR_MODE) { SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL, &new_mode, sizeof (new_mode)); zp->z_mode = ZTOI(zp)->i_mode = new_mode; ASSERT3P(aclp, !=, NULL); err = zfs_aclset_common(zp, aclp, cr, tx); ASSERT0(err); if (zp->z_acl_cached) zfs_acl_free(zp->z_acl_cached); zp->z_acl_cached = aclp; aclp = NULL; } if ((mask & ATTR_ATIME) || zp->z_atime_dirty) { zp->z_atime_dirty = 0; ZFS_TIME_ENCODE(&ip->i_atime, atime); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL, &atime, sizeof (atime)); } if (mask & (ATTR_MTIME | ATTR_SIZE)) { ZFS_TIME_ENCODE(&vap->va_mtime, mtime); ZTOI(zp)->i_mtime = timespec_trunc(vap->va_mtime, ZTOI(zp)->i_sb->s_time_gran); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL, mtime, sizeof (mtime)); } if (mask & (ATTR_CTIME | ATTR_SIZE)) { ZFS_TIME_ENCODE(&vap->va_ctime, ctime); ZTOI(zp)->i_ctime = timespec_trunc(vap->va_ctime, ZTOI(zp)->i_sb->s_time_gran); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL, ctime, sizeof (ctime)); } if (attrzp && mask) { SA_ADD_BULK_ATTR(xattr_bulk, xattr_count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, sizeof (ctime)); } /* * Do this after setting timestamps to prevent timestamp * update from toggling bit */ if (xoap && (mask & ATTR_XVATTR)) { /* * restore trimmed off masks * so that return masks can be set for caller. */ if (XVA_ISSET_REQ(tmpxvattr, XAT_APPENDONLY)) { XVA_SET_REQ(xvap, XAT_APPENDONLY); } if (XVA_ISSET_REQ(tmpxvattr, XAT_NOUNLINK)) { XVA_SET_REQ(xvap, XAT_NOUNLINK); } if (XVA_ISSET_REQ(tmpxvattr, XAT_IMMUTABLE)) { XVA_SET_REQ(xvap, XAT_IMMUTABLE); } if (XVA_ISSET_REQ(tmpxvattr, XAT_NODUMP)) { XVA_SET_REQ(xvap, XAT_NODUMP); } if (XVA_ISSET_REQ(tmpxvattr, XAT_AV_MODIFIED)) { XVA_SET_REQ(xvap, XAT_AV_MODIFIED); } if (XVA_ISSET_REQ(tmpxvattr, XAT_AV_QUARANTINED)) { XVA_SET_REQ(xvap, XAT_AV_QUARANTINED); } if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP)) ASSERT(S_ISREG(ip->i_mode)); zfs_xvattr_set(zp, xvap, tx); } if (fuid_dirtied) zfs_fuid_sync(zfsvfs, tx); if (mask != 0) zfs_log_setattr(zilog, tx, TX_SETATTR, zp, vap, mask, fuidp); mutex_exit(&zp->z_lock); if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE)) mutex_exit(&zp->z_acl_lock); if (attrzp) { if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE)) mutex_exit(&attrzp->z_acl_lock); mutex_exit(&attrzp->z_lock); } out: if (err == 0 && attrzp) { err2 = sa_bulk_update(attrzp->z_sa_hdl, xattr_bulk, xattr_count, tx); ASSERT(err2 == 0); } if (aclp) zfs_acl_free(aclp); if (fuidp) { zfs_fuid_info_free(fuidp); fuidp = NULL; } if (err) { dmu_tx_abort(tx); if (attrzp) iput(ZTOI(attrzp)); if (err == ERESTART) goto top; } else { err2 = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx); dmu_tx_commit(tx); if (attrzp) iput(ZTOI(attrzp)); zfs_inode_update(zp); } out2: if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); out3: kmem_free(xattr_bulk, sizeof (sa_bulk_attr_t) * 7); kmem_free(bulk, sizeof (sa_bulk_attr_t) * 7); kmem_free(tmpxvattr, sizeof (xvattr_t)); ZFS_EXIT(zfsvfs); return (err); } typedef struct zfs_zlock { krwlock_t *zl_rwlock; /* lock we acquired */ znode_t *zl_znode; /* znode we held */ struct zfs_zlock *zl_next; /* next in list */ } zfs_zlock_t; /* * Drop locks and release vnodes that were held by zfs_rename_lock(). */ static void zfs_rename_unlock(zfs_zlock_t **zlpp) { zfs_zlock_t *zl; while ((zl = *zlpp) != NULL) { if (zl->zl_znode != NULL) zfs_iput_async(ZTOI(zl->zl_znode)); rw_exit(zl->zl_rwlock); *zlpp = zl->zl_next; kmem_free(zl, sizeof (*zl)); } } /* * Search back through the directory tree, using the ".." entries. * Lock each directory in the chain to prevent concurrent renames. * Fail any attempt to move a directory into one of its own descendants. * XXX - z_parent_lock can overlap with map or grow locks */ static int zfs_rename_lock(znode_t *szp, znode_t *tdzp, znode_t *sdzp, zfs_zlock_t **zlpp) { zfs_zlock_t *zl; znode_t *zp = tdzp; uint64_t rootid = ZTOZSB(zp)->z_root; uint64_t oidp = zp->z_id; krwlock_t *rwlp = &szp->z_parent_lock; krw_t rw = RW_WRITER; /* * First pass write-locks szp and compares to zp->z_id. * Later passes read-lock zp and compare to zp->z_parent. */ do { if (!rw_tryenter(rwlp, rw)) { /* * Another thread is renaming in this path. * Note that if we are a WRITER, we don't have any * parent_locks held yet. */ if (rw == RW_READER && zp->z_id > szp->z_id) { /* * Drop our locks and restart */ zfs_rename_unlock(&zl); *zlpp = NULL; zp = tdzp; oidp = zp->z_id; rwlp = &szp->z_parent_lock; rw = RW_WRITER; continue; } else { /* * Wait for other thread to drop its locks */ rw_enter(rwlp, rw); } } zl = kmem_alloc(sizeof (*zl), KM_SLEEP); zl->zl_rwlock = rwlp; zl->zl_znode = NULL; zl->zl_next = *zlpp; *zlpp = zl; if (oidp == szp->z_id) /* We're a descendant of szp */ return (SET_ERROR(EINVAL)); if (oidp == rootid) /* We've hit the top */ return (0); if (rw == RW_READER) { /* i.e. not the first pass */ int error = zfs_zget(ZTOZSB(zp), oidp, &zp); if (error) return (error); zl->zl_znode = zp; } (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(ZTOZSB(zp)), &oidp, sizeof (oidp)); rwlp = &zp->z_parent_lock; rw = RW_READER; } while (zp->z_id != sdzp->z_id); return (0); } /* * Move an entry from the provided source directory to the target * directory. Change the entry name as indicated. * * IN: sdip - Source directory containing the "old entry". * snm - Old entry name. * tdip - Target directory to contain the "new entry". * tnm - New entry name. * cr - credentials of caller. * flags - case flags * * RETURN: 0 on success, error code on failure. * * Timestamps: * sdip,tdip - ctime|mtime updated */ /*ARGSUSED*/ int zfs_rename(struct inode *sdip, char *snm, struct inode *tdip, char *tnm, cred_t *cr, int flags) { znode_t *tdzp, *szp, *tzp; znode_t *sdzp = ITOZ(sdip); zfsvfs_t *zfsvfs = ITOZSB(sdip); zilog_t *zilog; zfs_dirlock_t *sdl, *tdl; dmu_tx_t *tx; zfs_zlock_t *zl; int cmp, serr, terr; int error = 0; int zflg = 0; boolean_t waited = B_FALSE; if (snm == NULL || tnm == NULL) return (SET_ERROR(EINVAL)); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(sdzp); zilog = zfsvfs->z_log; tdzp = ITOZ(tdip); ZFS_VERIFY_ZP(tdzp); /* * We check i_sb because snapshots and the ctldir must have different * super blocks. */ if (tdip->i_sb != sdip->i_sb || zfsctl_is_node(tdip)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EXDEV)); } if (zfsvfs->z_utf8 && u8_validate(tnm, strlen(tnm), NULL, U8_VALIDATE_ENTIRE, &error) < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EILSEQ)); } if (flags & FIGNORECASE) zflg |= ZCILOOK; top: szp = NULL; tzp = NULL; zl = NULL; /* * This is to prevent the creation of links into attribute space * by renaming a linked file into/outof an attribute directory. * See the comment in zfs_link() for why this is considered bad. */ if ((tdzp->z_pflags & ZFS_XATTR) != (sdzp->z_pflags & ZFS_XATTR)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } /* * Lock source and target directory entries. To prevent deadlock, * a lock ordering must be defined. We lock the directory with * the smallest object id first, or if it's a tie, the one with * the lexically first name. */ if (sdzp->z_id < tdzp->z_id) { cmp = -1; } else if (sdzp->z_id > tdzp->z_id) { cmp = 1; } else { /* * First compare the two name arguments without * considering any case folding. */ int nofold = (zfsvfs->z_norm & ~U8_TEXTPREP_TOUPPER); cmp = u8_strcmp(snm, tnm, 0, nofold, U8_UNICODE_LATEST, &error); ASSERT(error == 0 || !zfsvfs->z_utf8); if (cmp == 0) { /* * POSIX: "If the old argument and the new argument * both refer to links to the same existing file, * the rename() function shall return successfully * and perform no other action." */ ZFS_EXIT(zfsvfs); return (0); } /* * If the file system is case-folding, then we may * have some more checking to do. A case-folding file * system is either supporting mixed case sensitivity * access or is completely case-insensitive. Note * that the file system is always case preserving. * * In mixed sensitivity mode case sensitive behavior * is the default. FIGNORECASE must be used to * explicitly request case insensitive behavior. * * If the source and target names provided differ only * by case (e.g., a request to rename 'tim' to 'Tim'), * we will treat this as a special case in the * case-insensitive mode: as long as the source name * is an exact match, we will allow this to proceed as * a name-change request. */ if ((zfsvfs->z_case == ZFS_CASE_INSENSITIVE || (zfsvfs->z_case == ZFS_CASE_MIXED && flags & FIGNORECASE)) && u8_strcmp(snm, tnm, 0, zfsvfs->z_norm, U8_UNICODE_LATEST, &error) == 0) { /* * case preserving rename request, require exact * name matches */ zflg |= ZCIEXACT; zflg &= ~ZCILOOK; } } /* * If the source and destination directories are the same, we should * grab the z_name_lock of that directory only once. */ if (sdzp == tdzp) { zflg |= ZHAVELOCK; rw_enter(&sdzp->z_name_lock, RW_READER); } if (cmp < 0) { serr = zfs_dirent_lock(&sdl, sdzp, snm, &szp, ZEXISTS | zflg, NULL, NULL); terr = zfs_dirent_lock(&tdl, tdzp, tnm, &tzp, ZRENAMING | zflg, NULL, NULL); } else { terr = zfs_dirent_lock(&tdl, tdzp, tnm, &tzp, zflg, NULL, NULL); serr = zfs_dirent_lock(&sdl, sdzp, snm, &szp, ZEXISTS | ZRENAMING | zflg, NULL, NULL); } if (serr) { /* * Source entry invalid or not there. */ if (!terr) { zfs_dirent_unlock(tdl); if (tzp) iput(ZTOI(tzp)); } if (sdzp == tdzp) rw_exit(&sdzp->z_name_lock); if (strcmp(snm, "..") == 0) serr = EINVAL; ZFS_EXIT(zfsvfs); return (serr); } if (terr) { zfs_dirent_unlock(sdl); iput(ZTOI(szp)); if (sdzp == tdzp) rw_exit(&sdzp->z_name_lock); if (strcmp(tnm, "..") == 0) terr = EINVAL; ZFS_EXIT(zfsvfs); return (terr); } /* * Must have write access at the source to remove the old entry * and write access at the target to create the new entry. * Note that if target and source are the same, this can be * done in a single check. */ if ((error = zfs_zaccess_rename(sdzp, szp, tdzp, tzp, cr))) goto out; if (S_ISDIR(ZTOI(szp)->i_mode)) { /* * Check to make sure rename is valid. * Can't do a move like this: /usr/a/b to /usr/a/b/c/d */ if ((error = zfs_rename_lock(szp, tdzp, sdzp, &zl))) goto out; } /* * Does target exist? */ if (tzp) { /* * Source and target must be the same type. */ if (S_ISDIR(ZTOI(szp)->i_mode)) { if (!S_ISDIR(ZTOI(tzp)->i_mode)) { error = SET_ERROR(ENOTDIR); goto out; } } else { if (S_ISDIR(ZTOI(tzp)->i_mode)) { error = SET_ERROR(EISDIR); goto out; } } /* * POSIX dictates that when the source and target * entries refer to the same file object, rename * must do nothing and exit without error. */ if (szp->z_id == tzp->z_id) { error = 0; goto out; } } tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE); dmu_tx_hold_sa(tx, sdzp->z_sa_hdl, B_FALSE); dmu_tx_hold_zap(tx, sdzp->z_id, FALSE, snm); dmu_tx_hold_zap(tx, tdzp->z_id, TRUE, tnm); if (sdzp != tdzp) { dmu_tx_hold_sa(tx, tdzp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, tdzp); } if (tzp) { dmu_tx_hold_sa(tx, tzp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, tzp); } zfs_sa_upgrade_txholds(tx, szp); dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL); error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { if (zl != NULL) zfs_rename_unlock(&zl); zfs_dirent_unlock(sdl); zfs_dirent_unlock(tdl); if (sdzp == tdzp) rw_exit(&sdzp->z_name_lock); if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); iput(ZTOI(szp)); if (tzp) iput(ZTOI(tzp)); goto top; } dmu_tx_abort(tx); iput(ZTOI(szp)); if (tzp) iput(ZTOI(tzp)); ZFS_EXIT(zfsvfs); return (error); } if (tzp) /* Attempt to remove the existing target */ error = zfs_link_destroy(tdl, tzp, tx, zflg, NULL); if (error == 0) { error = zfs_link_create(tdl, szp, tx, ZRENAMING); if (error == 0) { szp->z_pflags |= ZFS_AV_MODIFIED; error = sa_update(szp->z_sa_hdl, SA_ZPL_FLAGS(zfsvfs), (void *)&szp->z_pflags, sizeof (uint64_t), tx); ASSERT0(error); error = zfs_link_destroy(sdl, szp, tx, ZRENAMING, NULL); if (error == 0) { zfs_log_rename(zilog, tx, TX_RENAME | (flags & FIGNORECASE ? TX_CI : 0), sdzp, sdl->dl_name, tdzp, tdl->dl_name, szp); } else { /* * At this point, we have successfully created * the target name, but have failed to remove * the source name. Since the create was done * with the ZRENAMING flag, there are * complications; for one, the link count is * wrong. The easiest way to deal with this * is to remove the newly created target, and * return the original error. This must * succeed; fortunately, it is very unlikely to * fail, since we just created it. */ VERIFY3U(zfs_link_destroy(tdl, szp, tx, ZRENAMING, NULL), ==, 0); } } } dmu_tx_commit(tx); out: if (zl != NULL) zfs_rename_unlock(&zl); zfs_dirent_unlock(sdl); zfs_dirent_unlock(tdl); zfs_inode_update(sdzp); if (sdzp == tdzp) rw_exit(&sdzp->z_name_lock); if (sdzp != tdzp) zfs_inode_update(tdzp); zfs_inode_update(szp); iput(ZTOI(szp)); if (tzp) { zfs_inode_update(tzp); iput(ZTOI(tzp)); } if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); ZFS_EXIT(zfsvfs); return (error); } /* * Insert the indicated symbolic reference entry into the directory. * * IN: dip - Directory to contain new symbolic link. * link - Name for new symlink entry. * vap - Attributes of new entry. * target - Target path of new symlink. * * cr - credentials of caller. * flags - case flags * * RETURN: 0 on success, error code on failure. * * Timestamps: * dip - ctime|mtime updated */ /*ARGSUSED*/ int zfs_symlink(struct inode *dip, char *name, vattr_t *vap, char *link, struct inode **ipp, cred_t *cr, int flags) { znode_t *zp, *dzp = ITOZ(dip); zfs_dirlock_t *dl; dmu_tx_t *tx; zfsvfs_t *zfsvfs = ITOZSB(dip); zilog_t *zilog; uint64_t len = strlen(link); int error; int zflg = ZNEW; zfs_acl_ids_t acl_ids; boolean_t fuid_dirtied; uint64_t txtype = TX_SYMLINK; boolean_t waited = B_FALSE; ASSERT(S_ISLNK(vap->va_mode)); if (name == NULL) return (SET_ERROR(EINVAL)); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(dzp); zilog = zfsvfs->z_log; if (zfsvfs->z_utf8 && u8_validate(name, strlen(name), NULL, U8_VALIDATE_ENTIRE, &error) < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EILSEQ)); } if (flags & FIGNORECASE) zflg |= ZCILOOK; if (len > MAXPATHLEN) { ZFS_EXIT(zfsvfs); return (SET_ERROR(ENAMETOOLONG)); } if ((error = zfs_acl_ids_create(dzp, 0, vap, cr, NULL, &acl_ids)) != 0) { ZFS_EXIT(zfsvfs); return (error); } top: *ipp = NULL; /* * Attempt to lock directory; fail if entry already exists. */ error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg, NULL, NULL); if (error) { zfs_acl_ids_free(&acl_ids); ZFS_EXIT(zfsvfs); return (error); } if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) { zfs_acl_ids_free(&acl_ids); zfs_dirent_unlock(dl); ZFS_EXIT(zfsvfs); return (error); } if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) { zfs_acl_ids_free(&acl_ids); zfs_dirent_unlock(dl); ZFS_EXIT(zfsvfs); return (SET_ERROR(EDQUOT)); } tx = dmu_tx_create(zfsvfs->z_os); fuid_dirtied = zfsvfs->z_fuid_dirty; dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, MAX(1, len)); dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name); dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes + ZFS_SA_BASE_ATTR_SIZE + len); dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE); if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) { dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, acl_ids.z_aclp->z_acl_bytes); } if (fuid_dirtied) zfs_fuid_txhold(zfsvfs, tx); error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { zfs_dirent_unlock(dl); if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); goto top; } zfs_acl_ids_free(&acl_ids); dmu_tx_abort(tx); ZFS_EXIT(zfsvfs); return (error); } /* * Create a new object for the symlink. * for version 4 ZPL datsets the symlink will be an SA attribute */ zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids); if (fuid_dirtied) zfs_fuid_sync(zfsvfs, tx); mutex_enter(&zp->z_lock); if (zp->z_is_sa) error = sa_update(zp->z_sa_hdl, SA_ZPL_SYMLINK(zfsvfs), link, len, tx); else zfs_sa_symlink(zp, link, len, tx); mutex_exit(&zp->z_lock); zp->z_size = len; (void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs), &zp->z_size, sizeof (zp->z_size), tx); /* * Insert the new object into the directory. */ (void) zfs_link_create(dl, zp, tx, ZNEW); if (flags & FIGNORECASE) txtype |= TX_CI; zfs_log_symlink(zilog, tx, txtype, dzp, zp, name, link); zfs_inode_update(dzp); zfs_inode_update(zp); zfs_acl_ids_free(&acl_ids); dmu_tx_commit(tx); zfs_dirent_unlock(dl); *ipp = ZTOI(zp); if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); ZFS_EXIT(zfsvfs); return (error); } /* * Return, in the buffer contained in the provided uio structure, * the symbolic path referred to by ip. * * IN: ip - inode of symbolic link * uio - structure to contain the link path. * cr - credentials of caller. * * RETURN: 0 if success * error code if failure * * Timestamps: * ip - atime updated */ /* ARGSUSED */ int zfs_readlink(struct inode *ip, uio_t *uio, cred_t *cr) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); int error; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); mutex_enter(&zp->z_lock); if (zp->z_is_sa) error = sa_lookup_uio(zp->z_sa_hdl, SA_ZPL_SYMLINK(zfsvfs), uio); else error = zfs_sa_readlink(zp, uio); mutex_exit(&zp->z_lock); ZFS_EXIT(zfsvfs); return (error); } /* * Insert a new entry into directory tdip referencing sip. * * IN: tdip - Directory to contain new entry. * sip - inode of new entry. * name - name of new entry. * cr - credentials of caller. * * RETURN: 0 if success * error code if failure * * Timestamps: * tdip - ctime|mtime updated * sip - ctime updated */ /* ARGSUSED */ int zfs_link(struct inode *tdip, struct inode *sip, char *name, cred_t *cr, int flags) { znode_t *dzp = ITOZ(tdip); znode_t *tzp, *szp; zfsvfs_t *zfsvfs = ITOZSB(tdip); zilog_t *zilog; zfs_dirlock_t *dl; dmu_tx_t *tx; int error; int zf = ZNEW; uint64_t parent; uid_t owner; boolean_t waited = B_FALSE; boolean_t is_tmpfile = 0; uint64_t txg; #ifdef HAVE_TMPFILE is_tmpfile = (sip->i_nlink == 0 && (sip->i_state & I_LINKABLE)); #endif ASSERT(S_ISDIR(tdip->i_mode)); if (name == NULL) return (SET_ERROR(EINVAL)); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(dzp); zilog = zfsvfs->z_log; /* * POSIX dictates that we return EPERM here. * Better choices include ENOTSUP or EISDIR. */ if (S_ISDIR(sip->i_mode)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } szp = ITOZ(sip); ZFS_VERIFY_ZP(szp); /* * We check i_sb because snapshots and the ctldir must have different * super blocks. */ if (sip->i_sb != tdip->i_sb || zfsctl_is_node(sip)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EXDEV)); } /* Prevent links to .zfs/shares files */ if ((error = sa_lookup(szp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs), &parent, sizeof (uint64_t))) != 0) { ZFS_EXIT(zfsvfs); return (error); } if (parent == zfsvfs->z_shares_dir) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } if (zfsvfs->z_utf8 && u8_validate(name, strlen(name), NULL, U8_VALIDATE_ENTIRE, &error) < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EILSEQ)); } if (flags & FIGNORECASE) zf |= ZCILOOK; /* * We do not support links between attributes and non-attributes * because of the potential security risk of creating links * into "normal" file space in order to circumvent restrictions * imposed in attribute space. */ if ((szp->z_pflags & ZFS_XATTR) != (dzp->z_pflags & ZFS_XATTR)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } owner = zfs_fuid_map_id(zfsvfs, KUID_TO_SUID(sip->i_uid), cr, ZFS_OWNER); if (owner != crgetuid(cr) && secpolicy_basic_link(cr) != 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) { ZFS_EXIT(zfsvfs); return (error); } top: /* * Attempt to lock directory; fail if entry already exists. */ error = zfs_dirent_lock(&dl, dzp, name, &tzp, zf, NULL, NULL); if (error) { ZFS_EXIT(zfsvfs); return (error); } tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE); dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name); if (is_tmpfile) dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL); zfs_sa_upgrade_txholds(tx, szp); zfs_sa_upgrade_txholds(tx, dzp); error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { zfs_dirent_unlock(dl); if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); goto top; } dmu_tx_abort(tx); ZFS_EXIT(zfsvfs); return (error); } /* unmark z_unlinked so zfs_link_create will not reject */ if (is_tmpfile) szp->z_unlinked = 0; error = zfs_link_create(dl, szp, tx, 0); if (error == 0) { uint64_t txtype = TX_LINK; /* * tmpfile is created to be in z_unlinkedobj, so remove it. * Also, we don't log in ZIL, be cause all previous file * operation on the tmpfile are ignored by ZIL. Instead we * always wait for txg to sync to make sure all previous * operation are sync safe. */ if (is_tmpfile) { VERIFY(zap_remove_int(zfsvfs->z_os, zfsvfs->z_unlinkedobj, szp->z_id, tx) == 0); } else { if (flags & FIGNORECASE) txtype |= TX_CI; zfs_log_link(zilog, tx, txtype, dzp, szp, name); } } else if (is_tmpfile) { /* restore z_unlinked since when linking failed */ szp->z_unlinked = 1; } txg = dmu_tx_get_txg(tx); dmu_tx_commit(tx); zfs_dirent_unlock(dl); if (!is_tmpfile && zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); if (is_tmpfile) txg_wait_synced(dmu_objset_pool(zfsvfs->z_os), txg); zfs_inode_update(dzp); zfs_inode_update(szp); ZFS_EXIT(zfsvfs); return (error); } static void zfs_putpage_commit_cb(void *arg) { struct page *pp = arg; ClearPageError(pp); end_page_writeback(pp); } /* * Push a page out to disk, once the page is on stable storage the * registered commit callback will be run as notification of completion. * * IN: ip - page mapped for inode. * pp - page to push (page is locked) * wbc - writeback control data * * RETURN: 0 if success * error code if failure * * Timestamps: * ip - ctime|mtime updated */ /* ARGSUSED */ int zfs_putpage(struct inode *ip, struct page *pp, struct writeback_control *wbc) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); loff_t offset; loff_t pgoff; unsigned int pglen; rl_t *rl; dmu_tx_t *tx; caddr_t va; int err = 0; uint64_t mtime[2], ctime[2]; sa_bulk_attr_t bulk[3]; int cnt = 0; struct address_space *mapping; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); ASSERT(PageLocked(pp)); pgoff = page_offset(pp); /* Page byte-offset in file */ offset = i_size_read(ip); /* File length in bytes */ pglen = MIN(PAGE_SIZE, /* Page length in bytes */ P2ROUNDUP(offset, PAGE_SIZE)-pgoff); /* Page is beyond end of file */ if (pgoff >= offset) { unlock_page(pp); ZFS_EXIT(zfsvfs); return (0); } /* Truncate page length to end of file */ if (pgoff + pglen > offset) pglen = offset - pgoff; #if 0 /* * FIXME: Allow mmap writes past its quota. The correct fix * is to register a page_mkwrite() handler to count the page * against its quota when it is about to be dirtied. */ if (zfs_owner_overquota(zfsvfs, zp, B_FALSE) || zfs_owner_overquota(zfsvfs, zp, B_TRUE)) { err = EDQUOT; } #endif /* * The ordering here is critical and must adhere to the following * rules in order to avoid deadlocking in either zfs_read() or * zfs_free_range() due to a lock inversion. * * 1) The page must be unlocked prior to acquiring the range lock. * This is critical because zfs_read() calls find_lock_page() * which may block on the page lock while holding the range lock. * * 2) Before setting or clearing write back on a page the range lock * must be held in order to prevent a lock inversion with the * zfs_free_range() function. * * This presents a problem because upon entering this function the * page lock is already held. To safely acquire the range lock the * page lock must be dropped. This creates a window where another * process could truncate, invalidate, dirty, or write out the page. * * Therefore, after successfully reacquiring the range and page locks * the current page state is checked. In the common case everything * will be as is expected and it can be written out. However, if * the page state has changed it must be handled accordingly. */ mapping = pp->mapping; redirty_page_for_writepage(wbc, pp); unlock_page(pp); rl = zfs_range_lock(&zp->z_range_lock, pgoff, pglen, RL_WRITER); lock_page(pp); /* Page mapping changed or it was no longer dirty, we're done */ if (unlikely((mapping != pp->mapping) || !PageDirty(pp))) { unlock_page(pp); zfs_range_unlock(rl); ZFS_EXIT(zfsvfs); return (0); } /* Another process started write block if required */ if (PageWriteback(pp)) { unlock_page(pp); zfs_range_unlock(rl); if (wbc->sync_mode != WB_SYNC_NONE) wait_on_page_writeback(pp); ZFS_EXIT(zfsvfs); return (0); } /* Clear the dirty flag the required locks are held */ if (!clear_page_dirty_for_io(pp)) { unlock_page(pp); zfs_range_unlock(rl); ZFS_EXIT(zfsvfs); return (0); } /* * Counterpart for redirty_page_for_writepage() above. This page * was in fact not skipped and should not be counted as if it were. */ wbc->pages_skipped--; set_page_writeback(pp); unlock_page(pp); tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_write(tx, zp->z_id, pgoff, pglen); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, zp); err = dmu_tx_assign(tx, TXG_NOWAIT); if (err != 0) { if (err == ERESTART) dmu_tx_wait(tx); dmu_tx_abort(tx); __set_page_dirty_nobuffers(pp); ClearPageError(pp); end_page_writeback(pp); zfs_range_unlock(rl); ZFS_EXIT(zfsvfs); return (err); } va = kmap(pp); ASSERT3U(pglen, <=, PAGE_SIZE); dmu_write(zfsvfs->z_os, zp->z_id, pgoff, pglen, va, tx); kunmap(pp); SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16); SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16); SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_FLAGS(zfsvfs), NULL, &zp->z_pflags, 8); /* Preserve the mtime and ctime provided by the inode */ ZFS_TIME_ENCODE(&ip->i_mtime, mtime); ZFS_TIME_ENCODE(&ip->i_ctime, ctime); zp->z_atime_dirty = 0; zp->z_seq++; err = sa_bulk_update(zp->z_sa_hdl, bulk, cnt, tx); zfs_log_write(zfsvfs->z_log, tx, TX_WRITE, zp, pgoff, pglen, 0, zfs_putpage_commit_cb, pp); dmu_tx_commit(tx); zfs_range_unlock(rl); if (wbc->sync_mode != WB_SYNC_NONE) { /* * Note that this is rarely called under writepages(), because * writepages() normally handles the entire commit for * performance reasons. */ zil_commit(zfsvfs->z_log, zp->z_id); } ZFS_EXIT(zfsvfs); return (err); } /* * Update the system attributes when the inode has been dirtied. For the * moment we only update the mode, atime, mtime, and ctime. */ int zfs_dirty_inode(struct inode *ip, int flags) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); dmu_tx_t *tx; uint64_t mode, atime[2], mtime[2], ctime[2]; sa_bulk_attr_t bulk[4]; int error = 0; int cnt = 0; if (zfs_is_readonly(zfsvfs) || dmu_objset_is_snapshot(zfsvfs->z_os)) return (0); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); #ifdef I_DIRTY_TIME /* * This is the lazytime semantic indroduced in Linux 4.0 * This flag will only be called from update_time when lazytime is set. * (Note, I_DIRTY_SYNC will also set if not lazytime) * Fortunately mtime and ctime are managed within ZFS itself, so we * only need to dirty atime. */ if (flags == I_DIRTY_TIME) { zp->z_atime_dirty = 1; goto out; } #endif tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, zp); error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); goto out; } mutex_enter(&zp->z_lock); zp->z_atime_dirty = 0; SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MODE(zfsvfs), NULL, &mode, 8); SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_ATIME(zfsvfs), NULL, &atime, 16); SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16); SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16); /* Preserve the mode, mtime and ctime provided by the inode */ ZFS_TIME_ENCODE(&ip->i_atime, atime); ZFS_TIME_ENCODE(&ip->i_mtime, mtime); ZFS_TIME_ENCODE(&ip->i_ctime, ctime); mode = ip->i_mode; zp->z_mode = mode; error = sa_bulk_update(zp->z_sa_hdl, bulk, cnt, tx); mutex_exit(&zp->z_lock); dmu_tx_commit(tx); out: ZFS_EXIT(zfsvfs); return (error); } /*ARGSUSED*/ void zfs_inactive(struct inode *ip) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); uint64_t atime[2]; int error; int need_unlock = 0; /* Only read lock if we haven't already write locked, e.g. rollback */ if (!RW_WRITE_HELD(&zfsvfs->z_teardown_inactive_lock)) { need_unlock = 1; rw_enter(&zfsvfs->z_teardown_inactive_lock, RW_READER); } if (zp->z_sa_hdl == NULL) { if (need_unlock) rw_exit(&zfsvfs->z_teardown_inactive_lock); return; } if (zp->z_atime_dirty && zp->z_unlinked == 0) { dmu_tx_t *tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, zp); error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); } else { ZFS_TIME_ENCODE(&ip->i_atime, atime); mutex_enter(&zp->z_lock); (void) sa_update(zp->z_sa_hdl, SA_ZPL_ATIME(zfsvfs), (void *)&atime, sizeof (atime), tx); zp->z_atime_dirty = 0; mutex_exit(&zp->z_lock); dmu_tx_commit(tx); } } zfs_zinactive(zp); if (need_unlock) rw_exit(&zfsvfs->z_teardown_inactive_lock); } /* * Bounds-check the seek operation. * * IN: ip - inode seeking within * ooff - old file offset * noffp - pointer to new file offset * ct - caller context * * RETURN: 0 if success * EINVAL if new offset invalid */ /* ARGSUSED */ int zfs_seek(struct inode *ip, offset_t ooff, offset_t *noffp) { if (S_ISDIR(ip->i_mode)) return (0); return ((*noffp < 0 || *noffp > MAXOFFSET_T) ? EINVAL : 0); } /* * Fill pages with data from the disk. */ static int zfs_fillpage(struct inode *ip, struct page *pl[], int nr_pages) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); objset_t *os; struct page *cur_pp; u_offset_t io_off, total; size_t io_len; loff_t i_size; unsigned page_idx; int err; os = zfsvfs->z_os; io_len = nr_pages << PAGE_SHIFT; i_size = i_size_read(ip); io_off = page_offset(pl[0]); if (io_off + io_len > i_size) io_len = i_size - io_off; /* * Iterate over list of pages and read each page individually. */ page_idx = 0; for (total = io_off + io_len; io_off < total; io_off += PAGESIZE) { caddr_t va; cur_pp = pl[page_idx++]; va = kmap(cur_pp); err = dmu_read(os, zp->z_id, io_off, PAGESIZE, va, DMU_READ_PREFETCH); kunmap(cur_pp); if (err) { /* convert checksum errors into IO errors */ if (err == ECKSUM) err = SET_ERROR(EIO); return (err); } } return (0); } /* * Uses zfs_fillpage to read data from the file and fill the pages. * * IN: ip - inode of file to get data from. * pl - list of pages to read * nr_pages - number of pages to read * * RETURN: 0 on success, error code on failure. * * Timestamps: * vp - atime updated */ /* ARGSUSED */ int zfs_getpage(struct inode *ip, struct page *pl[], int nr_pages) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); int err; if (pl == NULL) return (0); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); err = zfs_fillpage(ip, pl, nr_pages); ZFS_EXIT(zfsvfs); return (err); } /* * Check ZFS specific permissions to memory map a section of a file. * * IN: ip - inode of the file to mmap * off - file offset * addrp - start address in memory region * len - length of memory region * vm_flags- address flags * * RETURN: 0 if success * error code if failure */ /*ARGSUSED*/ int zfs_map(struct inode *ip, offset_t off, caddr_t *addrp, size_t len, unsigned long vm_flags) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); if ((vm_flags & VM_WRITE) && (zp->z_pflags & (ZFS_IMMUTABLE | ZFS_READONLY | ZFS_APPENDONLY))) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } if ((vm_flags & (VM_READ | VM_EXEC)) && (zp->z_pflags & ZFS_AV_QUARANTINED)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EACCES)); } if (off < 0 || len > MAXOFFSET_T - off) { ZFS_EXIT(zfsvfs); return (SET_ERROR(ENXIO)); } ZFS_EXIT(zfsvfs); return (0); } /* * convoff - converts the given data (start, whence) to the * given whence. */ int convoff(struct inode *ip, flock64_t *lckdat, int whence, offset_t offset) { vattr_t vap; int error; if ((lckdat->l_whence == 2) || (whence == 2)) { if ((error = zfs_getattr(ip, &vap, 0, CRED()))) return (error); } switch (lckdat->l_whence) { case 1: lckdat->l_start += offset; break; case 2: lckdat->l_start += vap.va_size; /* FALLTHRU */ case 0: break; default: return (SET_ERROR(EINVAL)); } if (lckdat->l_start < 0) return (SET_ERROR(EINVAL)); switch (whence) { case 1: lckdat->l_start -= offset; break; case 2: lckdat->l_start -= vap.va_size; /* FALLTHRU */ case 0: break; default: return (SET_ERROR(EINVAL)); } lckdat->l_whence = (short)whence; return (0); } /* * Free or allocate space in a file. Currently, this function only * supports the `F_FREESP' command. However, this command is somewhat * misnamed, as its functionality includes the ability to allocate as * well as free space. * * IN: ip - inode of file to free data in. * cmd - action to take (only F_FREESP supported). * bfp - section of file to free/alloc. * flag - current file open mode flags. * offset - current file offset. * cr - credentials of caller [UNUSED]. * * RETURN: 0 on success, error code on failure. * * Timestamps: * ip - ctime|mtime updated */ /* ARGSUSED */ int zfs_space(struct inode *ip, int cmd, flock64_t *bfp, int flag, offset_t offset, cred_t *cr) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); uint64_t off, len; int error; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); if (cmd != F_FREESP) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } /* * 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); return (SET_ERROR(EROFS)); } if ((error = convoff(ip, bfp, 0, offset))) { ZFS_EXIT(zfsvfs); return (error); } if (bfp->l_len < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } /* * Permissions aren't checked on Solaris because on this OS * zfs_space() can only be called with an opened file handle. * On Linux we can get here through truncate_range() which * operates directly on inodes, so we need to check access rights. */ if ((error = zfs_zaccess(zp, ACE_WRITE_DATA, 0, B_FALSE, cr))) { ZFS_EXIT(zfsvfs); return (error); } off = bfp->l_start; len = bfp->l_len; /* 0 means from off to end of file */ error = zfs_freesp(zp, off, len, flag, TRUE); ZFS_EXIT(zfsvfs); return (error); } /*ARGSUSED*/ int zfs_fid(struct inode *ip, fid_t *fidp) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); uint32_t gen; uint64_t gen64; uint64_t object = zp->z_id; zfid_short_t *zfid; int size, i, error; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zfsvfs), &gen64, sizeof (uint64_t))) != 0) { ZFS_EXIT(zfsvfs); return (error); } gen = (uint32_t)gen64; size = SHORT_FID_LEN; zfid = (zfid_short_t *)fidp; zfid->zf_len = size; for (i = 0; i < sizeof (zfid->zf_object); i++) zfid->zf_object[i] = (uint8_t)(object >> (8 * i)); /* Must have a non-zero generation number to distinguish from .zfs */ if (gen == 0) gen = 1; for (i = 0; i < sizeof (zfid->zf_gen); i++) zfid->zf_gen[i] = (uint8_t)(gen >> (8 * i)); ZFS_EXIT(zfsvfs); return (0); } /*ARGSUSED*/ int zfs_getsecattr(struct inode *ip, vsecattr_t *vsecp, int flag, cred_t *cr) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); int error; boolean_t skipaclchk = (flag & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); error = zfs_getacl(zp, vsecp, skipaclchk, cr); ZFS_EXIT(zfsvfs); return (error); } /*ARGSUSED*/ int zfs_setsecattr(struct inode *ip, vsecattr_t *vsecp, int flag, cred_t *cr) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); int error; boolean_t skipaclchk = (flag & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE; zilog_t *zilog = zfsvfs->z_log; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); error = zfs_setacl(zp, vsecp, skipaclchk, cr); if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); ZFS_EXIT(zfsvfs); return (error); } #ifdef HAVE_UIO_ZEROCOPY /* * Tunable, both must be a power of 2. * * zcr_blksz_min: the smallest read we may consider to loan out an arcbuf * zcr_blksz_max: if set to less than the file block size, allow loaning out of * an arcbuf for a partial block read */ int zcr_blksz_min = (1 << 10); /* 1K */ int zcr_blksz_max = (1 << 17); /* 128K */ /*ARGSUSED*/ static int zfs_reqzcbuf(struct inode *ip, enum uio_rw ioflag, xuio_t *xuio, cred_t *cr) { znode_t *zp = ITOZ(ip); zfsvfs_t *zfsvfs = ITOZSB(ip); int max_blksz = zfsvfs->z_max_blksz; uio_t *uio = &xuio->xu_uio; ssize_t size = uio->uio_resid; offset_t offset = uio->uio_loffset; int blksz; int fullblk, i; arc_buf_t *abuf; ssize_t maxsize; int preamble, postamble; if (xuio->xu_type != UIOTYPE_ZEROCOPY) return (SET_ERROR(EINVAL)); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); switch (ioflag) { case UIO_WRITE: /* * Loan out an arc_buf for write if write size is bigger than * max_blksz, and the file's block size is also max_blksz. */ blksz = max_blksz; if (size < blksz || zp->z_blksz != blksz) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } /* * Caller requests buffers for write before knowing where the * write offset might be (e.g. NFS TCP write). */ if (offset == -1) { preamble = 0; } else { preamble = P2PHASE(offset, blksz); if (preamble) { preamble = blksz - preamble; size -= preamble; } } postamble = P2PHASE(size, blksz); size -= postamble; fullblk = size / blksz; (void) dmu_xuio_init(xuio, (preamble != 0) + fullblk + (postamble != 0)); /* * Have to fix iov base/len for partial buffers. They * currently represent full arc_buf's. */ if (preamble) { /* data begins in the middle of the arc_buf */ abuf = dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl), blksz); ASSERT(abuf); (void) dmu_xuio_add(xuio, abuf, blksz - preamble, preamble); } for (i = 0; i < fullblk; i++) { abuf = dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl), blksz); ASSERT(abuf); (void) dmu_xuio_add(xuio, abuf, 0, blksz); } if (postamble) { /* data ends in the middle of the arc_buf */ abuf = dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl), blksz); ASSERT(abuf); (void) dmu_xuio_add(xuio, abuf, 0, postamble); } break; case UIO_READ: /* * Loan out an arc_buf for read if the read size is larger than * the current file block size. Block alignment is not * considered. Partial arc_buf will be loaned out for read. */ blksz = zp->z_blksz; if (blksz < zcr_blksz_min) blksz = zcr_blksz_min; if (blksz > zcr_blksz_max) blksz = zcr_blksz_max; /* avoid potential complexity of dealing with it */ if (blksz > max_blksz) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } maxsize = zp->z_size - uio->uio_loffset; if (size > maxsize) size = maxsize; if (size < blksz) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } break; default: ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } uio->uio_extflg = UIO_XUIO; XUIO_XUZC_RW(xuio) = ioflag; ZFS_EXIT(zfsvfs); return (0); } /*ARGSUSED*/ static int zfs_retzcbuf(struct inode *ip, xuio_t *xuio, cred_t *cr) { int i; arc_buf_t *abuf; int ioflag = XUIO_XUZC_RW(xuio); ASSERT(xuio->xu_type == UIOTYPE_ZEROCOPY); i = dmu_xuio_cnt(xuio); while (i-- > 0) { abuf = dmu_xuio_arcbuf(xuio, i); /* * if abuf == NULL, it must be a write buffer * that has been returned in zfs_write(). */ if (abuf) dmu_return_arcbuf(abuf); ASSERT(abuf || ioflag == UIO_WRITE); } dmu_xuio_fini(xuio); return (0); } #endif /* HAVE_UIO_ZEROCOPY */ #if defined(_KERNEL) && defined(HAVE_SPL) EXPORT_SYMBOL(zfs_open); EXPORT_SYMBOL(zfs_close); EXPORT_SYMBOL(zfs_read); EXPORT_SYMBOL(zfs_write); EXPORT_SYMBOL(zfs_access); EXPORT_SYMBOL(zfs_lookup); EXPORT_SYMBOL(zfs_create); EXPORT_SYMBOL(zfs_tmpfile); EXPORT_SYMBOL(zfs_remove); EXPORT_SYMBOL(zfs_mkdir); EXPORT_SYMBOL(zfs_rmdir); EXPORT_SYMBOL(zfs_readdir); EXPORT_SYMBOL(zfs_fsync); EXPORT_SYMBOL(zfs_getattr); EXPORT_SYMBOL(zfs_getattr_fast); EXPORT_SYMBOL(zfs_setattr); EXPORT_SYMBOL(zfs_rename); EXPORT_SYMBOL(zfs_symlink); EXPORT_SYMBOL(zfs_readlink); EXPORT_SYMBOL(zfs_link); EXPORT_SYMBOL(zfs_inactive); EXPORT_SYMBOL(zfs_space); EXPORT_SYMBOL(zfs_fid); EXPORT_SYMBOL(zfs_getsecattr); EXPORT_SYMBOL(zfs_setsecattr); EXPORT_SYMBOL(zfs_getpage); EXPORT_SYMBOL(zfs_putpage); EXPORT_SYMBOL(zfs_dirty_inode); EXPORT_SYMBOL(zfs_map); /* CSTYLED */ module_param(zfs_delete_blocks, ulong, 0644); MODULE_PARM_DESC(zfs_delete_blocks, "Delete files larger than N blocks async"); module_param(zfs_read_chunk_size, long, 0644); MODULE_PARM_DESC(zfs_read_chunk_size, "Bytes to read per chunk"); #endif diff --git a/module/zfs/zil.c b/module/zfs/zil.c index 9f4312a5a28d..5300b3408a14 100644 --- a/module/zfs/zil.c +++ b/module/zfs/zil.c @@ -1,2394 +1,3412 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2017 by Delphix. All rights reserved. * Copyright (c) 2014 Integros [integros.com] */ /* Portions Copyright 2010 Robert Milkowski */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* - * The zfs intent log (ZIL) saves transaction records of system calls - * that change the file system in memory with enough information - * to be able to replay them. These are stored in memory until - * either the DMU transaction group (txg) commits them to the stable pool - * and they can be discarded, or they are flushed to the stable log - * (also in the pool) due to a fsync, O_DSYNC or other synchronous - * requirement. In the event of a panic or power fail then those log - * records (transactions) are replayed. + * The ZFS Intent Log (ZIL) saves "transaction records" (itxs) of system + * calls that change the file system. Each itx has enough information to + * be able to replay them after a system crash, power loss, or + * equivalent failure mode. These are stored in memory until either: * - * There is one ZIL per file system. Its on-disk (pool) format consists - * of 3 parts: + * 1. they are committed to the pool by the DMU transaction group + * (txg), at which point they can be discarded; or + * 2. they are committed to the on-disk ZIL for the dataset being + * modified (e.g. due to an fsync, O_DSYNC, or other synchronous + * requirement). * - * - ZIL header - * - ZIL blocks - * - ZIL records + * In the event of a crash or power loss, the itxs contained by each + * dataset's on-disk ZIL will be replayed when that dataset is first + * instantianted (e.g. if the dataset is a normal fileystem, when it is + * first mounted). * - * A log record holds a system call transaction. Log blocks can - * hold many log records and the blocks are chained together. - * Each ZIL block contains a block pointer (blkptr_t) to the next - * ZIL block in the chain. The ZIL header points to the first - * block in the chain. Note there is not a fixed place in the pool - * to hold blocks. They are dynamically allocated and freed as - * needed from the blocks available. Figure X shows the ZIL structure: + * As hinted at above, there is one ZIL per dataset (both the in-memory + * representation, and the on-disk representation). The on-disk format + * consists of 3 parts: + * + * - a single, per-dataset, ZIL header; which points to a chain of + * - zero or more ZIL blocks; each of which contains + * - zero or more ZIL records + * + * A ZIL record holds the information necessary to replay a single + * system call transaction. A ZIL block can hold many ZIL records, and + * the blocks are chained together, similarly to a singly linked list. + * + * Each ZIL block contains a block pointer (blkptr_t) to the next ZIL + * block in the chain, and the ZIL header points to the first block in + * the chain. + * + * Note, there is not a fixed place in the pool to hold these ZIL + * blocks; they are dynamically allocated and freed as needed from the + * blocks available on the pool, though they can be preferentially + * allocated from a dedicated "log" vdev. */ +/* + * This controls the amount of time that a ZIL block (lwb) will remain + * "open" when it isn't "full", and it has a thread waiting for it to be + * committed to stable storage. Please refer to the zil_commit_waiter() + * function (and the comments within it) for more details. + */ +int zfs_commit_timeout_pct = 5; + /* * See zil.h for more information about these fields. */ zil_stats_t zil_stats = { { "zil_commit_count", KSTAT_DATA_UINT64 }, { "zil_commit_writer_count", KSTAT_DATA_UINT64 }, { "zil_itx_count", KSTAT_DATA_UINT64 }, { "zil_itx_indirect_count", KSTAT_DATA_UINT64 }, { "zil_itx_indirect_bytes", KSTAT_DATA_UINT64 }, { "zil_itx_copied_count", KSTAT_DATA_UINT64 }, { "zil_itx_copied_bytes", KSTAT_DATA_UINT64 }, { "zil_itx_needcopy_count", KSTAT_DATA_UINT64 }, { "zil_itx_needcopy_bytes", KSTAT_DATA_UINT64 }, { "zil_itx_metaslab_normal_count", KSTAT_DATA_UINT64 }, { "zil_itx_metaslab_normal_bytes", KSTAT_DATA_UINT64 }, { "zil_itx_metaslab_slog_count", KSTAT_DATA_UINT64 }, { "zil_itx_metaslab_slog_bytes", KSTAT_DATA_UINT64 }, }; static kstat_t *zil_ksp; /* * Disable intent logging replay. This global ZIL switch affects all pools. */ int zil_replay_disable = 0; /* * Tunable parameter for debugging or performance analysis. Setting * zfs_nocacheflush will cause corruption on power loss if a volatile * out-of-order write cache is enabled. */ int zfs_nocacheflush = 0; /* * Limit SLOG write size per commit executed with synchronous priority. * Any writes above that will be executed with lower (asynchronous) priority * to limit potential SLOG device abuse by single active ZIL writer. */ unsigned long zil_slog_bulk = 768 * 1024; static kmem_cache_t *zil_lwb_cache; +static kmem_cache_t *zil_zcw_cache; static void zil_async_to_sync(zilog_t *zilog, uint64_t foid); #define LWB_EMPTY(lwb) ((BP_GET_LSIZE(&lwb->lwb_blk) - \ sizeof (zil_chain_t)) == (lwb->lwb_sz - lwb->lwb_nused)) static int zil_bp_compare(const void *x1, const void *x2) { const dva_t *dva1 = &((zil_bp_node_t *)x1)->zn_dva; const dva_t *dva2 = &((zil_bp_node_t *)x2)->zn_dva; int cmp = AVL_CMP(DVA_GET_VDEV(dva1), DVA_GET_VDEV(dva2)); if (likely(cmp)) return (cmp); return (AVL_CMP(DVA_GET_OFFSET(dva1), DVA_GET_OFFSET(dva2))); } static void zil_bp_tree_init(zilog_t *zilog) { avl_create(&zilog->zl_bp_tree, zil_bp_compare, sizeof (zil_bp_node_t), offsetof(zil_bp_node_t, zn_node)); } static void zil_bp_tree_fini(zilog_t *zilog) { avl_tree_t *t = &zilog->zl_bp_tree; zil_bp_node_t *zn; void *cookie = NULL; while ((zn = avl_destroy_nodes(t, &cookie)) != NULL) kmem_free(zn, sizeof (zil_bp_node_t)); avl_destroy(t); } int zil_bp_tree_add(zilog_t *zilog, const blkptr_t *bp) { avl_tree_t *t = &zilog->zl_bp_tree; const dva_t *dva; zil_bp_node_t *zn; avl_index_t where; if (BP_IS_EMBEDDED(bp)) return (0); dva = BP_IDENTITY(bp); if (avl_find(t, dva, &where) != NULL) return (SET_ERROR(EEXIST)); zn = kmem_alloc(sizeof (zil_bp_node_t), KM_SLEEP); zn->zn_dva = *dva; avl_insert(t, zn, where); return (0); } static zil_header_t * zil_header_in_syncing_context(zilog_t *zilog) { return ((zil_header_t *)zilog->zl_header); } static void zil_init_log_chain(zilog_t *zilog, blkptr_t *bp) { zio_cksum_t *zc = &bp->blk_cksum; zc->zc_word[ZIL_ZC_GUID_0] = spa_get_random(-1ULL); zc->zc_word[ZIL_ZC_GUID_1] = spa_get_random(-1ULL); zc->zc_word[ZIL_ZC_OBJSET] = dmu_objset_id(zilog->zl_os); zc->zc_word[ZIL_ZC_SEQ] = 1ULL; } /* * Read a log block and make sure it's valid. */ static int zil_read_log_block(zilog_t *zilog, boolean_t decrypt, const blkptr_t *bp, blkptr_t *nbp, void *dst, char **end) { enum zio_flag zio_flags = ZIO_FLAG_CANFAIL; arc_flags_t aflags = ARC_FLAG_WAIT; arc_buf_t *abuf = NULL; zbookmark_phys_t zb; int error; if (zilog->zl_header->zh_claim_txg == 0) zio_flags |= ZIO_FLAG_SPECULATIVE | ZIO_FLAG_SCRUB; if (!(zilog->zl_header->zh_flags & ZIL_CLAIM_LR_SEQ_VALID)) zio_flags |= ZIO_FLAG_SPECULATIVE; if (!decrypt) zio_flags |= ZIO_FLAG_RAW; SET_BOOKMARK(&zb, bp->blk_cksum.zc_word[ZIL_ZC_OBJSET], ZB_ZIL_OBJECT, ZB_ZIL_LEVEL, bp->blk_cksum.zc_word[ZIL_ZC_SEQ]); error = arc_read(NULL, zilog->zl_spa, bp, arc_getbuf_func, &abuf, ZIO_PRIORITY_SYNC_READ, zio_flags, &aflags, &zb); if (error == 0) { zio_cksum_t cksum = bp->blk_cksum; /* * Validate the checksummed log block. * * Sequence numbers should be... sequential. The checksum * verifier for the next block should be bp's checksum plus 1. * * Also check the log chain linkage and size used. */ cksum.zc_word[ZIL_ZC_SEQ]++; if (BP_GET_CHECKSUM(bp) == ZIO_CHECKSUM_ZILOG2) { zil_chain_t *zilc = abuf->b_data; char *lr = (char *)(zilc + 1); uint64_t len = zilc->zc_nused - sizeof (zil_chain_t); if (bcmp(&cksum, &zilc->zc_next_blk.blk_cksum, sizeof (cksum)) || BP_IS_HOLE(&zilc->zc_next_blk)) { error = SET_ERROR(ECKSUM); } else { ASSERT3U(len, <=, SPA_OLD_MAXBLOCKSIZE); bcopy(lr, dst, len); *end = (char *)dst + len; *nbp = zilc->zc_next_blk; } } else { char *lr = abuf->b_data; uint64_t size = BP_GET_LSIZE(bp); zil_chain_t *zilc = (zil_chain_t *)(lr + size) - 1; if (bcmp(&cksum, &zilc->zc_next_blk.blk_cksum, sizeof (cksum)) || BP_IS_HOLE(&zilc->zc_next_blk) || (zilc->zc_nused > (size - sizeof (*zilc)))) { error = SET_ERROR(ECKSUM); } else { ASSERT3U(zilc->zc_nused, <=, SPA_OLD_MAXBLOCKSIZE); bcopy(lr, dst, zilc->zc_nused); *end = (char *)dst + zilc->zc_nused; *nbp = zilc->zc_next_blk; } } arc_buf_destroy(abuf, &abuf); } return (error); } /* * Read a TX_WRITE log data block. */ static int zil_read_log_data(zilog_t *zilog, const lr_write_t *lr, void *wbuf) { enum zio_flag zio_flags = ZIO_FLAG_CANFAIL; const blkptr_t *bp = &lr->lr_blkptr; arc_flags_t aflags = ARC_FLAG_WAIT; arc_buf_t *abuf = NULL; zbookmark_phys_t zb; int error; if (BP_IS_HOLE(bp)) { if (wbuf != NULL) bzero(wbuf, MAX(BP_GET_LSIZE(bp), lr->lr_length)); return (0); } if (zilog->zl_header->zh_claim_txg == 0) zio_flags |= ZIO_FLAG_SPECULATIVE | ZIO_FLAG_SCRUB; /* * If we are not using the resulting data, we are just checking that * it hasn't been corrupted so we don't need to waste CPU time * decompressing and decrypting it. */ if (wbuf == NULL) zio_flags |= ZIO_FLAG_RAW; SET_BOOKMARK(&zb, dmu_objset_id(zilog->zl_os), lr->lr_foid, ZB_ZIL_LEVEL, lr->lr_offset / BP_GET_LSIZE(bp)); error = arc_read(NULL, zilog->zl_spa, bp, arc_getbuf_func, &abuf, ZIO_PRIORITY_SYNC_READ, zio_flags, &aflags, &zb); if (error == 0) { if (wbuf != NULL) bcopy(abuf->b_data, wbuf, arc_buf_size(abuf)); arc_buf_destroy(abuf, &abuf); } return (error); } /* * Parse the intent log, and call parse_func for each valid record within. */ int zil_parse(zilog_t *zilog, zil_parse_blk_func_t *parse_blk_func, zil_parse_lr_func_t *parse_lr_func, void *arg, uint64_t txg, boolean_t decrypt) { const zil_header_t *zh = zilog->zl_header; boolean_t claimed = !!zh->zh_claim_txg; uint64_t claim_blk_seq = claimed ? zh->zh_claim_blk_seq : UINT64_MAX; uint64_t claim_lr_seq = claimed ? zh->zh_claim_lr_seq : UINT64_MAX; uint64_t max_blk_seq = 0; uint64_t max_lr_seq = 0; uint64_t blk_count = 0; uint64_t lr_count = 0; blkptr_t blk, next_blk; char *lrbuf, *lrp; int error = 0; bzero(&next_blk, sizeof (blkptr_t)); /* * Old logs didn't record the maximum zh_claim_lr_seq. */ if (!(zh->zh_flags & ZIL_CLAIM_LR_SEQ_VALID)) claim_lr_seq = UINT64_MAX; /* * Starting at the block pointed to by zh_log we read the log chain. * For each block in the chain we strongly check that block to * ensure its validity. We stop when an invalid block is found. * For each block pointer in the chain we call parse_blk_func(). * For each record in each valid block we call parse_lr_func(). * If the log has been claimed, stop if we encounter a sequence * number greater than the highest claimed sequence number. */ lrbuf = zio_buf_alloc(SPA_OLD_MAXBLOCKSIZE); zil_bp_tree_init(zilog); for (blk = zh->zh_log; !BP_IS_HOLE(&blk); blk = next_blk) { uint64_t blk_seq = blk.blk_cksum.zc_word[ZIL_ZC_SEQ]; int reclen; char *end = NULL; if (blk_seq > claim_blk_seq) break; error = parse_blk_func(zilog, &blk, arg, txg); if (error != 0) break; ASSERT3U(max_blk_seq, <, blk_seq); max_blk_seq = blk_seq; blk_count++; if (max_lr_seq == claim_lr_seq && max_blk_seq == claim_blk_seq) break; error = zil_read_log_block(zilog, decrypt, &blk, &next_blk, lrbuf, &end); if (error != 0) break; for (lrp = lrbuf; lrp < end; lrp += reclen) { lr_t *lr = (lr_t *)lrp; reclen = lr->lrc_reclen; ASSERT3U(reclen, >=, sizeof (lr_t)); if (lr->lrc_seq > claim_lr_seq) goto done; error = parse_lr_func(zilog, lr, arg, txg); if (error != 0) goto done; ASSERT3U(max_lr_seq, <, lr->lrc_seq); max_lr_seq = lr->lrc_seq; lr_count++; } } done: zilog->zl_parse_error = error; zilog->zl_parse_blk_seq = max_blk_seq; zilog->zl_parse_lr_seq = max_lr_seq; zilog->zl_parse_blk_count = blk_count; zilog->zl_parse_lr_count = lr_count; ASSERT(!claimed || !(zh->zh_flags & ZIL_CLAIM_LR_SEQ_VALID) || (max_blk_seq == claim_blk_seq && max_lr_seq == claim_lr_seq) || (decrypt && error == EIO)); zil_bp_tree_fini(zilog); zio_buf_free(lrbuf, SPA_OLD_MAXBLOCKSIZE); return (error); } static int zil_claim_log_block(zilog_t *zilog, blkptr_t *bp, void *tx, uint64_t first_txg) { /* * Claim log block if not already committed and not already claimed. * If tx == NULL, just verify that the block is claimable. */ if (BP_IS_HOLE(bp) || bp->blk_birth < first_txg || zil_bp_tree_add(zilog, bp) != 0) return (0); return (zio_wait(zio_claim(NULL, zilog->zl_spa, tx == NULL ? 0 : first_txg, bp, spa_claim_notify, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_SCRUB))); } static int zil_claim_log_record(zilog_t *zilog, lr_t *lrc, void *tx, uint64_t first_txg) { lr_write_t *lr = (lr_write_t *)lrc; int error; if (lrc->lrc_txtype != TX_WRITE) return (0); /* * If the block is not readable, don't claim it. This can happen * in normal operation when a log block is written to disk before * some of the dmu_sync() blocks it points to. In this case, the * transaction cannot have been committed to anyone (we would have * waited for all writes to be stable first), so it is semantically * correct to declare this the end of the log. */ if (lr->lr_blkptr.blk_birth >= first_txg) { error = zil_read_log_data(zilog, lr, NULL); if (error != 0) return (error); } return (zil_claim_log_block(zilog, &lr->lr_blkptr, tx, first_txg)); } /* ARGSUSED */ static int zil_free_log_block(zilog_t *zilog, blkptr_t *bp, void *tx, uint64_t claim_txg) { zio_free_zil(zilog->zl_spa, dmu_tx_get_txg(tx), bp); return (0); } static int zil_free_log_record(zilog_t *zilog, lr_t *lrc, void *tx, uint64_t claim_txg) { lr_write_t *lr = (lr_write_t *)lrc; blkptr_t *bp = &lr->lr_blkptr; /* * If we previously claimed it, we need to free it. */ if (claim_txg != 0 && lrc->lrc_txtype == TX_WRITE && bp->blk_birth >= claim_txg && zil_bp_tree_add(zilog, bp) == 0 && !BP_IS_HOLE(bp)) zio_free(zilog->zl_spa, dmu_tx_get_txg(tx), bp); return (0); } +static int +zil_lwb_vdev_compare(const void *x1, const void *x2) +{ + const uint64_t v1 = ((zil_vdev_node_t *)x1)->zv_vdev; + const uint64_t v2 = ((zil_vdev_node_t *)x2)->zv_vdev; + + return (AVL_CMP(v1, v2)); +} + static lwb_t * zil_alloc_lwb(zilog_t *zilog, blkptr_t *bp, boolean_t slog, uint64_t txg, boolean_t fastwrite) { lwb_t *lwb; lwb = kmem_cache_alloc(zil_lwb_cache, KM_SLEEP); lwb->lwb_zilog = zilog; lwb->lwb_blk = *bp; lwb->lwb_fastwrite = fastwrite; lwb->lwb_slog = slog; + lwb->lwb_state = LWB_STATE_CLOSED; lwb->lwb_buf = zio_buf_alloc(BP_GET_LSIZE(bp)); lwb->lwb_max_txg = txg; - lwb->lwb_zio = NULL; + lwb->lwb_write_zio = NULL; + lwb->lwb_root_zio = NULL; lwb->lwb_tx = NULL; + lwb->lwb_issued_timestamp = 0; if (BP_GET_CHECKSUM(bp) == ZIO_CHECKSUM_ZILOG2) { lwb->lwb_nused = sizeof (zil_chain_t); lwb->lwb_sz = BP_GET_LSIZE(bp); } else { lwb->lwb_nused = 0; lwb->lwb_sz = BP_GET_LSIZE(bp) - sizeof (zil_chain_t); } mutex_enter(&zilog->zl_lock); list_insert_tail(&zilog->zl_lwb_list, lwb); mutex_exit(&zilog->zl_lock); + ASSERT(!MUTEX_HELD(&lwb->lwb_vdev_lock)); + ASSERT(avl_is_empty(&lwb->lwb_vdev_tree)); + ASSERT(list_is_empty(&lwb->lwb_waiters)); + ASSERT(list_is_empty(&lwb->lwb_itxs)); + return (lwb); } +static void +zil_free_lwb(zilog_t *zilog, lwb_t *lwb) +{ + ASSERT(MUTEX_HELD(&zilog->zl_lock)); + ASSERT(!MUTEX_HELD(&lwb->lwb_vdev_lock)); + ASSERT(list_is_empty(&lwb->lwb_waiters)); + + if (lwb->lwb_state == LWB_STATE_OPENED) { + avl_tree_t *t = &lwb->lwb_vdev_tree; + void *cookie = NULL; + zil_vdev_node_t *zv; + itx_t *itx; + + while ((zv = avl_destroy_nodes(t, &cookie)) != NULL) + kmem_free(zv, sizeof (*zv)); + + while ((itx = list_head(&lwb->lwb_itxs)) != NULL) { + if (itx->itx_callback != NULL) + itx->itx_callback(itx->itx_callback_data); + list_remove(&lwb->lwb_itxs, itx); + zil_itx_destroy(itx); + } + + ASSERT3P(lwb->lwb_root_zio, !=, NULL); + ASSERT3P(lwb->lwb_write_zio, !=, NULL); + + zio_cancel(lwb->lwb_root_zio); + zio_cancel(lwb->lwb_write_zio); + + lwb->lwb_root_zio = NULL; + lwb->lwb_write_zio = NULL; + } else { + ASSERT3S(lwb->lwb_state, !=, LWB_STATE_ISSUED); + } + + ASSERT(avl_is_empty(&lwb->lwb_vdev_tree)); + ASSERT(list_is_empty(&lwb->lwb_itxs)); + ASSERT3P(lwb->lwb_write_zio, ==, NULL); + ASSERT3P(lwb->lwb_root_zio, ==, NULL); + + /* + * Clear the zilog's field to indicate this lwb is no longer + * valid, and prevent use-after-free errors. + */ + if (zilog->zl_last_lwb_opened == lwb) + zilog->zl_last_lwb_opened = NULL; + + kmem_cache_free(zil_lwb_cache, lwb); +} + /* * Called when we create in-memory log transactions so that we know * to cleanup the itxs at the end of spa_sync(). */ void zilog_dirty(zilog_t *zilog, uint64_t txg) { dsl_pool_t *dp = zilog->zl_dmu_pool; dsl_dataset_t *ds = dmu_objset_ds(zilog->zl_os); + ASSERT(spa_writeable(zilog->zl_spa)); + if (ds->ds_is_snapshot) panic("dirtying snapshot!"); if (txg_list_add(&dp->dp_dirty_zilogs, zilog, txg)) { /* up the hold count until we can be written out */ dmu_buf_add_ref(ds->ds_dbuf, zilog); + + zilog->zl_dirty_max_txg = MAX(txg, zilog->zl_dirty_max_txg); } } /* * Determine if the zil is dirty in the specified txg. Callers wanting to * ensure that the dirty state does not change must hold the itxg_lock for * the specified txg. Holding the lock will ensure that the zil cannot be * dirtied (zil_itx_assign) or cleaned (zil_clean) while we check its current * state. */ boolean_t zilog_is_dirty_in_txg(zilog_t *zilog, uint64_t txg) { dsl_pool_t *dp = zilog->zl_dmu_pool; if (txg_list_member(&dp->dp_dirty_zilogs, zilog, txg & TXG_MASK)) return (B_TRUE); return (B_FALSE); } /* * Determine if the zil is dirty. The zil is considered dirty if it has * any pending itx records that have not been cleaned by zil_clean(). */ boolean_t zilog_is_dirty(zilog_t *zilog) { dsl_pool_t *dp = zilog->zl_dmu_pool; for (int t = 0; t < TXG_SIZE; t++) { if (txg_list_member(&dp->dp_dirty_zilogs, zilog, t)) return (B_TRUE); } return (B_FALSE); } /* * Create an on-disk intent log. */ static lwb_t * zil_create(zilog_t *zilog) { const zil_header_t *zh = zilog->zl_header; lwb_t *lwb = NULL; uint64_t txg = 0; dmu_tx_t *tx = NULL; blkptr_t blk; int error = 0; boolean_t fastwrite = FALSE; boolean_t slog = FALSE; /* * Wait for any previous destroy to complete. */ txg_wait_synced(zilog->zl_dmu_pool, zilog->zl_destroy_txg); ASSERT(zh->zh_claim_txg == 0); ASSERT(zh->zh_replay_seq == 0); blk = zh->zh_log; /* * Allocate an initial log block if: * - there isn't one already * - the existing block is the wrong endianness */ if (BP_IS_HOLE(&blk) || BP_SHOULD_BYTESWAP(&blk)) { tx = dmu_tx_create(zilog->zl_os); - VERIFY(dmu_tx_assign(tx, TXG_WAIT) == 0); + VERIFY0(dmu_tx_assign(tx, TXG_WAIT)); dsl_dataset_dirty(dmu_objset_ds(zilog->zl_os), tx); txg = dmu_tx_get_txg(tx); if (!BP_IS_HOLE(&blk)) { zio_free_zil(zilog->zl_spa, txg, &blk); BP_ZERO(&blk); } error = zio_alloc_zil(zilog->zl_spa, zilog->zl_os, txg, &blk, ZIL_MIN_BLKSZ, &slog); fastwrite = TRUE; if (error == 0) zil_init_log_chain(zilog, &blk); } /* - * Allocate a log write buffer (lwb) for the first log block. + * Allocate a log write block (lwb) for the first log block. */ if (error == 0) lwb = zil_alloc_lwb(zilog, &blk, slog, txg, fastwrite); /* * If we just allocated the first log block, commit our transaction * and wait for zil_sync() to stuff the block poiner into zh_log. * (zh is part of the MOS, so we cannot modify it in open context.) */ if (tx != NULL) { dmu_tx_commit(tx); txg_wait_synced(zilog->zl_dmu_pool, txg); } ASSERT(bcmp(&blk, &zh->zh_log, sizeof (blk)) == 0); return (lwb); } /* - * In one tx, free all log blocks and clear the log header. - * If keep_first is set, then we're replaying a log with no content. - * We want to keep the first block, however, so that the first - * synchronous transaction doesn't require a txg_wait_synced() - * in zil_create(). We don't need to txg_wait_synced() here either - * when keep_first is set, because both zil_create() and zil_destroy() - * will wait for any in-progress destroys to complete. + * In one tx, free all log blocks and clear the log header. If keep_first + * is set, then we're replaying a log with no content. We want to keep the + * first block, however, so that the first synchronous transaction doesn't + * require a txg_wait_synced() in zil_create(). We don't need to + * txg_wait_synced() here either when keep_first is set, because both + * zil_create() and zil_destroy() will wait for any in-progress destroys + * to complete. */ void zil_destroy(zilog_t *zilog, boolean_t keep_first) { const zil_header_t *zh = zilog->zl_header; lwb_t *lwb; dmu_tx_t *tx; uint64_t txg; /* * Wait for any previous destroy to complete. */ txg_wait_synced(zilog->zl_dmu_pool, zilog->zl_destroy_txg); zilog->zl_old_header = *zh; /* debugging aid */ if (BP_IS_HOLE(&zh->zh_log)) return; tx = dmu_tx_create(zilog->zl_os); - VERIFY(dmu_tx_assign(tx, TXG_WAIT) == 0); + VERIFY0(dmu_tx_assign(tx, TXG_WAIT)); dsl_dataset_dirty(dmu_objset_ds(zilog->zl_os), tx); txg = dmu_tx_get_txg(tx); mutex_enter(&zilog->zl_lock); ASSERT3U(zilog->zl_destroy_txg, <, txg); zilog->zl_destroy_txg = txg; zilog->zl_keep_first = keep_first; if (!list_is_empty(&zilog->zl_lwb_list)) { ASSERT(zh->zh_claim_txg == 0); VERIFY(!keep_first); while ((lwb = list_head(&zilog->zl_lwb_list)) != NULL) { - ASSERT(lwb->lwb_zio == NULL); if (lwb->lwb_fastwrite) metaslab_fastwrite_unmark(zilog->zl_spa, &lwb->lwb_blk); + list_remove(&zilog->zl_lwb_list, lwb); if (lwb->lwb_buf != NULL) zio_buf_free(lwb->lwb_buf, lwb->lwb_sz); - zio_free_zil(zilog->zl_spa, txg, &lwb->lwb_blk); - kmem_cache_free(zil_lwb_cache, lwb); + zio_free(zilog->zl_spa, txg, &lwb->lwb_blk); + zil_free_lwb(zilog, lwb); } } else if (!keep_first) { zil_destroy_sync(zilog, tx); } mutex_exit(&zilog->zl_lock); dmu_tx_commit(tx); } void zil_destroy_sync(zilog_t *zilog, dmu_tx_t *tx) { ASSERT(list_is_empty(&zilog->zl_lwb_list)); (void) zil_parse(zilog, zil_free_log_block, zil_free_log_record, tx, zilog->zl_header->zh_claim_txg, B_FALSE); } int zil_claim(dsl_pool_t *dp, dsl_dataset_t *ds, void *txarg) { dmu_tx_t *tx = txarg; uint64_t first_txg = dmu_tx_get_txg(tx); zilog_t *zilog; zil_header_t *zh; objset_t *os; int error; error = dmu_objset_own_obj(dp, ds->ds_object, DMU_OST_ANY, B_FALSE, B_FALSE, FTAG, &os); if (error != 0) { /* * EBUSY indicates that the objset is inconsistent, in which * case it can not have a ZIL. */ if (error != EBUSY) { cmn_err(CE_WARN, "can't open objset for %llu, error %u", (unsigned long long)ds->ds_object, error); } return (0); } zilog = dmu_objset_zil(os); zh = zil_header_in_syncing_context(zilog); if (spa_get_log_state(zilog->zl_spa) == SPA_LOG_CLEAR) { if (!BP_IS_HOLE(&zh->zh_log)) zio_free_zil(zilog->zl_spa, first_txg, &zh->zh_log); BP_ZERO(&zh->zh_log); if (os->os_encrypted) os->os_next_write_raw = B_TRUE; dsl_dataset_dirty(dmu_objset_ds(os), tx); dmu_objset_disown(os, B_FALSE, FTAG); return (0); } /* * Claim all log blocks if we haven't already done so, and remember * the highest claimed sequence number. This ensures that if we can * read only part of the log now (e.g. due to a missing device), * but we can read the entire log later, we will not try to replay * or destroy beyond the last block we successfully claimed. */ ASSERT3U(zh->zh_claim_txg, <=, first_txg); if (zh->zh_claim_txg == 0 && !BP_IS_HOLE(&zh->zh_log)) { (void) zil_parse(zilog, zil_claim_log_block, zil_claim_log_record, tx, first_txg, B_FALSE); zh->zh_claim_txg = first_txg; zh->zh_claim_blk_seq = zilog->zl_parse_blk_seq; zh->zh_claim_lr_seq = zilog->zl_parse_lr_seq; if (zilog->zl_parse_lr_count || zilog->zl_parse_blk_count > 1) zh->zh_flags |= ZIL_REPLAY_NEEDED; zh->zh_flags |= ZIL_CLAIM_LR_SEQ_VALID; dsl_dataset_dirty(dmu_objset_ds(os), tx); } ASSERT3U(first_txg, ==, (spa_last_synced_txg(zilog->zl_spa) + 1)); dmu_objset_disown(os, B_FALSE, FTAG); return (0); } /* * Check the log by walking the log chain. * Checksum errors are ok as they indicate the end of the chain. * Any other error (no device or read failure) returns an error. */ /* ARGSUSED */ int zil_check_log_chain(dsl_pool_t *dp, dsl_dataset_t *ds, void *tx) { zilog_t *zilog; objset_t *os; blkptr_t *bp; int error; ASSERT(tx == NULL); error = dmu_objset_from_ds(ds, &os); if (error != 0) { cmn_err(CE_WARN, "can't open objset %llu, error %d", (unsigned long long)ds->ds_object, error); return (0); } zilog = dmu_objset_zil(os); bp = (blkptr_t *)&zilog->zl_header->zh_log; /* * Check the first block and determine if it's on a log device * which may have been removed or faulted prior to loading this * pool. If so, there's no point in checking the rest of the log * as its content should have already been synced to the pool. */ if (!BP_IS_HOLE(bp)) { vdev_t *vd; boolean_t valid = B_TRUE; spa_config_enter(os->os_spa, SCL_STATE, FTAG, RW_READER); vd = vdev_lookup_top(os->os_spa, DVA_GET_VDEV(&bp->blk_dva[0])); if (vd->vdev_islog && vdev_is_dead(vd)) valid = vdev_log_state_valid(vd); spa_config_exit(os->os_spa, SCL_STATE, FTAG); if (!valid) return (0); } /* * Because tx == NULL, zil_claim_log_block() will not actually claim * any blocks, but just determine whether it is possible to do so. * In addition to checking the log chain, zil_claim_log_block() * will invoke zio_claim() with a done func of spa_claim_notify(), * which will update spa_max_claim_txg. See spa_load() for details. */ error = zil_parse(zilog, zil_claim_log_block, zil_claim_log_record, tx, zilog->zl_header->zh_claim_txg ? -1ULL : spa_first_txg(os->os_spa), B_FALSE); return ((error == ECKSUM || error == ENOENT) ? 0 : error); } -static int -zil_vdev_compare(const void *x1, const void *x2) +/* + * When an itx is "skipped", this function is used to properly mark the + * waiter as "done, and signal any thread(s) waiting on it. An itx can + * be skipped (and not committed to an lwb) for a variety of reasons, + * one of them being that the itx was committed via spa_sync(), prior to + * it being committed to an lwb; this can happen if a thread calling + * zil_commit() is racing with spa_sync(). + */ +static void +zil_commit_waiter_skip(zil_commit_waiter_t *zcw) { - const uint64_t v1 = ((zil_vdev_node_t *)x1)->zv_vdev; - const uint64_t v2 = ((zil_vdev_node_t *)x2)->zv_vdev; + mutex_enter(&zcw->zcw_lock); + ASSERT3B(zcw->zcw_done, ==, B_FALSE); + zcw->zcw_done = B_TRUE; + cv_broadcast(&zcw->zcw_cv); + mutex_exit(&zcw->zcw_lock); +} - return (AVL_CMP(v1, v2)); +/* + * This function is used when the given waiter is to be linked into an + * lwb's "lwb_waiter" list; i.e. when the itx is committed to the lwb. + * At this point, the waiter will no longer be referenced by the itx, + * and instead, will be referenced by the lwb. + */ +static void +zil_commit_waiter_link_lwb(zil_commit_waiter_t *zcw, lwb_t *lwb) +{ + mutex_enter(&zcw->zcw_lock); + ASSERT(!list_link_active(&zcw->zcw_node)); + ASSERT3P(zcw->zcw_lwb, ==, NULL); + ASSERT3P(lwb, !=, NULL); + ASSERT(lwb->lwb_state == LWB_STATE_OPENED || + lwb->lwb_state == LWB_STATE_ISSUED); + + list_insert_tail(&lwb->lwb_waiters, zcw); + zcw->zcw_lwb = lwb; + mutex_exit(&zcw->zcw_lock); +} + +/* + * This function is used when zio_alloc_zil() fails to allocate a ZIL + * block, and the given waiter must be linked to the "nolwb waiters" + * list inside of zil_process_commit_list(). + */ +static void +zil_commit_waiter_link_nolwb(zil_commit_waiter_t *zcw, list_t *nolwb) +{ + mutex_enter(&zcw->zcw_lock); + ASSERT(!list_link_active(&zcw->zcw_node)); + ASSERT3P(zcw->zcw_lwb, ==, NULL); + list_insert_tail(nolwb, zcw); + mutex_exit(&zcw->zcw_lock); } void -zil_add_block(zilog_t *zilog, const blkptr_t *bp) +zil_lwb_add_block(lwb_t *lwb, const blkptr_t *bp) { - avl_tree_t *t = &zilog->zl_vdev_tree; + avl_tree_t *t = &lwb->lwb_vdev_tree; avl_index_t where; zil_vdev_node_t *zv, zvsearch; int ndvas = BP_GET_NDVAS(bp); int i; if (zfs_nocacheflush) return; - ASSERT(zilog->zl_writer); - - /* - * Even though we're zl_writer, we still need a lock because the - * zl_get_data() callbacks may have dmu_sync() done callbacks - * that will run concurrently. - */ - mutex_enter(&zilog->zl_vdev_lock); + mutex_enter(&lwb->lwb_vdev_lock); for (i = 0; i < ndvas; i++) { zvsearch.zv_vdev = DVA_GET_VDEV(&bp->blk_dva[i]); if (avl_find(t, &zvsearch, &where) == NULL) { zv = kmem_alloc(sizeof (*zv), KM_SLEEP); zv->zv_vdev = zvsearch.zv_vdev; avl_insert(t, zv, where); } } - mutex_exit(&zilog->zl_vdev_lock); + mutex_exit(&lwb->lwb_vdev_lock); } +void +zil_lwb_add_txg(lwb_t *lwb, uint64_t txg) +{ + lwb->lwb_max_txg = MAX(lwb->lwb_max_txg, txg); +} + +/* + * This function is a called after all VDEVs associated with a given lwb + * write have completed their DKIOCFLUSHWRITECACHE command; or as soon + * as the lwb write completes, if "zfs_nocacheflush" is set. + * + * The intention is for this function to be called as soon as the + * contents of an lwb are considered "stable" on disk, and will survive + * any sudden loss of power. At this point, any threads waiting for the + * lwb to reach this state are signalled, and the "waiter" structures + * are marked "done". + */ static void -zil_flush_vdevs(zilog_t *zilog) +zil_lwb_flush_vdevs_done(zio_t *zio) { - spa_t *spa = zilog->zl_spa; - avl_tree_t *t = &zilog->zl_vdev_tree; - void *cookie = NULL; - zil_vdev_node_t *zv; - zio_t *zio; + lwb_t *lwb = zio->io_private; + zilog_t *zilog = lwb->lwb_zilog; + dmu_tx_t *tx = lwb->lwb_tx; + zil_commit_waiter_t *zcw; + itx_t *itx; + + spa_config_exit(zilog->zl_spa, SCL_STATE, lwb); + + zio_buf_free(lwb->lwb_buf, lwb->lwb_sz); - ASSERT(zilog->zl_writer); + mutex_enter(&zilog->zl_lock); /* - * We don't need zl_vdev_lock here because we're the zl_writer, - * and all zl_get_data() callbacks are done. + * Ensure the lwb buffer pointer is cleared before releasing the + * txg. If we have had an allocation failure and the txg is + * waiting to sync then we want zil_sync() to remove the lwb so + * that it's not picked up as the next new one in + * zil_process_commit_list(). zil_sync() will only remove the + * lwb if lwb_buf is null. */ - if (avl_numnodes(t) == 0) - return; + lwb->lwb_buf = NULL; + lwb->lwb_tx = NULL; - spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); + ASSERT3U(lwb->lwb_issued_timestamp, >, 0); + zilog->zl_last_lwb_latency = gethrtime() - lwb->lwb_issued_timestamp; - zio = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL); + lwb->lwb_root_zio = NULL; + lwb->lwb_state = LWB_STATE_DONE; - while ((zv = avl_destroy_nodes(t, &cookie)) != NULL) { - vdev_t *vd = vdev_lookup_top(spa, zv->zv_vdev); - if (vd != NULL) - zio_flush(zio, vd); - kmem_free(zv, sizeof (*zv)); + if (zilog->zl_last_lwb_opened == lwb) { + /* + * Remember the highest committed log sequence number + * for ztest. We only update this value when all the log + * writes succeeded, because ztest wants to ASSERT that + * it got the whole log chain. + */ + zilog->zl_commit_lr_seq = zilog->zl_lr_seq; + } + + while ((itx = list_head(&lwb->lwb_itxs)) != NULL) { + list_remove(&lwb->lwb_itxs, itx); + zil_itx_destroy(itx); + } + + while ((zcw = list_head(&lwb->lwb_waiters)) != NULL) { + mutex_enter(&zcw->zcw_lock); + + ASSERT(list_link_active(&zcw->zcw_node)); + list_remove(&lwb->lwb_waiters, zcw); + + ASSERT3P(zcw->zcw_lwb, ==, lwb); + zcw->zcw_lwb = NULL; + + zcw->zcw_zio_error = zio->io_error; + + ASSERT3B(zcw->zcw_done, ==, B_FALSE); + zcw->zcw_done = B_TRUE; + cv_broadcast(&zcw->zcw_cv); + + mutex_exit(&zcw->zcw_lock); } + mutex_exit(&zilog->zl_lock); + /* - * Wait for all the flushes to complete. Not all devices actually - * support the DKIOCFLUSHWRITECACHE ioctl, so it's OK if it fails. + * Now that we've written this log block, we have a stable pointer + * to the next block in the chain, so it's OK to let the txg in + * which we allocated the next block sync. */ - (void) zio_wait(zio); - - spa_config_exit(spa, SCL_STATE, FTAG); + dmu_tx_commit(tx); } /* - * Function called when a log block write completes + * This is called when an lwb write completes. This means, this specific + * lwb was written to disk, and all dependent lwb have also been + * written to disk. + * + * At this point, a DKIOCFLUSHWRITECACHE command hasn't been issued to + * the VDEVs involved in writing out this specific lwb. The lwb will be + * "done" once zil_lwb_flush_vdevs_done() is called, which occurs in the + * zio completion callback for the lwb's root zio. */ static void zil_lwb_write_done(zio_t *zio) { lwb_t *lwb = zio->io_private; + spa_t *spa = zio->io_spa; zilog_t *zilog = lwb->lwb_zilog; - dmu_tx_t *tx = lwb->lwb_tx; + avl_tree_t *t = &lwb->lwb_vdev_tree; + void *cookie = NULL; + zil_vdev_node_t *zv; + + ASSERT3S(spa_config_held(spa, SCL_STATE, RW_READER), !=, 0); ASSERT(BP_GET_COMPRESS(zio->io_bp) == ZIO_COMPRESS_OFF); ASSERT(BP_GET_TYPE(zio->io_bp) == DMU_OT_INTENT_LOG); ASSERT(BP_GET_LEVEL(zio->io_bp) == 0); ASSERT(BP_GET_BYTEORDER(zio->io_bp) == ZFS_HOST_BYTEORDER); ASSERT(!BP_IS_GANG(zio->io_bp)); ASSERT(!BP_IS_HOLE(zio->io_bp)); ASSERT(BP_GET_FILL(zio->io_bp) == 0); - /* - * Ensure the lwb buffer pointer is cleared before releasing - * the txg. If we have had an allocation failure and - * the txg is waiting to sync then we want want zil_sync() - * to remove the lwb so that it's not picked up as the next new - * one in zil_commit_writer(). zil_sync() will only remove - * the lwb if lwb_buf is null. - */ abd_put(zio->io_abd); - zio_buf_free(lwb->lwb_buf, lwb->lwb_sz); + + ASSERT3S(lwb->lwb_state, ==, LWB_STATE_ISSUED); + mutex_enter(&zilog->zl_lock); - lwb->lwb_zio = NULL; + lwb->lwb_write_zio = NULL; lwb->lwb_fastwrite = FALSE; - lwb->lwb_buf = NULL; - lwb->lwb_tx = NULL; mutex_exit(&zilog->zl_lock); + if (avl_numnodes(t) == 0) + return; + /* - * Now that we've written this log block, we have a stable pointer - * to the next block in the chain, so it's OK to let the txg in - * which we allocated the next block sync. + * If there was an IO error, we're not going to call zio_flush() + * on these vdevs, so we simply empty the tree and free the + * nodes. We avoid calling zio_flush() since there isn't any + * good reason for doing so, after the lwb block failed to be + * written out. */ - dmu_tx_commit(tx); + if (zio->io_error != 0) { + while ((zv = avl_destroy_nodes(t, &cookie)) != NULL) + kmem_free(zv, sizeof (*zv)); + return; + } + + while ((zv = avl_destroy_nodes(t, &cookie)) != NULL) { + vdev_t *vd = vdev_lookup_top(spa, zv->zv_vdev); + if (vd != NULL) + zio_flush(lwb->lwb_root_zio, vd); + kmem_free(zv, sizeof (*zv)); + } } /* - * Initialize the io for a log block. + * This function's purpose is to "open" an lwb such that it is ready to + * accept new itxs being committed to it. To do this, the lwb's zio + * structures are created, and linked to the lwb. This function is + * idempotent; if the passed in lwb has already been opened, this + * function is essentially a no-op. */ static void -zil_lwb_write_init(zilog_t *zilog, lwb_t *lwb) +zil_lwb_write_open(zilog_t *zilog, lwb_t *lwb) { zbookmark_phys_t zb; zio_priority_t prio; + ASSERT(MUTEX_HELD(&zilog->zl_writer_lock)); + ASSERT3P(lwb, !=, NULL); + EQUIV(lwb->lwb_root_zio == NULL, lwb->lwb_state == LWB_STATE_CLOSED); + EQUIV(lwb->lwb_root_zio != NULL, lwb->lwb_state == LWB_STATE_OPENED); + SET_BOOKMARK(&zb, lwb->lwb_blk.blk_cksum.zc_word[ZIL_ZC_OBJSET], ZB_ZIL_OBJECT, ZB_ZIL_LEVEL, lwb->lwb_blk.blk_cksum.zc_word[ZIL_ZC_SEQ]); - if (zilog->zl_root_zio == NULL) { - zilog->zl_root_zio = zio_root(zilog->zl_spa, NULL, NULL, - ZIO_FLAG_CANFAIL); - } - /* Lock so zil_sync() doesn't fastwrite_unmark after zio is created */ mutex_enter(&zilog->zl_lock); - if (lwb->lwb_zio == NULL) { + if (lwb->lwb_root_zio == NULL) { abd_t *lwb_abd = abd_get_from_buf(lwb->lwb_buf, BP_GET_LSIZE(&lwb->lwb_blk)); + if (!lwb->lwb_fastwrite) { metaslab_fastwrite_mark(zilog->zl_spa, &lwb->lwb_blk); lwb->lwb_fastwrite = 1; } + if (!lwb->lwb_slog || zilog->zl_cur_used <= zil_slog_bulk) prio = ZIO_PRIORITY_SYNC_WRITE; else prio = ZIO_PRIORITY_ASYNC_WRITE; - lwb->lwb_zio = zio_rewrite(zilog->zl_root_zio, zilog->zl_spa, - 0, &lwb->lwb_blk, lwb_abd, BP_GET_LSIZE(&lwb->lwb_blk), - zil_lwb_write_done, lwb, prio, - ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_PROPAGATE | + + lwb->lwb_root_zio = zio_root(zilog->zl_spa, + zil_lwb_flush_vdevs_done, lwb, ZIO_FLAG_CANFAIL); + ASSERT3P(lwb->lwb_root_zio, !=, NULL); + + lwb->lwb_write_zio = zio_rewrite(lwb->lwb_root_zio, + zilog->zl_spa, 0, &lwb->lwb_blk, lwb_abd, + BP_GET_LSIZE(&lwb->lwb_blk), zil_lwb_write_done, lwb, + prio, ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_FASTWRITE, &zb); + ASSERT3P(lwb->lwb_write_zio, !=, NULL); + + lwb->lwb_state = LWB_STATE_OPENED; + + /* + * The zilog's "zl_last_lwb_opened" field is used to + * build the lwb/zio dependency chain, which is used to + * preserve the ordering of lwb completions that is + * required by the semantics of the ZIL. Each new lwb + * zio becomes a parent of the "previous" lwb zio, such + * that the new lwb's zio cannot complete until the + * "previous" lwb's zio completes. + * + * This is required by the semantics of zil_commit(); + * the commit waiters attached to the lwbs will be woken + * in the lwb zio's completion callback, so this zio + * dependency graph ensures the waiters are woken in the + * correct order (the same order the lwbs were created). + */ + lwb_t *last_lwb_opened = zilog->zl_last_lwb_opened; + if (last_lwb_opened != NULL && + last_lwb_opened->lwb_state != LWB_STATE_DONE) { + ASSERT(last_lwb_opened->lwb_state == LWB_STATE_OPENED || + last_lwb_opened->lwb_state == LWB_STATE_ISSUED); + ASSERT3P(last_lwb_opened->lwb_root_zio, !=, NULL); + zio_add_child(lwb->lwb_root_zio, + last_lwb_opened->lwb_root_zio); + } + zilog->zl_last_lwb_opened = lwb; } mutex_exit(&zilog->zl_lock); + + ASSERT3P(lwb->lwb_root_zio, !=, NULL); + ASSERT3P(lwb->lwb_write_zio, !=, NULL); + ASSERT3S(lwb->lwb_state, ==, LWB_STATE_OPENED); } /* * Define a limited set of intent log block sizes. * * These must be a multiple of 4KB. Note only the amount used (again * aligned to 4KB) actually gets written. However, we can't always just * allocate SPA_OLD_MAXBLOCKSIZE as the slog space could be exhausted. */ uint64_t zil_block_buckets[] = { 4096, /* non TX_WRITE */ 8192+4096, /* data base */ 32*1024 + 4096, /* NFS writes */ UINT64_MAX }; /* * Start a log block write and advance to the next log block. * Calls are serialized. */ static lwb_t * -zil_lwb_write_start(zilog_t *zilog, lwb_t *lwb) +zil_lwb_write_issue(zilog_t *zilog, lwb_t *lwb) { lwb_t *nlwb = NULL; zil_chain_t *zilc; spa_t *spa = zilog->zl_spa; blkptr_t *bp; dmu_tx_t *tx; uint64_t txg; uint64_t zil_blksz, wsz; int i, error; boolean_t slog; + ASSERT(MUTEX_HELD(&zilog->zl_writer_lock)); + ASSERT3P(lwb->lwb_root_zio, !=, NULL); + ASSERT3P(lwb->lwb_write_zio, !=, NULL); + ASSERT3S(lwb->lwb_state, ==, LWB_STATE_OPENED); + if (BP_GET_CHECKSUM(&lwb->lwb_blk) == ZIO_CHECKSUM_ZILOG2) { zilc = (zil_chain_t *)lwb->lwb_buf; bp = &zilc->zc_next_blk; } else { zilc = (zil_chain_t *)(lwb->lwb_buf + lwb->lwb_sz); bp = &zilc->zc_next_blk; } ASSERT(lwb->lwb_nused <= lwb->lwb_sz); /* * Allocate the next block and save its address in this block * before writing it in order to establish the log chain. * Note that if the allocation of nlwb synced before we wrote * the block that points at it (lwb), we'd leak it if we crashed. * Therefore, we don't do dmu_tx_commit() until zil_lwb_write_done(). * We dirty the dataset to ensure that zil_sync() will be called * to clean up in the event of allocation failure or I/O failure. */ + tx = dmu_tx_create(zilog->zl_os); /* * Since we are not going to create any new dirty data and we can even * help with clearing the existing dirty data, we should not be subject * to the dirty data based delays. * We (ab)use TXG_WAITED to bypass the delay mechanism. * One side effect from using TXG_WAITED is that dmu_tx_assign() can * fail if the pool is suspended. Those are dramatic circumstances, * so we return NULL to signal that the normal ZIL processing is not * possible and txg_wait_synced() should be used to ensure that the data * is on disk. */ error = dmu_tx_assign(tx, TXG_WAITED); if (error != 0) { - ASSERT3S(error, ==, EIO); + ASSERT(error == EIO || error == ERESTART); dmu_tx_abort(tx); return (NULL); } dsl_dataset_dirty(dmu_objset_ds(zilog->zl_os), tx); txg = dmu_tx_get_txg(tx); lwb->lwb_tx = tx; /* * Log blocks are pre-allocated. Here we select the size of the next * block, based on size used in the last block. * - first find the smallest bucket that will fit the block from a * limited set of block sizes. This is because it's faster to write * blocks allocated from the same metaslab as they are adjacent or * close. * - next find the maximum from the new suggested size and an array of * previous sizes. This lessens a picket fence effect of wrongly * guesssing the size if we have a stream of say 2k, 64k, 2k, 64k * requests. * * Note we only write what is used, but we can't just allocate * the maximum block size because we can exhaust the available * pool log space. */ zil_blksz = zilog->zl_cur_used + sizeof (zil_chain_t); for (i = 0; zil_blksz > zil_block_buckets[i]; i++) continue; zil_blksz = zil_block_buckets[i]; if (zil_blksz == UINT64_MAX) zil_blksz = SPA_OLD_MAXBLOCKSIZE; zilog->zl_prev_blks[zilog->zl_prev_rotor] = zil_blksz; for (i = 0; i < ZIL_PREV_BLKS; i++) zil_blksz = MAX(zil_blksz, zilog->zl_prev_blks[i]); zilog->zl_prev_rotor = (zilog->zl_prev_rotor + 1) & (ZIL_PREV_BLKS - 1); BP_ZERO(bp); error = zio_alloc_zil(spa, zilog->zl_os, txg, bp, zil_blksz, &slog); if (slog) { ZIL_STAT_BUMP(zil_itx_metaslab_slog_count); ZIL_STAT_INCR(zil_itx_metaslab_slog_bytes, lwb->lwb_nused); } else { ZIL_STAT_BUMP(zil_itx_metaslab_normal_count); ZIL_STAT_INCR(zil_itx_metaslab_normal_bytes, lwb->lwb_nused); } if (error == 0) { ASSERT3U(bp->blk_birth, ==, txg); bp->blk_cksum = lwb->lwb_blk.blk_cksum; bp->blk_cksum.zc_word[ZIL_ZC_SEQ]++; /* - * Allocate a new log write buffer (lwb). + * Allocate a new log write block (lwb). */ nlwb = zil_alloc_lwb(zilog, bp, slog, txg, TRUE); - - /* Record the block for later vdev flushing */ - zil_add_block(zilog, &lwb->lwb_blk); } if (BP_GET_CHECKSUM(&lwb->lwb_blk) == ZIO_CHECKSUM_ZILOG2) { /* For Slim ZIL only write what is used. */ wsz = P2ROUNDUP_TYPED(lwb->lwb_nused, ZIL_MIN_BLKSZ, uint64_t); ASSERT3U(wsz, <=, lwb->lwb_sz); - zio_shrink(lwb->lwb_zio, wsz); + zio_shrink(lwb->lwb_write_zio, wsz); } else { wsz = lwb->lwb_sz; } zilc->zc_pad = 0; zilc->zc_nused = lwb->lwb_nused; zilc->zc_eck.zec_cksum = lwb->lwb_blk.blk_cksum; /* * clear unused data for security */ bzero(lwb->lwb_buf + lwb->lwb_nused, wsz - lwb->lwb_nused); - zio_nowait(lwb->lwb_zio); /* Kick off the write for the old log block */ + spa_config_enter(zilog->zl_spa, SCL_STATE, lwb, RW_READER); + + zil_lwb_add_block(lwb, &lwb->lwb_blk); + lwb->lwb_issued_timestamp = gethrtime(); + lwb->lwb_state = LWB_STATE_ISSUED; + + zio_nowait(lwb->lwb_root_zio); + zio_nowait(lwb->lwb_write_zio); /* * If there was an allocation failure then nlwb will be null which * forces a txg_wait_synced(). */ return (nlwb); } static lwb_t * zil_lwb_commit(zilog_t *zilog, itx_t *itx, lwb_t *lwb) { lr_t *lrcb, *lrc; lr_write_t *lrwb, *lrw; char *lr_buf; uint64_t dlen, dnow, lwb_sp, reclen, txg; - if (lwb == NULL) - return (NULL); + ASSERT(MUTEX_HELD(&zilog->zl_writer_lock)); + ASSERT3P(lwb, !=, NULL); + ASSERT3P(lwb->lwb_buf, !=, NULL); + + zil_lwb_write_open(zilog, lwb); - ASSERT(lwb->lwb_buf != NULL); + lrc = &itx->itx_lr; + lrw = (lr_write_t *)lrc; + + /* + * A commit itx doesn't represent any on-disk state; instead + * it's simply used as a place holder on the commit list, and + * provides a mechanism for attaching a "commit waiter" onto the + * correct lwb (such that the waiter can be signalled upon + * completion of that lwb). Thus, we don't process this itx's + * log record if it's a commit itx (these itx's don't have log + * records), and instead link the itx's waiter onto the lwb's + * list of waiters. + * + * For more details, see the comment above zil_commit(). + */ + if (lrc->lrc_txtype == TX_COMMIT) { + zil_commit_waiter_link_lwb(itx->itx_private, lwb); + itx->itx_private = NULL; + return (lwb); + } - lrc = &itx->itx_lr; /* Common log record inside itx. */ - lrw = (lr_write_t *)lrc; /* Write log record inside itx. */ if (lrc->lrc_txtype == TX_WRITE && itx->itx_wr_state == WR_NEED_COPY) { dlen = P2ROUNDUP_TYPED( lrw->lr_length, sizeof (uint64_t), uint64_t); } else { dlen = 0; } reclen = lrc->lrc_reclen; zilog->zl_cur_used += (reclen + dlen); txg = lrc->lrc_txg; - zil_lwb_write_init(zilog, lwb); + ASSERT3U(zilog->zl_cur_used, <, UINT64_MAX - (reclen + dlen)); cont: /* * If this record won't fit in the current log block, start a new one. * For WR_NEED_COPY optimize layout for minimal number of chunks. */ lwb_sp = lwb->lwb_sz - lwb->lwb_nused; if (reclen > lwb_sp || (reclen + dlen > lwb_sp && lwb_sp < ZIL_MAX_WASTE_SPACE && (dlen % ZIL_MAX_LOG_DATA == 0 || lwb_sp < reclen + dlen % ZIL_MAX_LOG_DATA))) { - lwb = zil_lwb_write_start(zilog, lwb); + lwb = zil_lwb_write_issue(zilog, lwb); if (lwb == NULL) return (NULL); - zil_lwb_write_init(zilog, lwb); + zil_lwb_write_open(zilog, lwb); ASSERT(LWB_EMPTY(lwb)); lwb_sp = lwb->lwb_sz - lwb->lwb_nused; ASSERT3U(reclen + MIN(dlen, sizeof (uint64_t)), <=, lwb_sp); } dnow = MIN(dlen, lwb_sp - reclen); lr_buf = lwb->lwb_buf + lwb->lwb_nused; bcopy(lrc, lr_buf, reclen); lrcb = (lr_t *)lr_buf; /* Like lrc, but inside lwb. */ lrwb = (lr_write_t *)lrcb; /* Like lrw, but inside lwb. */ ZIL_STAT_BUMP(zil_itx_count); /* * If it's a write, fetch the data or get its blkptr as appropriate. */ if (lrc->lrc_txtype == TX_WRITE) { if (txg > spa_freeze_txg(zilog->zl_spa)) txg_wait_synced(zilog->zl_dmu_pool, txg); if (itx->itx_wr_state == WR_COPIED) { ZIL_STAT_BUMP(zil_itx_copied_count); ZIL_STAT_INCR(zil_itx_copied_bytes, lrw->lr_length); } else { char *dbuf; int error; if (itx->itx_wr_state == WR_NEED_COPY) { dbuf = lr_buf + reclen; lrcb->lrc_reclen += dnow; if (lrwb->lr_length > dnow) lrwb->lr_length = dnow; lrw->lr_offset += dnow; lrw->lr_length -= dnow; ZIL_STAT_BUMP(zil_itx_needcopy_count); ZIL_STAT_INCR(zil_itx_needcopy_bytes, lrw->lr_length); } else { - ASSERT(itx->itx_wr_state == WR_INDIRECT); + ASSERT3S(itx->itx_wr_state, ==, WR_INDIRECT); dbuf = NULL; ZIL_STAT_BUMP(zil_itx_indirect_count); ZIL_STAT_INCR(zil_itx_indirect_bytes, lrw->lr_length); } - error = zilog->zl_get_data( - itx->itx_private, lrwb, dbuf, lwb->lwb_zio); + + /* + * We pass in the "lwb_write_zio" rather than + * "lwb_root_zio" so that the "lwb_write_zio" + * becomes the parent of any zio's created by + * the "zl_get_data" callback. The vdevs are + * flushed after the "lwb_write_zio" completes, + * so we want to make sure that completion + * callback waits for these additional zio's, + * such that the vdevs used by those zio's will + * be included in the lwb's vdev tree, and those + * vdevs will be properly flushed. If we passed + * in "lwb_root_zio" here, then these additional + * vdevs may not be flushed; e.g. if these zio's + * completed after "lwb_write_zio" completed. + */ + error = zilog->zl_get_data(itx->itx_private, + lrwb, dbuf, lwb, lwb->lwb_write_zio); + if (error == EIO) { txg_wait_synced(zilog->zl_dmu_pool, txg); return (lwb); } if (error != 0) { ASSERT(error == ENOENT || error == EEXIST || error == EALREADY); return (lwb); } } } /* * We're actually making an entry, so update lrc_seq to be the * log record sequence number. Note that this is generally not * equal to the itx sequence number because not all transactions * are synchronous, and sometimes spa_sync() gets there first. */ - lrcb->lrc_seq = ++zilog->zl_lr_seq; /* we are single threaded */ + lrcb->lrc_seq = ++zilog->zl_lr_seq; lwb->lwb_nused += reclen + dnow; - lwb->lwb_max_txg = MAX(lwb->lwb_max_txg, txg); + + zil_lwb_add_txg(lwb, txg); + ASSERT3U(lwb->lwb_nused, <=, lwb->lwb_sz); ASSERT0(P2PHASE(lwb->lwb_nused, sizeof (uint64_t))); dlen -= dnow; if (dlen > 0) { zilog->zl_cur_used += reclen; goto cont; } return (lwb); } itx_t * zil_itx_create(uint64_t txtype, size_t lrsize) { size_t itxsize; itx_t *itx; lrsize = P2ROUNDUP_TYPED(lrsize, sizeof (uint64_t), size_t); itxsize = offsetof(itx_t, itx_lr) + lrsize; itx = zio_data_buf_alloc(itxsize); itx->itx_lr.lrc_txtype = txtype; itx->itx_lr.lrc_reclen = lrsize; itx->itx_lr.lrc_seq = 0; /* defensive */ itx->itx_sync = B_TRUE; /* default is synchronous */ itx->itx_callback = NULL; itx->itx_callback_data = NULL; itx->itx_size = itxsize; return (itx); } void zil_itx_destroy(itx_t *itx) { + IMPLY(itx->itx_lr.lrc_txtype == TX_COMMIT, itx->itx_callback == NULL); + IMPLY(itx->itx_callback != NULL, itx->itx_lr.lrc_txtype != TX_COMMIT); + + if (itx->itx_callback != NULL) + itx->itx_callback(itx->itx_callback_data); + zio_data_buf_free(itx, itx->itx_size); } /* * Free up the sync and async itxs. The itxs_t has already been detached * so no locks are needed. */ static void zil_itxg_clean(itxs_t *itxs) { itx_t *itx; list_t *list; avl_tree_t *t; void *cookie; itx_async_node_t *ian; list = &itxs->i_sync_list; while ((itx = list_head(list)) != NULL) { - if (itx->itx_callback != NULL) - itx->itx_callback(itx->itx_callback_data); + /* + * In the general case, commit itxs will not be found + * here, as they'll be committed to an lwb via + * zil_lwb_commit(), and free'd in that function. Having + * said that, it is still possible for commit itxs to be + * found here, due to the following race: + * + * - a thread calls zil_commit() which assigns the + * commit itx to a per-txg i_sync_list + * - zil_itxg_clean() is called (e.g. via spa_sync()) + * while the waiter is still on the i_sync_list + * + * There's nothing to prevent syncing the txg while the + * waiter is on the i_sync_list. This normally doesn't + * happen because spa_sync() is slower than zil_commit(), + * but if zil_commit() calls txg_wait_synced() (e.g. + * because zil_create() or zil_commit_writer_stall() is + * called) we will hit this case. + */ + if (itx->itx_lr.lrc_txtype == TX_COMMIT) + zil_commit_waiter_skip(itx->itx_private); + list_remove(list, itx); zil_itx_destroy(itx); } cookie = NULL; t = &itxs->i_async_tree; while ((ian = avl_destroy_nodes(t, &cookie)) != NULL) { list = &ian->ia_list; while ((itx = list_head(list)) != NULL) { - if (itx->itx_callback != NULL) - itx->itx_callback(itx->itx_callback_data); list_remove(list, itx); + /* commit itxs should never be on the async lists. */ + ASSERT3U(itx->itx_lr.lrc_txtype, !=, TX_COMMIT); zil_itx_destroy(itx); } list_destroy(list); kmem_free(ian, sizeof (itx_async_node_t)); } avl_destroy(t); kmem_free(itxs, sizeof (itxs_t)); } static int zil_aitx_compare(const void *x1, const void *x2) { const uint64_t o1 = ((itx_async_node_t *)x1)->ia_foid; const uint64_t o2 = ((itx_async_node_t *)x2)->ia_foid; return (AVL_CMP(o1, o2)); } /* * Remove all async itx with the given oid. */ static void zil_remove_async(zilog_t *zilog, uint64_t oid) { uint64_t otxg, txg; itx_async_node_t *ian; avl_tree_t *t; avl_index_t where; list_t clean_list; itx_t *itx; ASSERT(oid != 0); list_create(&clean_list, sizeof (itx_t), offsetof(itx_t, itx_node)); if (spa_freeze_txg(zilog->zl_spa) != UINT64_MAX) /* ziltest support */ otxg = ZILTEST_TXG; else otxg = spa_last_synced_txg(zilog->zl_spa) + 1; for (txg = otxg; txg < (otxg + TXG_CONCURRENT_STATES); txg++) { itxg_t *itxg = &zilog->zl_itxg[txg & TXG_MASK]; mutex_enter(&itxg->itxg_lock); if (itxg->itxg_txg != txg) { mutex_exit(&itxg->itxg_lock); continue; } /* * Locate the object node and append its list. */ t = &itxg->itxg_itxs->i_async_tree; ian = avl_find(t, &oid, &where); if (ian != NULL) list_move_tail(&clean_list, &ian->ia_list); mutex_exit(&itxg->itxg_lock); } while ((itx = list_head(&clean_list)) != NULL) { - if (itx->itx_callback != NULL) - itx->itx_callback(itx->itx_callback_data); list_remove(&clean_list, itx); + /* commit itxs should never be on the async lists. */ + ASSERT3U(itx->itx_lr.lrc_txtype, !=, TX_COMMIT); zil_itx_destroy(itx); } list_destroy(&clean_list); } void zil_itx_assign(zilog_t *zilog, itx_t *itx, dmu_tx_t *tx) { uint64_t txg; itxg_t *itxg; itxs_t *itxs, *clean = NULL; /* * Object ids can be re-instantiated in the next txg so * remove any async transactions to avoid future leaks. * This can happen if a fsync occurs on the re-instantiated * object for a WR_INDIRECT or WR_NEED_COPY write, which gets * the new file data and flushes a write record for the old object. */ if ((itx->itx_lr.lrc_txtype & ~TX_CI) == TX_REMOVE) zil_remove_async(zilog, itx->itx_oid); /* * Ensure the data of a renamed file is committed before the rename. */ if ((itx->itx_lr.lrc_txtype & ~TX_CI) == TX_RENAME) zil_async_to_sync(zilog, itx->itx_oid); if (spa_freeze_txg(zilog->zl_spa) != UINT64_MAX) txg = ZILTEST_TXG; else txg = dmu_tx_get_txg(tx); itxg = &zilog->zl_itxg[txg & TXG_MASK]; mutex_enter(&itxg->itxg_lock); itxs = itxg->itxg_itxs; if (itxg->itxg_txg != txg) { if (itxs != NULL) { /* * The zil_clean callback hasn't got around to cleaning * this itxg. Save the itxs for release below. * This should be rare. */ zfs_dbgmsg("zil_itx_assign: missed itx cleanup for " "txg %llu", itxg->itxg_txg); clean = itxg->itxg_itxs; } itxg->itxg_txg = txg; itxs = itxg->itxg_itxs = kmem_zalloc(sizeof (itxs_t), KM_SLEEP); list_create(&itxs->i_sync_list, sizeof (itx_t), offsetof(itx_t, itx_node)); avl_create(&itxs->i_async_tree, zil_aitx_compare, sizeof (itx_async_node_t), offsetof(itx_async_node_t, ia_node)); } if (itx->itx_sync) { list_insert_tail(&itxs->i_sync_list, itx); } else { avl_tree_t *t = &itxs->i_async_tree; uint64_t foid = LR_FOID_GET_OBJ(((lr_ooo_t *)&itx->itx_lr)->lr_foid); itx_async_node_t *ian; avl_index_t where; ian = avl_find(t, &foid, &where); if (ian == NULL) { ian = kmem_alloc(sizeof (itx_async_node_t), KM_SLEEP); list_create(&ian->ia_list, sizeof (itx_t), offsetof(itx_t, itx_node)); ian->ia_foid = foid; avl_insert(t, ian, where); } list_insert_tail(&ian->ia_list, itx); } itx->itx_lr.lrc_txg = dmu_tx_get_txg(tx); - zilog_dirty(zilog, txg); + + /* + * We don't want to dirty the ZIL using ZILTEST_TXG, because + * zil_clean() will never be called using ZILTEST_TXG. Thus, we + * need to be careful to always dirty the ZIL using the "real" + * TXG (not itxg_txg) even when the SPA is frozen. + */ + zilog_dirty(zilog, dmu_tx_get_txg(tx)); mutex_exit(&itxg->itxg_lock); /* Release the old itxs now we've dropped the lock */ if (clean != NULL) zil_itxg_clean(clean); } /* * If there are any in-memory intent log transactions which have now been * synced then start up a taskq to free them. We should only do this after we * have written out the uberblocks (i.e. txg has been comitted) so that * don't inadvertently clean out in-memory log records that would be required * by zil_commit(). */ void zil_clean(zilog_t *zilog, uint64_t synced_txg) { itxg_t *itxg = &zilog->zl_itxg[synced_txg & TXG_MASK]; itxs_t *clean_me; + ASSERT3U(synced_txg, <, ZILTEST_TXG); + mutex_enter(&itxg->itxg_lock); if (itxg->itxg_itxs == NULL || itxg->itxg_txg == ZILTEST_TXG) { mutex_exit(&itxg->itxg_lock); return; } ASSERT3U(itxg->itxg_txg, <=, synced_txg); ASSERT3U(itxg->itxg_txg, !=, 0); clean_me = itxg->itxg_itxs; itxg->itxg_itxs = NULL; itxg->itxg_txg = 0; mutex_exit(&itxg->itxg_lock); /* * Preferably start a task queue to free up the old itxs but * if taskq_dispatch can't allocate resources to do that then * free it in-line. This should be rare. Note, using TQ_SLEEP * created a bad performance problem. */ ASSERT3P(zilog->zl_dmu_pool, !=, NULL); ASSERT3P(zilog->zl_dmu_pool->dp_zil_clean_taskq, !=, NULL); taskqid_t id = taskq_dispatch(zilog->zl_dmu_pool->dp_zil_clean_taskq, (void (*)(void *))zil_itxg_clean, clean_me, TQ_NOSLEEP); if (id == TASKQID_INVALID) zil_itxg_clean(clean_me); } /* - * Get the list of itxs to commit into zl_itx_commit_list. + * This function will traverse the queue of itxs that need to be + * committed, and move them onto the ZIL's zl_itx_commit_list. */ static void zil_get_commit_list(zilog_t *zilog) { uint64_t otxg, txg; list_t *commit_list = &zilog->zl_itx_commit_list; + ASSERT(MUTEX_HELD(&zilog->zl_writer_lock)); + if (spa_freeze_txg(zilog->zl_spa) != UINT64_MAX) /* ziltest support */ otxg = ZILTEST_TXG; else otxg = spa_last_synced_txg(zilog->zl_spa) + 1; /* * This is inherently racy, since there is nothing to prevent * the last synced txg from changing. That's okay since we'll * only commit things in the future. */ for (txg = otxg; txg < (otxg + TXG_CONCURRENT_STATES); txg++) { itxg_t *itxg = &zilog->zl_itxg[txg & TXG_MASK]; mutex_enter(&itxg->itxg_lock); if (itxg->itxg_txg != txg) { mutex_exit(&itxg->itxg_lock); continue; } /* * If we're adding itx records to the zl_itx_commit_list, * then the zil better be dirty in this "txg". We can assert * that here since we're holding the itxg_lock which will * prevent spa_sync from cleaning it. Once we add the itxs * to the zl_itx_commit_list we must commit it to disk even * if it's unnecessary (i.e. the txg was synced). */ ASSERT(zilog_is_dirty_in_txg(zilog, txg) || spa_freeze_txg(zilog->zl_spa) != UINT64_MAX); list_move_tail(commit_list, &itxg->itxg_itxs->i_sync_list); mutex_exit(&itxg->itxg_lock); } } /* * Move the async itxs for a specified object to commit into sync lists. */ static void zil_async_to_sync(zilog_t *zilog, uint64_t foid) { uint64_t otxg, txg; itx_async_node_t *ian; avl_tree_t *t; avl_index_t where; if (spa_freeze_txg(zilog->zl_spa) != UINT64_MAX) /* ziltest support */ otxg = ZILTEST_TXG; else otxg = spa_last_synced_txg(zilog->zl_spa) + 1; /* * This is inherently racy, since there is nothing to prevent * the last synced txg from changing. */ for (txg = otxg; txg < (otxg + TXG_CONCURRENT_STATES); txg++) { itxg_t *itxg = &zilog->zl_itxg[txg & TXG_MASK]; mutex_enter(&itxg->itxg_lock); if (itxg->itxg_txg != txg) { mutex_exit(&itxg->itxg_lock); continue; } /* * If a foid is specified then find that node and append its * list. Otherwise walk the tree appending all the lists * to the sync list. We add to the end rather than the * beginning to ensure the create has happened. */ t = &itxg->itxg_itxs->i_async_tree; if (foid != 0) { ian = avl_find(t, &foid, &where); if (ian != NULL) { list_move_tail(&itxg->itxg_itxs->i_sync_list, &ian->ia_list); } } else { void *cookie = NULL; while ((ian = avl_destroy_nodes(t, &cookie)) != NULL) { list_move_tail(&itxg->itxg_itxs->i_sync_list, &ian->ia_list); list_destroy(&ian->ia_list); kmem_free(ian, sizeof (itx_async_node_t)); } } mutex_exit(&itxg->itxg_lock); } } +/* + * This function will prune commit itxs that are at the head of the + * commit list (it won't prune past the first non-commit itx), and + * either: a) attach them to the last lwb that's still pending + * completion, or b) skip them altogether. + * + * This is used as a performance optimization to prevent commit itxs + * from generating new lwbs when it's unnecessary to do so. + */ static void -zil_commit_writer(zilog_t *zilog) +zil_prune_commit_list(zilog_t *zilog) { - uint64_t txg; itx_t *itx; - lwb_t *lwb; - spa_t *spa = zilog->zl_spa; - int error = 0; - ASSERT(zilog->zl_root_zio == NULL); + ASSERT(MUTEX_HELD(&zilog->zl_writer_lock)); - mutex_exit(&zilog->zl_lock); + while ((itx = list_head(&zilog->zl_itx_commit_list)) != NULL) { + lr_t *lrc = &itx->itx_lr; + if (lrc->lrc_txtype != TX_COMMIT) + break; - zil_get_commit_list(zilog); + mutex_enter(&zilog->zl_lock); + + lwb_t *last_lwb = zilog->zl_last_lwb_opened; + if (last_lwb == NULL || last_lwb->lwb_state == LWB_STATE_DONE) { + /* + * All of the itxs this waiter was waiting on + * must have already completed (or there were + * never any itx's for it to wait on), so it's + * safe to skip this waiter and mark it done. + */ + zil_commit_waiter_skip(itx->itx_private); + } else { + zil_commit_waiter_link_lwb(itx->itx_private, last_lwb); + itx->itx_private = NULL; + } + + mutex_exit(&zilog->zl_lock); + + list_remove(&zilog->zl_itx_commit_list, itx); + zil_itx_destroy(itx); + } + + IMPLY(itx != NULL, itx->itx_lr.lrc_txtype != TX_COMMIT); +} + +static void +zil_commit_writer_stall(zilog_t *zilog) +{ + /* + * When zio_alloc_zil() fails to allocate the next lwb block on + * disk, we must call txg_wait_synced() to ensure all of the + * lwbs in the zilog's zl_lwb_list are synced and then freed (in + * zil_sync()), such that any subsequent ZIL writer (i.e. a call + * to zil_process_commit_list()) will have to call zil_create(), + * and start a new ZIL chain. + * + * Since zil_alloc_zil() failed, the lwb that was previously + * issued does not have a pointer to the "next" lwb on disk. + * Thus, if another ZIL writer thread was to allocate the "next" + * on-disk lwb, that block could be leaked in the event of a + * crash (because the previous lwb on-disk would not point to + * it). + * + * We must hold the zilog's zl_writer_lock while we do this, to + * ensure no new threads enter zil_process_commit_list() until + * all lwb's in the zl_lwb_list have been synced and freed + * (which is achieved via the txg_wait_synced() call). + */ + ASSERT(MUTEX_HELD(&zilog->zl_writer_lock)); + txg_wait_synced(zilog->zl_dmu_pool, 0); + ASSERT3P(list_tail(&zilog->zl_lwb_list), ==, NULL); +} + +/* + * This function will traverse the commit list, creating new lwbs as + * needed, and committing the itxs from the commit list to these newly + * created lwbs. Additionally, as a new lwb is created, the previous + * lwb will be issued to the zio layer to be written to disk. + */ +static void +zil_process_commit_list(zilog_t *zilog) +{ + spa_t *spa = zilog->zl_spa; + list_t nolwb_itxs; + list_t nolwb_waiters; + lwb_t *lwb; + itx_t *itx; + + ASSERT(MUTEX_HELD(&zilog->zl_writer_lock)); /* * Return if there's nothing to commit before we dirty the fs by * calling zil_create(). */ - if (list_head(&zilog->zl_itx_commit_list) == NULL) { - mutex_enter(&zilog->zl_lock); + if (list_head(&zilog->zl_itx_commit_list) == NULL) return; - } - if (zilog->zl_suspend) { - lwb = NULL; + list_create(&nolwb_itxs, sizeof (itx_t), offsetof(itx_t, itx_node)); + list_create(&nolwb_waiters, sizeof (zil_commit_waiter_t), + offsetof(zil_commit_waiter_t, zcw_node)); + + lwb = list_tail(&zilog->zl_lwb_list); + if (lwb == NULL) { + lwb = zil_create(zilog); } else { - lwb = list_tail(&zilog->zl_lwb_list); - if (lwb == NULL) - lwb = zil_create(zilog); + ASSERT3S(lwb->lwb_state, !=, LWB_STATE_ISSUED); + ASSERT3S(lwb->lwb_state, !=, LWB_STATE_DONE); } - DTRACE_PROBE1(zil__cw1, zilog_t *, zilog); - for (itx = list_head(&zilog->zl_itx_commit_list); itx != NULL; - itx = list_next(&zilog->zl_itx_commit_list, itx)) { - txg = itx->itx_lr.lrc_txg; + while ((itx = list_head(&zilog->zl_itx_commit_list)) != NULL) { + lr_t *lrc = &itx->itx_lr; + uint64_t txg = lrc->lrc_txg; + ASSERT3U(txg, !=, 0); + if (lrc->lrc_txtype == TX_COMMIT) { + DTRACE_PROBE2(zil__process__commit__itx, + zilog_t *, zilog, itx_t *, itx); + } else { + DTRACE_PROBE2(zil__process__normal__itx, + zilog_t *, zilog, itx_t *, itx); + } + + list_remove(&zilog->zl_itx_commit_list, itx); + /* * This is inherently racy and may result in us writing - * out a log block for a txg that was just synced. This is - * ok since we'll end cleaning up that log block the next - * time we call zil_sync(). + * out a log block for a txg that was just synced. This + * is ok since we'll end cleaning up that log block the + * next time we call zil_sync(). */ - if (txg > spa_last_synced_txg(spa) || txg > spa_freeze_txg(spa)) - lwb = zil_lwb_commit(zilog, itx, lwb); + boolean_t synced = txg <= spa_last_synced_txg(spa); + boolean_t frozen = txg > spa_freeze_txg(spa); + + if (!synced || frozen) { + if (lwb != NULL) { + lwb = zil_lwb_commit(zilog, itx, lwb); + + if (lwb == NULL) + list_insert_tail(&nolwb_itxs, itx); + else + list_insert_tail(&lwb->lwb_itxs, itx); + } else { + if (lrc->lrc_txtype == TX_COMMIT) { + zil_commit_waiter_link_nolwb( + itx->itx_private, &nolwb_waiters); + } + + list_insert_tail(&nolwb_itxs, itx); + } + } else { + /* + * If this is a commit itx, then there will be a + * thread that is either: already waiting for + * it, or soon will be waiting. + * + * This itx has already been committed to disk + * via spa_sync() so we don't bother committing + * it to an lwb. As a result, we cannot use the + * lwb zio callback to signal the waiter and + * mark it as done, so we must do that here. + */ + if (lrc->lrc_txtype == TX_COMMIT) + zil_commit_waiter_skip(itx->itx_private); + + zil_itx_destroy(itx); + } } - DTRACE_PROBE1(zil__cw2, zilog_t *, zilog); - /* write the last block out */ - if (lwb != NULL && lwb->lwb_zio != NULL) - lwb = zil_lwb_write_start(zilog, lwb); + if (lwb == NULL) { + /* + * This indicates zio_alloc_zil() failed to allocate the + * "next" lwb on-disk. When this happens, we must stall + * the ZIL write pipeline; see the comment within + * zil_commit_writer_stall() for more details. + */ + zil_commit_writer_stall(zilog); - zilog->zl_cur_used = 0; + /* + * Additionally, we have to signal and mark the "nolwb" + * waiters as "done" here, since without an lwb, we + * can't do this via zil_lwb_flush_vdevs_done() like + * normal. + */ + zil_commit_waiter_t *zcw; + while ((zcw = list_head(&nolwb_waiters)) != NULL) { + zil_commit_waiter_skip(zcw); + list_remove(&nolwb_waiters, zcw); + } + + /* + * And finally, we have to destroy the itx's that + * couldn't be committed to an lwb; this will also call + * the itx's callback if one exists for the itx. + */ + while ((itx = list_head(&nolwb_itxs)) != NULL) { + list_remove(&nolwb_itxs, itx); + zil_itx_destroy(itx); + } + } else { + ASSERT(list_is_empty(&nolwb_waiters)); + ASSERT3P(lwb, !=, NULL); + ASSERT3S(lwb->lwb_state, !=, LWB_STATE_ISSUED); + ASSERT3S(lwb->lwb_state, !=, LWB_STATE_DONE); + + /* + * At this point, the ZIL block pointed at by the "lwb" + * variable is in one of the following states: "closed" + * or "open". + * + * If its "closed", then no itxs have been committed to + * it, so there's no point in issueing its zio (i.e. + * it's "empty"). + * + * If its "open" state, then it contains one or more + * itxs that eventually need to be committed to stable + * storage. In this case we intentionally do not issue + * the lwb's zio to disk yet, and instead rely on one of + * the following two mechanisms for issuing the zio: + * + * 1. Ideally, there will be more ZIL activity occuring + * on the system, such that this function will be + * immeidately called again (not necessarily by the same + * thread) and this lwb's zio will be issued via + * zil_lwb_commit(). This way, the lwb is guaranteed to + * be "full" when it is issued to disk, and we'll make + * use of the lwb's size the best we can. + * + * 2. If there isn't sufficient ZIL activity occuring on + * the system, such that this lwb's zio isn't issued via + * zil_lwb_commit(), zil_commit_waiter() will issue the + * lwb's zio. If this occurs, the lwb is not guaranteed + * to be "full" by the time its zio is issued, and means + * the size of the lwb was "too large" given the amount + * of ZIL activity occuring on the system at that time. + * + * We do this for a couple of reasons: + * + * 1. To try and reduce the number of IOPs needed to + * write the same number of itxs. If an lwb has space + * available in it's buffer for more itxs, and more itxs + * will be committed relatively soon (relative to the + * latency of performing a write), then it's beneficial + * to wait for these "next" itxs. This way, more itxs + * can be committed to stable storage with fewer writes. + * + * 2. To try and use the largest lwb block size that the + * incoming rate of itxs can support. Again, this is to + * try and pack as many itxs into as few lwbs as + * possible, without significantly impacting the latency + * of each individual itx. + */ + } +} + +/* + * This function is responsible for ensuring the passed in commit waiter + * (and associated commit itx) is committed to an lwb. If the waiter is + * not already committed to an lwb, all itxs in the zilog's queue of + * itxs will be processed. The assumption is the passed in waiter's + * commit itx will found in the queue just like the other non-commit + * itxs, such that when the entire queue is processed, the waiter will + * have been commited to an lwb. + * + * The lwb associated with the passed in waiter is not guaranteed to + * have been issued by the time this function completes. If the lwb is + * not issued, we rely on future calls to zil_commit_writer() to issue + * the lwb, or the timeout mechanism found in zil_commit_waiter(). + */ +static void +zil_commit_writer(zilog_t *zilog, zil_commit_waiter_t *zcw) +{ + ASSERT(!MUTEX_HELD(&zilog->zl_lock)); + ASSERT(spa_writeable(zilog->zl_spa)); + ASSERT0(zilog->zl_suspend); + + mutex_enter(&zilog->zl_writer_lock); + + if (zcw->zcw_lwb != NULL || zcw->zcw_done) { + /* + * It's possible that, while we were waiting to acquire + * the "zl_writer_lock", another thread committed this + * waiter to an lwb. If that occurs, we bail out early, + * without processing any of the zilog's queue of itxs. + * + * On certain workloads and system configurations, the + * "zl_writer_lock" can become highly contended. In an + * attempt to reduce this contention, we immediately drop + * the lock if the waiter has already been processed. + * + * We've measured this optimization to reduce CPU spent + * contending on this lock by up to 5%, using a system + * with 32 CPUs, low latency storage (~50 usec writes), + * and 1024 threads performing sync writes. + */ + goto out; + } + + ZIL_STAT_BUMP(zil_commit_writer_count); + + zil_get_commit_list(zilog); + zil_prune_commit_list(zilog); + zil_process_commit_list(zilog); + +out: + mutex_exit(&zilog->zl_writer_lock); +} + +static void +zil_commit_waiter_timeout(zilog_t *zilog, zil_commit_waiter_t *zcw) +{ + ASSERT(!MUTEX_HELD(&zilog->zl_writer_lock)); + ASSERT(MUTEX_HELD(&zcw->zcw_lock)); + ASSERT3B(zcw->zcw_done, ==, B_FALSE); + + lwb_t *lwb = zcw->zcw_lwb; + ASSERT3P(lwb, !=, NULL); + ASSERT3S(lwb->lwb_state, !=, LWB_STATE_CLOSED); /* - * Wait if necessary for the log blocks to be on stable storage. + * If the lwb has already been issued by another thread, we can + * immediately return since there's no work to be done (the + * point of this function is to issue the lwb). Additionally, we + * do this prior to acquiring the zl_writer_lock, to avoid + * acquiring it when it's not necessary to do so. */ - if (zilog->zl_root_zio) { - error = zio_wait(zilog->zl_root_zio); - zilog->zl_root_zio = NULL; - zil_flush_vdevs(zilog); - } + if (lwb->lwb_state == LWB_STATE_ISSUED || + lwb->lwb_state == LWB_STATE_DONE) + return; - if (error || lwb == NULL) - txg_wait_synced(zilog->zl_dmu_pool, 0); + /* + * In order to call zil_lwb_write_issue() we must hold the + * zilog's "zl_writer_lock". We can't simply acquire that lock, + * since we're already holding the commit waiter's "zcw_lock", + * and those two locks are aquired in the opposite order + * elsewhere. + */ + mutex_exit(&zcw->zcw_lock); + mutex_enter(&zilog->zl_writer_lock); + mutex_enter(&zcw->zcw_lock); - while ((itx = list_head(&zilog->zl_itx_commit_list))) { - txg = itx->itx_lr.lrc_txg; - ASSERT(txg); + /* + * Since we just dropped and re-acquired the commit waiter's + * lock, we have to re-check to see if the waiter was marked + * "done" during that process. If the waiter was marked "done", + * the "lwb" pointer is no longer valid (it can be free'd after + * the waiter is marked "done"), so without this check we could + * wind up with a use-after-free error below. + */ + if (zcw->zcw_done) + goto out; - if (itx->itx_callback != NULL) - itx->itx_callback(itx->itx_callback_data); - list_remove(&zilog->zl_itx_commit_list, itx); - zil_itx_destroy(itx); + ASSERT3P(lwb, ==, zcw->zcw_lwb); + + /* + * We've already checked this above, but since we hadn't + * acquired the zilog's zl_writer_lock, we have to perform this + * check a second time while holding the lock. We can't call + * zil_lwb_write_issue() if the lwb had already been issued. + */ + if (lwb->lwb_state == LWB_STATE_ISSUED || + lwb->lwb_state == LWB_STATE_DONE) + goto out; + + ASSERT3S(lwb->lwb_state, ==, LWB_STATE_OPENED); + + /* + * As described in the comments above zil_commit_waiter() and + * zil_process_commit_list(), we need to issue this lwb's zio + * since we've reached the commit waiter's timeout and it still + * hasn't been issued. + */ + lwb_t *nlwb = zil_lwb_write_issue(zilog, lwb); + + ASSERT3S(lwb->lwb_state, !=, LWB_STATE_OPENED); + + /* + * Since the lwb's zio hadn't been issued by the time this thread + * reached its timeout, we reset the zilog's "zl_cur_used" field + * to influence the zil block size selection algorithm. + * + * By having to issue the lwb's zio here, it means the size of the + * lwb was too large, given the incoming throughput of itxs. By + * setting "zl_cur_used" to zero, we communicate this fact to the + * block size selection algorithm, so it can take this informaiton + * into account, and potentially select a smaller size for the + * next lwb block that is allocated. + */ + zilog->zl_cur_used = 0; + + if (nlwb == NULL) { + /* + * When zil_lwb_write_issue() returns NULL, this + * indicates zio_alloc_zil() failed to allocate the + * "next" lwb on-disk. When this occurs, the ZIL write + * pipeline must be stalled; see the comment within the + * zil_commit_writer_stall() function for more details. + * + * We must drop the commit waiter's lock prior to + * calling zil_commit_writer_stall() or else we can wind + * up with the following deadlock: + * + * - This thread is waiting for the txg to sync while + * holding the waiter's lock; txg_wait_synced() is + * used within txg_commit_writer_stall(). + * + * - The txg can't sync because it is waiting for this + * lwb's zio callback to call dmu_tx_commit(). + * + * - The lwb's zio callback can't call dmu_tx_commit() + * because it's blocked trying to acquire the waiter's + * lock, which occurs prior to calling dmu_tx_commit() + */ + mutex_exit(&zcw->zcw_lock); + zil_commit_writer_stall(zilog); + mutex_enter(&zcw->zcw_lock); } - mutex_enter(&zilog->zl_lock); +out: + mutex_exit(&zilog->zl_writer_lock); + ASSERT(MUTEX_HELD(&zcw->zcw_lock)); +} + +/* + * This function is responsible for performing the following two tasks: + * + * 1. its primary responsibility is to block until the given "commit + * waiter" is considered "done". + * + * 2. its secondary responsibility is to issue the zio for the lwb that + * the given "commit waiter" is waiting on, if this function has + * waited "long enough" and the lwb is still in the "open" state. + * + * Given a sufficient amount of itxs being generated and written using + * the ZIL, the lwb's zio will be issued via the zil_lwb_commit() + * function. If this does not occur, this secondary responsibility will + * ensure the lwb is issued even if there is not other synchronous + * activity on the system. + * + * For more details, see zil_process_commit_list(); more specifically, + * the comment at the bottom of that function. + */ +static void +zil_commit_waiter(zilog_t *zilog, zil_commit_waiter_t *zcw) +{ + ASSERT(!MUTEX_HELD(&zilog->zl_lock)); + ASSERT(!MUTEX_HELD(&zilog->zl_writer_lock)); + ASSERT(spa_writeable(zilog->zl_spa)); + ASSERT0(zilog->zl_suspend); + + mutex_enter(&zcw->zcw_lock); /* - * Remember the highest committed log sequence number for ztest. - * We only update this value when all the log writes succeeded, - * because ztest wants to ASSERT that it got the whole log chain. + * The timeout is scaled based on the lwb latency to avoid + * significantly impacting the latency of each individual itx. + * For more details, see the comment at the bottom of the + * zil_process_commit_list() function. */ - if (error == 0 && lwb != NULL) - zilog->zl_commit_lr_seq = zilog->zl_lr_seq; + int pct = MAX(zfs_commit_timeout_pct, 1); + hrtime_t sleep = (zilog->zl_last_lwb_latency * pct) / 100; + hrtime_t wakeup = gethrtime() + sleep; + boolean_t timedout = B_FALSE; + + while (!zcw->zcw_done) { + ASSERT(MUTEX_HELD(&zcw->zcw_lock)); + + lwb_t *lwb = zcw->zcw_lwb; + + /* + * Usually, the waiter will have a non-NULL lwb field here, + * but it's possible for it to be NULL as a result of + * zil_commit() racing with spa_sync(). + * + * When zil_clean() is called, it's possible for the itxg + * list (which may be cleaned via a taskq) to contain + * commit itxs. When this occurs, the commit waiters linked + * off of these commit itxs will not be committed to an + * lwb. Additionally, these commit waiters will not be + * marked done until zil_commit_waiter_skip() is called via + * zil_itxg_clean(). + * + * Thus, it's possible for this commit waiter (i.e. the + * "zcw" variable) to be found in this "in between" state; + * where it's "zcw_lwb" field is NULL, and it hasn't yet + * been skipped, so it's "zcw_done" field is still B_FALSE. + */ + IMPLY(lwb != NULL, lwb->lwb_state != LWB_STATE_CLOSED); + + if (lwb != NULL && lwb->lwb_state == LWB_STATE_OPENED) { + ASSERT3B(timedout, ==, B_FALSE); + + /* + * If the lwb hasn't been issued yet, then we + * need to wait with a timeout, in case this + * function needs to issue the lwb after the + * timeout is reached; responsibility (2) from + * the comment above this function. + */ + clock_t timeleft = cv_timedwait_hires(&zcw->zcw_cv, + &zcw->zcw_lock, wakeup, USEC2NSEC(1), + CALLOUT_FLAG_ABSOLUTE); + + if (timeleft >= 0 || zcw->zcw_done) + continue; + + timedout = B_TRUE; + zil_commit_waiter_timeout(zilog, zcw); + + if (!zcw->zcw_done) { + /* + * If the commit waiter has already been + * marked "done", it's possible for the + * waiter's lwb structure to have already + * been freed. Thus, we can only reliably + * make these assertions if the waiter + * isn't done. + */ + ASSERT3P(lwb, ==, zcw->zcw_lwb); + ASSERT3S(lwb->lwb_state, !=, LWB_STATE_OPENED); + } + } else { + /* + * If the lwb isn't open, then it must have already + * been issued. In that case, there's no need to + * use a timeout when waiting for the lwb to + * complete. + * + * Additionally, if the lwb is NULL, the waiter + * will soon be signalled and marked done via + * zil_clean() and zil_itxg_clean(), so no timeout + * is required. + */ + + IMPLY(lwb != NULL, + lwb->lwb_state == LWB_STATE_ISSUED || + lwb->lwb_state == LWB_STATE_DONE); + cv_wait(&zcw->zcw_cv, &zcw->zcw_lock); + } + } + + mutex_exit(&zcw->zcw_lock); +} + +static zil_commit_waiter_t * +zil_alloc_commit_waiter(void) +{ + zil_commit_waiter_t *zcw = kmem_cache_alloc(zil_zcw_cache, KM_SLEEP); + + cv_init(&zcw->zcw_cv, NULL, CV_DEFAULT, NULL); + mutex_init(&zcw->zcw_lock, NULL, MUTEX_DEFAULT, NULL); + list_link_init(&zcw->zcw_node); + zcw->zcw_lwb = NULL; + zcw->zcw_done = B_FALSE; + zcw->zcw_zio_error = 0; + + return (zcw); +} + +static void +zil_free_commit_waiter(zil_commit_waiter_t *zcw) +{ + ASSERT(!list_link_active(&zcw->zcw_node)); + ASSERT3P(zcw->zcw_lwb, ==, NULL); + ASSERT3B(zcw->zcw_done, ==, B_TRUE); + mutex_destroy(&zcw->zcw_lock); + cv_destroy(&zcw->zcw_cv); + kmem_cache_free(zil_zcw_cache, zcw); } /* - * Commit zfs transactions to stable storage. - * If foid is 0 push out all transactions, otherwise push only those - * for that object or might reference that object. + * This function is used to create a TX_COMMIT itx and assign it. This + * way, it will be linked into the ZIL's list of synchronous itxs, and + * then later committed to an lwb (or skipped) when + * zil_process_commit_list() is called. + */ +static void +zil_commit_itx_assign(zilog_t *zilog, zil_commit_waiter_t *zcw) +{ + dmu_tx_t *tx = dmu_tx_create(zilog->zl_os); + VERIFY0(dmu_tx_assign(tx, TXG_WAIT)); + + itx_t *itx = zil_itx_create(TX_COMMIT, sizeof (lr_t)); + itx->itx_sync = B_TRUE; + itx->itx_private = zcw; + + zil_itx_assign(zilog, itx, tx); + + dmu_tx_commit(tx); +} + +/* + * Commit ZFS Intent Log transactions (itxs) to stable storage. + * + * When writing ZIL transactions to the on-disk representation of the + * ZIL, the itxs are committed to a Log Write Block (lwb). Multiple + * itxs can be committed to a single lwb. Once a lwb is written and + * committed to stable storage (i.e. the lwb is written, and vdevs have + * been flushed), each itx that was committed to that lwb is also + * considered to be committed to stable storage. + * + * When an itx is committed to an lwb, the log record (lr_t) contained + * by the itx is copied into the lwb's zio buffer, and once this buffer + * is written to disk, it becomes an on-disk ZIL block. + * + * As itxs are generated, they're inserted into the ZIL's queue of + * uncommitted itxs. The semantics of zil_commit() are such that it will + * block until all itxs that were in the queue when it was called, are + * committed to stable storage. + * + * If "foid" is zero, this means all "synchronous" and "asynchronous" + * itxs, for all objects in the dataset, will be committed to stable + * storage prior to zil_commit() returning. If "foid" is non-zero, all + * "synchronous" itxs for all objects, but only "asynchronous" itxs + * that correspond to the foid passed in, will be committed to stable + * storage prior to zil_commit() returning. + * + * Generally speaking, when zil_commit() is called, the consumer doesn't + * actually care about _all_ of the uncommitted itxs. Instead, they're + * simply trying to waiting for a specific itx to be committed to disk, + * but the interface(s) for interacting with the ZIL don't allow such + * fine-grained communication. A better interface would allow a consumer + * to create and assign an itx, and then pass a reference to this itx to + * zil_commit(); such that zil_commit() would return as soon as that + * specific itx was committed to disk (instead of waiting for _all_ + * itxs to be committed). + * + * When a thread calls zil_commit() a special "commit itx" will be + * generated, along with a corresponding "waiter" for this commit itx. + * zil_commit() will wait on this waiter's CV, such that when the waiter + * is marked done, and signalled, zil_commit() will return. + * + * This commit itx is inserted into the queue of uncommitted itxs. This + * provides an easy mechanism for determining which itxs were in the + * queue prior to zil_commit() having been called, and which itxs were + * added after zil_commit() was called. + * + * The commit it is special; it doesn't have any on-disk representation. + * When a commit itx is "committed" to an lwb, the waiter associated + * with it is linked onto the lwb's list of waiters. Then, when that lwb + * completes, each waiter on the lwb's list is marked done and signalled + * -- allowing the thread waiting on the waiter to return from zil_commit(). + * + * It's important to point out a few critical factors that allow us + * to make use of the commit itxs, commit waiters, per-lwb lists of + * commit waiters, and zio completion callbacks like we're doing: * - * itxs are committed in batches. In a heavily stressed zil there will be - * a commit writer thread who is writing out a bunch of itxs to the log - * for a set of committing threads (cthreads) in the same batch as the writer. - * Those cthreads are all waiting on the same cv for that batch. + * 1. The list of waiters for each lwb is traversed, and each commit + * waiter is marked "done" and signalled, in the zio completion + * callback of the lwb's zio[*]. * - * There will also be a different and growing batch of threads that are - * waiting to commit (qthreads). When the committing batch completes - * a transition occurs such that the cthreads exit and the qthreads become - * cthreads. One of the new cthreads becomes the writer thread for the - * batch. Any new threads arriving become new qthreads. + * * Actually, the waiters are signalled in the zio completion + * callback of the root zio for the DKIOCFLUSHWRITECACHE commands + * that are sent to the vdevs upon completion of the lwb zio. * - * Only 2 condition variables are needed and there's no transition - * between the two cvs needed. They just flip-flop between qthreads - * and cthreads. + * 2. When the itxs are inserted into the ZIL's queue of uncommitted + * itxs, the order in which they are inserted is preserved[*]; as + * itxs are added to the queue, they are added to the tail of + * in-memory linked lists. * - * Using this scheme we can efficiently wakeup up only those threads - * that have been committed. + * When committing the itxs to lwbs (to be written to disk), they + * are committed in the same order in which the itxs were added to + * the uncommitted queue's linked list(s); i.e. the linked list of + * itxs to commit is traversed from head to tail, and each itx is + * committed to an lwb in that order. + * + * * To clarify: + * + * - the order of "sync" itxs is preserved w.r.t. other + * "sync" itxs, regardless of the corresponding objects. + * - the order of "async" itxs is preserved w.r.t. other + * "async" itxs corresponding to the same object. + * - the order of "async" itxs is *not* preserved w.r.t. other + * "async" itxs corresponding to different objects. + * - the order of "sync" itxs w.r.t. "async" itxs (or vice + * versa) is *not* preserved, even for itxs that correspond + * to the same object. + * + * For more details, see: zil_itx_assign(), zil_async_to_sync(), + * zil_get_commit_list(), and zil_process_commit_list(). + * + * 3. The lwbs represent a linked list of blocks on disk. Thus, any + * lwb cannot be considered committed to stable storage, until its + * "previous" lwb is also committed to stable storage. This fact, + * coupled with the fact described above, means that itxs are + * committed in (roughly) the order in which they were generated. + * This is essential because itxs are dependent on prior itxs. + * Thus, we *must not* deem an itx as being committed to stable + * storage, until *all* prior itxs have also been committed to + * stable storage. + * + * To enforce this ordering of lwb zio's, while still leveraging as + * much of the underlying storage performance as possible, we rely + * on two fundamental concepts: + * + * 1. The creation and issuance of lwb zio's is protected by + * the zilog's "zl_writer_lock", which ensures only a single + * thread is creating and/or issuing lwb's at a time + * 2. The "previous" lwb is a child of the "current" lwb + * (leveraging the zio parent-child depenency graph) + * + * By relying on this parent-child zio relationship, we can have + * many lwb zio's concurrently issued to the underlying storage, + * but the order in which they complete will be the same order in + * which they were created. */ void zil_commit(zilog_t *zilog, uint64_t foid) { - uint64_t mybatch; + /* + * We should never attempt to call zil_commit on a snapshot for + * a couple of reasons: + * + * 1. A snapshot may never be modified, thus it cannot have any + * in-flight itxs that would have modified the dataset. + * + * 2. By design, when zil_commit() is called, a commit itx will + * be assigned to this zilog; as a result, the zilog will be + * dirtied. We must not dirty the zilog of a snapshot; there's + * checks in the code that enforce this invariant, and will + * cause a panic if it's not upheld. + */ + ASSERT3B(dmu_objset_is_snapshot(zilog->zl_os), ==, B_FALSE); if (zilog->zl_sync == ZFS_SYNC_DISABLED) return; + if (!spa_writeable(zilog->zl_spa)) { + /* + * If the SPA is not writable, there should never be any + * pending itxs waiting to be committed to disk. If that + * weren't true, we'd skip writing those itxs out, and + * would break the sematics of zil_commit(); thus, we're + * verifying that truth before we return to the caller. + */ + ASSERT(list_is_empty(&zilog->zl_lwb_list)); + ASSERT3P(zilog->zl_last_lwb_opened, ==, NULL); + for (int i = 0; i < TXG_SIZE; i++) + ASSERT3P(zilog->zl_itxg[i].itxg_itxs, ==, NULL); + return; + } + + /* + * If the ZIL is suspended, we don't want to dirty it by calling + * zil_commit_itx_assign() below, nor can we write out + * lwbs like would be done in zil_commit_write(). Thus, we + * simply rely on txg_wait_synced() to maintain the necessary + * semantics, and avoid calling those functions altogether. + */ + if (zilog->zl_suspend > 0) { + txg_wait_synced(zilog->zl_dmu_pool, 0); + return; + } + ZIL_STAT_BUMP(zil_commit_count); - /* move the async itxs for the foid to the sync queues */ + /* + * Move the "async" itxs for the specified foid to the "sync" + * queues, such that they will be later committed (or skipped) + * to an lwb when zil_process_commit_list() is called. + * + * Since these "async" itxs must be committed prior to this + * call to zil_commit returning, we must perform this operation + * before we call zil_commit_itx_assign(). + */ zil_async_to_sync(zilog, foid); - mutex_enter(&zilog->zl_lock); - mybatch = zilog->zl_next_batch; - while (zilog->zl_writer) { - cv_wait(&zilog->zl_cv_batch[mybatch & 1], &zilog->zl_lock); - if (mybatch <= zilog->zl_com_batch) { - mutex_exit(&zilog->zl_lock); - return; - } - } - - zilog->zl_next_batch++; - zilog->zl_writer = B_TRUE; - ZIL_STAT_BUMP(zil_commit_writer_count); - zil_commit_writer(zilog); - zilog->zl_com_batch = mybatch; - zilog->zl_writer = B_FALSE; + /* + * We allocate a new "waiter" structure which will initially be + * linked to the commit itx using the itx's "itx_private" field. + * Since the commit itx doesn't represent any on-disk state, + * when it's committed to an lwb, rather than copying the its + * lr_t into the lwb's buffer, the commit itx's "waiter" will be + * added to the lwb's list of waiters. Then, when the lwb is + * committed to stable storage, each waiter in the lwb's list of + * waiters will be marked "done", and signalled. + * + * We must create the waiter and assign the commit itx prior to + * calling zil_commit_writer(), or else our specific commit itx + * is not guaranteed to be committed to an lwb prior to calling + * zil_commit_waiter(). + */ + zil_commit_waiter_t *zcw = zil_alloc_commit_waiter(); + zil_commit_itx_assign(zilog, zcw); - /* wake up one thread to become the next writer */ - cv_signal(&zilog->zl_cv_batch[(mybatch+1) & 1]); + zil_commit_writer(zilog, zcw); + zil_commit_waiter(zilog, zcw); - /* wake up all threads waiting for this batch to be committed */ - cv_broadcast(&zilog->zl_cv_batch[mybatch & 1]); + if (zcw->zcw_zio_error != 0) { + /* + * If there was an error writing out the ZIL blocks that + * this thread is waiting on, then we fallback to + * relying on spa_sync() to write out the data this + * thread is waiting on. Obviously this has performance + * implications, but the expectation is for this to be + * an exceptional case, and shouldn't occur often. + */ + DTRACE_PROBE2(zil__commit__io__error, + zilog_t *, zilog, zil_commit_waiter_t *, zcw); + txg_wait_synced(zilog->zl_dmu_pool, 0); + } - mutex_exit(&zilog->zl_lock); + zil_free_commit_waiter(zcw); } /* * Called in syncing context to free committed log blocks and update log header. */ void zil_sync(zilog_t *zilog, dmu_tx_t *tx) { zil_header_t *zh = zil_header_in_syncing_context(zilog); uint64_t txg = dmu_tx_get_txg(tx); spa_t *spa = zilog->zl_spa; uint64_t *replayed_seq = &zilog->zl_replayed_seq[txg & TXG_MASK]; lwb_t *lwb; /* * We don't zero out zl_destroy_txg, so make sure we don't try * to destroy it twice. */ if (spa_sync_pass(spa) != 1) return; mutex_enter(&zilog->zl_lock); ASSERT(zilog->zl_stop_sync == 0); if (*replayed_seq != 0) { ASSERT(zh->zh_replay_seq < *replayed_seq); zh->zh_replay_seq = *replayed_seq; *replayed_seq = 0; } if (zilog->zl_destroy_txg == txg) { blkptr_t blk = zh->zh_log; ASSERT(list_head(&zilog->zl_lwb_list) == NULL); bzero(zh, sizeof (zil_header_t)); bzero(zilog->zl_replayed_seq, sizeof (zilog->zl_replayed_seq)); if (zilog->zl_keep_first) { /* * If this block was part of log chain that couldn't * be claimed because a device was missing during * zil_claim(), but that device later returns, * then this block could erroneously appear valid. * To guard against this, assign a new GUID to the new * log chain so it doesn't matter what blk points to. */ zil_init_log_chain(zilog, &blk); zh->zh_log = blk; } } while ((lwb = list_head(&zilog->zl_lwb_list)) != NULL) { zh->zh_log = lwb->lwb_blk; if (lwb->lwb_buf != NULL || lwb->lwb_max_txg > txg) break; - - ASSERT(lwb->lwb_zio == NULL); - list_remove(&zilog->zl_lwb_list, lwb); - zio_free_zil(spa, txg, &lwb->lwb_blk); - kmem_cache_free(zil_lwb_cache, lwb); + zio_free(spa, txg, &lwb->lwb_blk); + zil_free_lwb(zilog, lwb); /* * If we don't have anything left in the lwb list then * we've had an allocation failure and we need to zero * out the zil_header blkptr so that we don't end * up freeing the same block twice. */ if (list_head(&zilog->zl_lwb_list) == NULL) BP_ZERO(&zh->zh_log); } /* * Remove fastwrite on any blocks that have been pre-allocated for * the next commit. This prevents fastwrite counter pollution by * unused, long-lived LWBs. */ for (; lwb != NULL; lwb = list_next(&zilog->zl_lwb_list, lwb)) { - if (lwb->lwb_fastwrite && !lwb->lwb_zio) { + if (lwb->lwb_fastwrite && !lwb->lwb_write_zio) { metaslab_fastwrite_unmark(zilog->zl_spa, &lwb->lwb_blk); lwb->lwb_fastwrite = 0; } } mutex_exit(&zilog->zl_lock); } +/* ARGSUSED */ +static int +zil_lwb_cons(void *vbuf, void *unused, int kmflag) +{ + lwb_t *lwb = vbuf; + list_create(&lwb->lwb_itxs, sizeof (itx_t), offsetof(itx_t, itx_node)); + list_create(&lwb->lwb_waiters, sizeof (zil_commit_waiter_t), + offsetof(zil_commit_waiter_t, zcw_node)); + avl_create(&lwb->lwb_vdev_tree, zil_lwb_vdev_compare, + sizeof (zil_vdev_node_t), offsetof(zil_vdev_node_t, zv_node)); + mutex_init(&lwb->lwb_vdev_lock, NULL, MUTEX_DEFAULT, NULL); + return (0); +} + +/* ARGSUSED */ +static void +zil_lwb_dest(void *vbuf, void *unused) +{ + lwb_t *lwb = vbuf; + mutex_destroy(&lwb->lwb_vdev_lock); + avl_destroy(&lwb->lwb_vdev_tree); + list_destroy(&lwb->lwb_waiters); + list_destroy(&lwb->lwb_itxs); +} + void zil_init(void) { zil_lwb_cache = kmem_cache_create("zil_lwb_cache", - sizeof (struct lwb), 0, NULL, NULL, NULL, NULL, NULL, 0); + sizeof (lwb_t), 0, zil_lwb_cons, zil_lwb_dest, NULL, NULL, NULL, 0); + + zil_zcw_cache = kmem_cache_create("zil_zcw_cache", + sizeof (zil_commit_waiter_t), 0, NULL, NULL, NULL, NULL, NULL, 0); zil_ksp = kstat_create("zfs", 0, "zil", "misc", KSTAT_TYPE_NAMED, sizeof (zil_stats) / sizeof (kstat_named_t), KSTAT_FLAG_VIRTUAL); if (zil_ksp != NULL) { zil_ksp->ks_data = &zil_stats; kstat_install(zil_ksp); } } void zil_fini(void) { + kmem_cache_destroy(zil_zcw_cache); kmem_cache_destroy(zil_lwb_cache); if (zil_ksp != NULL) { kstat_delete(zil_ksp); zil_ksp = NULL; } } void zil_set_sync(zilog_t *zilog, uint64_t sync) { zilog->zl_sync = sync; } void zil_set_logbias(zilog_t *zilog, uint64_t logbias) { zilog->zl_logbias = logbias; } zilog_t * zil_alloc(objset_t *os, zil_header_t *zh_phys) { zilog_t *zilog; zilog = kmem_zalloc(sizeof (zilog_t), KM_SLEEP); zilog->zl_header = zh_phys; zilog->zl_os = os; zilog->zl_spa = dmu_objset_spa(os); zilog->zl_dmu_pool = dmu_objset_pool(os); zilog->zl_destroy_txg = TXG_INITIAL - 1; zilog->zl_logbias = dmu_objset_logbias(os); zilog->zl_sync = dmu_objset_syncprop(os); - zilog->zl_next_batch = 1; + zilog->zl_dirty_max_txg = 0; + zilog->zl_last_lwb_opened = NULL; + zilog->zl_last_lwb_latency = 0; mutex_init(&zilog->zl_lock, NULL, MUTEX_DEFAULT, NULL); + mutex_init(&zilog->zl_writer_lock, NULL, MUTEX_DEFAULT, NULL); for (int i = 0; i < TXG_SIZE; i++) { mutex_init(&zilog->zl_itxg[i].itxg_lock, NULL, MUTEX_DEFAULT, NULL); } list_create(&zilog->zl_lwb_list, sizeof (lwb_t), offsetof(lwb_t, lwb_node)); list_create(&zilog->zl_itx_commit_list, sizeof (itx_t), offsetof(itx_t, itx_node)); - mutex_init(&zilog->zl_vdev_lock, NULL, MUTEX_DEFAULT, NULL); - - avl_create(&zilog->zl_vdev_tree, zil_vdev_compare, - sizeof (zil_vdev_node_t), offsetof(zil_vdev_node_t, zv_node)); - - cv_init(&zilog->zl_cv_writer, NULL, CV_DEFAULT, NULL); cv_init(&zilog->zl_cv_suspend, NULL, CV_DEFAULT, NULL); - cv_init(&zilog->zl_cv_batch[0], NULL, CV_DEFAULT, NULL); - cv_init(&zilog->zl_cv_batch[1], NULL, CV_DEFAULT, NULL); return (zilog); } void zil_free(zilog_t *zilog) { int i; zilog->zl_stop_sync = 1; ASSERT0(zilog->zl_suspend); ASSERT0(zilog->zl_suspending); ASSERT(list_is_empty(&zilog->zl_lwb_list)); list_destroy(&zilog->zl_lwb_list); - avl_destroy(&zilog->zl_vdev_tree); - mutex_destroy(&zilog->zl_vdev_lock); - ASSERT(list_is_empty(&zilog->zl_itx_commit_list)); list_destroy(&zilog->zl_itx_commit_list); for (i = 0; i < TXG_SIZE; i++) { /* * It's possible for an itx to be generated that doesn't dirty * a txg (e.g. ztest TX_TRUNCATE). So there's no zil_clean() * callback to remove the entry. We remove those here. * * Also free up the ziltest itxs. */ if (zilog->zl_itxg[i].itxg_itxs) zil_itxg_clean(zilog->zl_itxg[i].itxg_itxs); mutex_destroy(&zilog->zl_itxg[i].itxg_lock); } + mutex_destroy(&zilog->zl_writer_lock); mutex_destroy(&zilog->zl_lock); - cv_destroy(&zilog->zl_cv_writer); cv_destroy(&zilog->zl_cv_suspend); - cv_destroy(&zilog->zl_cv_batch[0]); - cv_destroy(&zilog->zl_cv_batch[1]); kmem_free(zilog, sizeof (zilog_t)); } /* * Open an intent log. */ zilog_t * zil_open(objset_t *os, zil_get_data_t *get_data) { zilog_t *zilog = dmu_objset_zil(os); - ASSERT(zilog->zl_get_data == NULL); + ASSERT3P(zilog->zl_get_data, ==, NULL); + ASSERT3P(zilog->zl_last_lwb_opened, ==, NULL); ASSERT(list_is_empty(&zilog->zl_lwb_list)); zilog->zl_get_data = get_data; return (zilog); } /* * Close an intent log. */ void zil_close(zilog_t *zilog) { lwb_t *lwb; - uint64_t txg = 0; + uint64_t txg; - zil_commit(zilog, 0); /* commit all itx */ + if (!dmu_objset_is_snapshot(zilog->zl_os)) { + zil_commit(zilog, 0); + } else { + ASSERT3P(list_tail(&zilog->zl_lwb_list), ==, NULL); + ASSERT0(zilog->zl_dirty_max_txg); + ASSERT3B(zilog_is_dirty(zilog), ==, B_FALSE); + } - /* - * The lwb_max_txg for the stubby lwb will reflect the last activity - * for the zil. After a txg_wait_synced() on the txg we know all the - * callbacks have occurred that may clean the zil. Only then can we - * destroy the zl_clean_taskq. - */ mutex_enter(&zilog->zl_lock); lwb = list_tail(&zilog->zl_lwb_list); - if (lwb != NULL) - txg = lwb->lwb_max_txg; + if (lwb == NULL) + txg = zilog->zl_dirty_max_txg; + else + txg = MAX(zilog->zl_dirty_max_txg, lwb->lwb_max_txg); mutex_exit(&zilog->zl_lock); - if (txg) + + /* + * We need to use txg_wait_synced() to wait long enough for the + * ZIL to be clean, and to wait for all pending lwbs to be + * written out. + */ + if (txg != 0) txg_wait_synced(zilog->zl_dmu_pool, txg); if (zilog_is_dirty(zilog)) zfs_dbgmsg("zil (%p) is dirty, txg %llu", zilog, txg); if (txg < spa_freeze_txg(zilog->zl_spa)) VERIFY(!zilog_is_dirty(zilog)); zilog->zl_get_data = NULL; /* - * We should have only one LWB left on the list; remove it now. + * We should have only one lwb left on the list; remove it now. */ mutex_enter(&zilog->zl_lock); lwb = list_head(&zilog->zl_lwb_list); if (lwb != NULL) { - ASSERT(lwb == list_tail(&zilog->zl_lwb_list)); - ASSERT(lwb->lwb_zio == NULL); + ASSERT3P(lwb, ==, list_tail(&zilog->zl_lwb_list)); + ASSERT3S(lwb->lwb_state, !=, LWB_STATE_ISSUED); + if (lwb->lwb_fastwrite) metaslab_fastwrite_unmark(zilog->zl_spa, &lwb->lwb_blk); + list_remove(&zilog->zl_lwb_list, lwb); zio_buf_free(lwb->lwb_buf, lwb->lwb_sz); - kmem_cache_free(zil_lwb_cache, lwb); + zil_free_lwb(zilog, lwb); } mutex_exit(&zilog->zl_lock); } static char *suspend_tag = "zil suspending"; /* * Suspend an intent log. While in suspended mode, we still honor * synchronous semantics, but we rely on txg_wait_synced() to do it. * On old version pools, we suspend the log briefly when taking a * snapshot so that it will have an empty intent log. * * Long holds are not really intended to be used the way we do here -- * held for such a short time. A concurrent caller of dsl_dataset_long_held() * could fail. Therefore we take pains to only put a long hold if it is * actually necessary. Fortunately, it will only be necessary if the * objset is currently mounted (or the ZVOL equivalent). In that case it * will already have a long hold, so we are not really making things any worse. * * Ideally, we would locate the existing long-holder (i.e. the zfsvfs_t or * zvol_state_t), and use their mechanism to prevent their hold from being * dropped (e.g. VFS_HOLD()). However, that would be even more pain for * very little gain. * * if cookiep == NULL, this does both the suspend & resume. * Otherwise, it returns with the dataset "long held", and the cookie * should be passed into zil_resume(). */ int zil_suspend(const char *osname, void **cookiep) { objset_t *os; zilog_t *zilog; const zil_header_t *zh; int error; error = dmu_objset_hold(osname, suspend_tag, &os); if (error != 0) return (error); zilog = dmu_objset_zil(os); mutex_enter(&zilog->zl_lock); zh = zilog->zl_header; if (zh->zh_flags & ZIL_REPLAY_NEEDED) { /* unplayed log */ mutex_exit(&zilog->zl_lock); dmu_objset_rele(os, suspend_tag); return (SET_ERROR(EBUSY)); } /* * Don't put a long hold in the cases where we can avoid it. This * is when there is no cookie so we are doing a suspend & resume * (i.e. called from zil_vdev_offline()), and there's nothing to do * for the suspend because it's already suspended, or there's no ZIL. */ if (cookiep == NULL && !zilog->zl_suspending && (zilog->zl_suspend > 0 || BP_IS_HOLE(&zh->zh_log))) { mutex_exit(&zilog->zl_lock); dmu_objset_rele(os, suspend_tag); return (0); } dsl_dataset_long_hold(dmu_objset_ds(os), suspend_tag); dsl_pool_rele(dmu_objset_pool(os), suspend_tag); zilog->zl_suspend++; if (zilog->zl_suspend > 1) { /* * Someone else is already suspending it. * Just wait for them to finish. */ while (zilog->zl_suspending) cv_wait(&zilog->zl_cv_suspend, &zilog->zl_lock); mutex_exit(&zilog->zl_lock); if (cookiep == NULL) zil_resume(os); else *cookiep = os; return (0); } /* * If there is no pointer to an on-disk block, this ZIL must not * be active (e.g. filesystem not mounted), so there's nothing * to clean up. */ if (BP_IS_HOLE(&zh->zh_log)) { ASSERT(cookiep != NULL); /* fast path already handled */ *cookiep = os; mutex_exit(&zilog->zl_lock); return (0); } /* * The ZIL has work to do. Ensure that the associated encryption * key will remain mapped while we are committing the log by * grabbing a reference to it. If the key isn't loaded we have no * choice but to return an error until the wrapping key is loaded. */ if (os->os_encrypted && spa_keystore_create_mapping(os->os_spa, dmu_objset_ds(os), FTAG) != 0) { zilog->zl_suspend--; mutex_exit(&zilog->zl_lock); dsl_dataset_long_rele(dmu_objset_ds(os), suspend_tag); dsl_dataset_rele(dmu_objset_ds(os), suspend_tag); return (SET_ERROR(EBUSY)); } zilog->zl_suspending = B_TRUE; mutex_exit(&zilog->zl_lock); zil_commit(zilog, 0); zil_destroy(zilog, B_FALSE); mutex_enter(&zilog->zl_lock); zilog->zl_suspending = B_FALSE; cv_broadcast(&zilog->zl_cv_suspend); mutex_exit(&zilog->zl_lock); if (os->os_encrypted) { /* * Encrypted datasets need to wait for all data to be * synced out before removing the mapping. * * XXX: Depending on the number of datasets with * outstanding ZIL data on a given log device, this * might cause spa_offline_log() to take a long time. */ txg_wait_synced(zilog->zl_dmu_pool, zilog->zl_destroy_txg); VERIFY0(spa_keystore_remove_mapping(os->os_spa, dmu_objset_id(os), FTAG)); } if (cookiep == NULL) zil_resume(os); else *cookiep = os; return (0); } void zil_resume(void *cookie) { objset_t *os = cookie; zilog_t *zilog = dmu_objset_zil(os); mutex_enter(&zilog->zl_lock); ASSERT(zilog->zl_suspend != 0); zilog->zl_suspend--; mutex_exit(&zilog->zl_lock); dsl_dataset_long_rele(dmu_objset_ds(os), suspend_tag); dsl_dataset_rele(dmu_objset_ds(os), suspend_tag); } typedef struct zil_replay_arg { zil_replay_func_t **zr_replay; void *zr_arg; boolean_t zr_byteswap; char *zr_lr; } zil_replay_arg_t; static int zil_replay_error(zilog_t *zilog, lr_t *lr, int error) { char name[ZFS_MAX_DATASET_NAME_LEN]; zilog->zl_replaying_seq--; /* didn't actually replay this one */ dmu_objset_name(zilog->zl_os, name); cmn_err(CE_WARN, "ZFS replay transaction error %d, " "dataset %s, seq 0x%llx, txtype %llu %s\n", error, name, (u_longlong_t)lr->lrc_seq, (u_longlong_t)(lr->lrc_txtype & ~TX_CI), (lr->lrc_txtype & TX_CI) ? "CI" : ""); return (error); } static int zil_replay_log_record(zilog_t *zilog, lr_t *lr, void *zra, uint64_t claim_txg) { zil_replay_arg_t *zr = zra; const zil_header_t *zh = zilog->zl_header; uint64_t reclen = lr->lrc_reclen; uint64_t txtype = lr->lrc_txtype; int error = 0; zilog->zl_replaying_seq = lr->lrc_seq; if (lr->lrc_seq <= zh->zh_replay_seq) /* already replayed */ return (0); if (lr->lrc_txg < claim_txg) /* already committed */ return (0); /* Strip case-insensitive bit, still present in log record */ txtype &= ~TX_CI; if (txtype == 0 || txtype >= TX_MAX_TYPE) return (zil_replay_error(zilog, lr, EINVAL)); /* * If this record type can be logged out of order, the object * (lr_foid) may no longer exist. That's legitimate, not an error. */ if (TX_OOO(txtype)) { error = dmu_object_info(zilog->zl_os, LR_FOID_GET_OBJ(((lr_ooo_t *)lr)->lr_foid), NULL); if (error == ENOENT || error == EEXIST) return (0); } /* * Make a copy of the data so we can revise and extend it. */ bcopy(lr, zr->zr_lr, reclen); /* * If this is a TX_WRITE with a blkptr, suck in the data. */ if (txtype == TX_WRITE && reclen == sizeof (lr_write_t)) { error = zil_read_log_data(zilog, (lr_write_t *)lr, zr->zr_lr + reclen); if (error != 0) return (zil_replay_error(zilog, lr, error)); } /* * The log block containing this lr may have been byteswapped * so that we can easily examine common fields like lrc_txtype. * However, the log is a mix of different record types, and only the * replay vectors know how to byteswap their records. Therefore, if * the lr was byteswapped, undo it before invoking the replay vector. */ if (zr->zr_byteswap) byteswap_uint64_array(zr->zr_lr, reclen); /* * We must now do two things atomically: replay this log record, * and update the log header sequence number to reflect the fact that * we did so. At the end of each replay function the sequence number * is updated if we are in replay mode. */ error = zr->zr_replay[txtype](zr->zr_arg, zr->zr_lr, zr->zr_byteswap); if (error != 0) { /* * The DMU's dnode layer doesn't see removes until the txg * commits, so a subsequent claim can spuriously fail with * EEXIST. So if we receive any error we try syncing out * any removes then retry the transaction. Note that we * specify B_FALSE for byteswap now, so we don't do it twice. */ txg_wait_synced(spa_get_dsl(zilog->zl_spa), 0); error = zr->zr_replay[txtype](zr->zr_arg, zr->zr_lr, B_FALSE); if (error != 0) return (zil_replay_error(zilog, lr, error)); } return (0); } /* ARGSUSED */ static int zil_incr_blks(zilog_t *zilog, blkptr_t *bp, void *arg, uint64_t claim_txg) { zilog->zl_replay_blks++; return (0); } /* * If this dataset has a non-empty intent log, replay it and destroy it. */ void zil_replay(objset_t *os, void *arg, zil_replay_func_t *replay_func[TX_MAX_TYPE]) { zilog_t *zilog = dmu_objset_zil(os); const zil_header_t *zh = zilog->zl_header; zil_replay_arg_t zr; if ((zh->zh_flags & ZIL_REPLAY_NEEDED) == 0) { zil_destroy(zilog, B_TRUE); return; } zr.zr_replay = replay_func; zr.zr_arg = arg; zr.zr_byteswap = BP_SHOULD_BYTESWAP(&zh->zh_log); zr.zr_lr = vmem_alloc(2 * SPA_MAXBLOCKSIZE, KM_SLEEP); /* * Wait for in-progress removes to sync before starting replay. */ txg_wait_synced(zilog->zl_dmu_pool, 0); zilog->zl_replay = B_TRUE; zilog->zl_replay_time = ddi_get_lbolt(); ASSERT(zilog->zl_replay_blks == 0); (void) zil_parse(zilog, zil_incr_blks, zil_replay_log_record, &zr, zh->zh_claim_txg, B_TRUE); vmem_free(zr.zr_lr, 2 * SPA_MAXBLOCKSIZE); zil_destroy(zilog, B_FALSE); txg_wait_synced(zilog->zl_dmu_pool, zilog->zl_destroy_txg); zilog->zl_replay = B_FALSE; } boolean_t zil_replaying(zilog_t *zilog, dmu_tx_t *tx) { if (zilog->zl_sync == ZFS_SYNC_DISABLED) return (B_TRUE); if (zilog->zl_replay) { dsl_dataset_dirty(dmu_objset_ds(zilog->zl_os), tx); zilog->zl_replayed_seq[dmu_tx_get_txg(tx) & TXG_MASK] = zilog->zl_replaying_seq; return (B_TRUE); } return (B_FALSE); } /* ARGSUSED */ int zil_vdev_offline(const char *osname, void *arg) { int error; error = zil_suspend(osname, NULL); if (error != 0) return (SET_ERROR(EEXIST)); return (0); } #if defined(_KERNEL) && defined(HAVE_SPL) EXPORT_SYMBOL(zil_alloc); EXPORT_SYMBOL(zil_free); EXPORT_SYMBOL(zil_open); EXPORT_SYMBOL(zil_close); EXPORT_SYMBOL(zil_replay); EXPORT_SYMBOL(zil_replaying); EXPORT_SYMBOL(zil_destroy); EXPORT_SYMBOL(zil_destroy_sync); EXPORT_SYMBOL(zil_itx_create); EXPORT_SYMBOL(zil_itx_destroy); EXPORT_SYMBOL(zil_itx_assign); EXPORT_SYMBOL(zil_commit); EXPORT_SYMBOL(zil_vdev_offline); EXPORT_SYMBOL(zil_claim); EXPORT_SYMBOL(zil_check_log_chain); EXPORT_SYMBOL(zil_sync); EXPORT_SYMBOL(zil_clean); EXPORT_SYMBOL(zil_suspend); EXPORT_SYMBOL(zil_resume); -EXPORT_SYMBOL(zil_add_block); +EXPORT_SYMBOL(zil_lwb_add_block); EXPORT_SYMBOL(zil_bp_tree_add); EXPORT_SYMBOL(zil_set_sync); EXPORT_SYMBOL(zil_set_logbias); /* BEGIN CSTYLED */ module_param(zil_replay_disable, int, 0644); MODULE_PARM_DESC(zil_replay_disable, "Disable intent logging replay"); module_param(zfs_nocacheflush, int, 0644); MODULE_PARM_DESC(zfs_nocacheflush, "Disable cache flushes"); module_param(zil_slog_bulk, ulong, 0644); MODULE_PARM_DESC(zil_slog_bulk, "Limit in bytes slog sync writes per commit"); /* END CSTYLED */ #endif diff --git a/module/zfs/zio.c b/module/zfs/zio.c index 311f79e23b93..92e5a8dd8303 100644 --- a/module/zfs/zio.c +++ b/module/zfs/zio.c @@ -1,4488 +1,4505 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2017 by Delphix. All rights reserved. * Copyright (c) 2011 Nexenta Systems, Inc. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * ========================================================================== * I/O type descriptions * ========================================================================== */ const char *zio_type_name[ZIO_TYPES] = { /* * Note: Linux kernel thread name length is limited * so these names will differ from upstream open zfs. */ "z_null", "z_rd", "z_wr", "z_fr", "z_cl", "z_ioctl" }; int zio_dva_throttle_enabled = B_TRUE; /* * ========================================================================== * I/O kmem caches * ========================================================================== */ kmem_cache_t *zio_cache; kmem_cache_t *zio_link_cache; kmem_cache_t *zio_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; kmem_cache_t *zio_data_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; #if defined(ZFS_DEBUG) && !defined(_KERNEL) uint64_t zio_buf_cache_allocs[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; uint64_t zio_buf_cache_frees[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; #endif int zio_delay_max = ZIO_DELAY_MAX; #define ZIO_PIPELINE_CONTINUE 0x100 #define ZIO_PIPELINE_STOP 0x101 #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. */ int zfs_sync_pass_deferred_free = 2; /* defer frees starting in this pass */ int zfs_sync_pass_dont_compress = 5; /* don't compress starting in this pass */ int zfs_sync_pass_rewrite = 2; /* rewrite new bps starting in this pass */ /* * An allocating zio is one that either currently has the DVA allocate * stage set or will have it later in its lifetime. */ #define IO_IS_ALLOCATING(zio) ((zio)->io_orig_pipeline & ZIO_STAGE_DVA_ALLOCATE) int zio_requeue_io_start_cut_in_line = 1; #ifdef ZFS_DEBUG int zio_buf_debug_limit = 16384; #else int zio_buf_debug_limit = 0; #endif static inline void __zio_execute(zio_t *zio); static void zio_taskq_dispatch(zio_t *, zio_taskq_type_t, boolean_t); void zio_init(void) { size_t c; vmem_t *data_alloc_arena = NULL; zio_cache = kmem_cache_create("zio_cache", sizeof (zio_t), 0, NULL, NULL, NULL, NULL, NULL, 0); zio_link_cache = kmem_cache_create("zio_link_cache", sizeof (zio_link_t), 0, NULL, NULL, NULL, NULL, NULL, 0); /* * For small buffers, we want a cache for each multiple of * SPA_MINBLOCKSIZE. For larger buffers, we want a cache * for each quarter-power of 2. */ for (c = 0; c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; c++) { size_t size = (c + 1) << SPA_MINBLOCKSHIFT; size_t p2 = size; size_t align = 0; size_t cflags = (size > zio_buf_debug_limit) ? KMC_NODEBUG : 0; #if defined(_ILP32) && defined(_KERNEL) /* * Cache size limited to 1M on 32-bit platforms until ARC * buffers no longer require virtual address space. */ if (size > zfs_max_recordsize) break; #endif while (!ISP2(p2)) p2 &= p2 - 1; #ifndef _KERNEL /* * If we are using watchpoints, put each buffer on its own page, * to eliminate the performance overhead of trapping to the * kernel when modifying a non-watched buffer that shares the * page with a watched buffer. */ if (arc_watch && !IS_P2ALIGNED(size, PAGESIZE)) continue; /* * Here's the problem - on 4K native devices in userland on * Linux using O_DIRECT, buffers must be 4K aligned or I/O * will fail with EINVAL, causing zdb (and others) to coredump. * Since userland probably doesn't need optimized buffer caches, * we just force 4K alignment on everything. */ align = 8 * SPA_MINBLOCKSIZE; #else if (size < PAGESIZE) { align = SPA_MINBLOCKSIZE; } else if (IS_P2ALIGNED(size, p2 >> 2)) { align = PAGESIZE; } #endif if (align != 0) { char name[36]; (void) sprintf(name, "zio_buf_%lu", (ulong_t)size); zio_buf_cache[c] = kmem_cache_create(name, size, align, NULL, NULL, NULL, NULL, NULL, cflags); (void) sprintf(name, "zio_data_buf_%lu", (ulong_t)size); zio_data_buf_cache[c] = kmem_cache_create(name, size, align, NULL, NULL, NULL, NULL, data_alloc_arena, cflags); } } while (--c != 0) { ASSERT(zio_buf_cache[c] != NULL); if (zio_buf_cache[c - 1] == NULL) zio_buf_cache[c - 1] = zio_buf_cache[c]; ASSERT(zio_data_buf_cache[c] != NULL); if (zio_data_buf_cache[c - 1] == NULL) zio_data_buf_cache[c - 1] = zio_data_buf_cache[c]; } zio_inject_init(); lz4_init(); } void zio_fini(void) { size_t c; kmem_cache_t *last_cache = NULL; kmem_cache_t *last_data_cache = NULL; for (c = 0; c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; c++) { #ifdef _ILP32 /* * Cache size limited to 1M on 32-bit platforms until ARC * buffers no longer require virtual address space. */ if (((c + 1) << SPA_MINBLOCKSHIFT) > zfs_max_recordsize) break; #endif #if defined(ZFS_DEBUG) && !defined(_KERNEL) if (zio_buf_cache_allocs[c] != zio_buf_cache_frees[c]) (void) printf("zio_fini: [%d] %llu != %llu\n", (int)((c + 1) << SPA_MINBLOCKSHIFT), (long long unsigned)zio_buf_cache_allocs[c], (long long unsigned)zio_buf_cache_frees[c]); #endif if (zio_buf_cache[c] != last_cache) { last_cache = zio_buf_cache[c]; kmem_cache_destroy(zio_buf_cache[c]); } zio_buf_cache[c] = NULL; if (zio_data_buf_cache[c] != last_data_cache) { last_data_cache = zio_data_buf_cache[c]; kmem_cache_destroy(zio_data_buf_cache[c]); } zio_data_buf_cache[c] = NULL; } kmem_cache_destroy(zio_link_cache); kmem_cache_destroy(zio_cache); zio_inject_fini(); lz4_fini(); } /* * ========================================================================== * Allocate and free I/O buffers * ========================================================================== */ /* * Use zio_buf_alloc to allocate ZFS metadata. This data will appear in a * crashdump if the kernel panics, so use it judiciously. Obviously, it's * useful to inspect ZFS metadata, but if possible, we should avoid keeping * excess / transient data in-core during a crashdump. */ void * zio_buf_alloc(size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); #if defined(ZFS_DEBUG) && !defined(_KERNEL) atomic_add_64(&zio_buf_cache_allocs[c], 1); #endif return (kmem_cache_alloc(zio_buf_cache[c], KM_PUSHPAGE)); } /* * Use zio_data_buf_alloc to allocate data. The data will not appear in a * crashdump if the kernel panics. This exists so that we will limit the amount * of ZFS data that shows up in a kernel crashdump. (Thus reducing the amount * of kernel heap dumped to disk when the kernel panics) */ void * zio_data_buf_alloc(size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); return (kmem_cache_alloc(zio_data_buf_cache[c], KM_PUSHPAGE)); } void zio_buf_free(void *buf, size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); #if defined(ZFS_DEBUG) && !defined(_KERNEL) atomic_add_64(&zio_buf_cache_frees[c], 1); #endif kmem_cache_free(zio_buf_cache[c], buf); } void zio_data_buf_free(void *buf, size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); kmem_cache_free(zio_data_buf_cache[c], buf); } static void zio_abd_free(void *abd, size_t size) { abd_free((abd_t *)abd); } /* * ========================================================================== * Push and pop I/O transform buffers * ========================================================================== */ void zio_push_transform(zio_t *zio, abd_t *data, uint64_t size, uint64_t bufsize, zio_transform_func_t *transform) { zio_transform_t *zt = kmem_alloc(sizeof (zio_transform_t), KM_SLEEP); /* * Ensure that anyone expecting this zio to contain a linear ABD isn't * going to get a nasty surprise when they try to access the data. */ IMPLY(abd_is_linear(zio->io_abd), abd_is_linear(data)); zt->zt_orig_abd = zio->io_abd; zt->zt_orig_size = zio->io_size; zt->zt_bufsize = bufsize; zt->zt_transform = transform; zt->zt_next = zio->io_transform_stack; zio->io_transform_stack = zt; zio->io_abd = data; zio->io_size = size; } void zio_pop_transforms(zio_t *zio) { zio_transform_t *zt; while ((zt = zio->io_transform_stack) != NULL) { if (zt->zt_transform != NULL) zt->zt_transform(zio, zt->zt_orig_abd, zt->zt_orig_size); if (zt->zt_bufsize != 0) abd_free(zio->io_abd); zio->io_abd = zt->zt_orig_abd; zio->io_size = zt->zt_orig_size; zio->io_transform_stack = zt->zt_next; kmem_free(zt, sizeof (zio_transform_t)); } } /* * ========================================================================== * I/O transform callbacks for subblocks, decompression, and decryption * ========================================================================== */ static void zio_subblock(zio_t *zio, abd_t *data, uint64_t size) { ASSERT(zio->io_size > size); if (zio->io_type == ZIO_TYPE_READ) abd_copy(data, zio->io_abd, size); } static void zio_decompress(zio_t *zio, abd_t *data, uint64_t size) { if (zio->io_error == 0) { void *tmp = abd_borrow_buf(data, size); int ret = zio_decompress_data(BP_GET_COMPRESS(zio->io_bp), zio->io_abd, tmp, zio->io_size, size); abd_return_buf_copy(data, tmp, size); 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; uint64_t lsize = BP_GET_LSIZE(bp); dmu_object_type_t ot = BP_GET_TYPE(bp); uint8_t salt[ZIO_DATA_SALT_LEN]; uint8_t iv[ZIO_DATA_IV_LEN]; uint8_t mac[ZIO_DATA_MAC_LEN]; boolean_t no_crypt = B_FALSE; ASSERT(BP_USES_CRYPT(bp)); ASSERT3U(size, !=, 0); if (zio->io_error != 0) return; /* * Verify the cksum of MACs stored in an indirect bp. It will always * be possible to verify this since it does not require an encryption * key. */ if (BP_HAS_INDIRECT_MAC_CKSUM(bp)) { zio_crypt_decode_mac_bp(bp, mac); if (BP_GET_COMPRESS(bp) != ZIO_COMPRESS_OFF) { /* * We haven't decompressed the data yet, but * zio_crypt_do_indirect_mac_checksum() requires * decompressed data to be able to parse out the MACs * from the indirect block. We decompress it now and * throw away the result after we are finished. */ tmp = zio_buf_alloc(lsize); ret = zio_decompress_data(BP_GET_COMPRESS(bp), zio->io_abd, tmp, zio->io_size, lsize); if (ret != 0) { ret = SET_ERROR(EIO); goto error; } ret = zio_crypt_do_indirect_mac_checksum(B_FALSE, tmp, lsize, BP_SHOULD_BYTESWAP(bp), mac); zio_buf_free(tmp, lsize); } else { ret = zio_crypt_do_indirect_mac_checksum_abd(B_FALSE, zio->io_abd, size, BP_SHOULD_BYTESWAP(bp), mac); } abd_copy(data, zio->io_abd, size); if (ret != 0) goto error; return; } /* * If this is an authenticated block, just check the MAC. It would be * nice to separate this out into its own flag, but for the moment * enum zio_flag is out of bits. */ if (BP_IS_AUTHENTICATED(bp)) { if (ot == DMU_OT_OBJSET) { ret = spa_do_crypt_objset_mac_abd(B_FALSE, zio->io_spa, zio->io_bookmark.zb_objset, zio->io_abd, size, BP_SHOULD_BYTESWAP(bp)); } else { zio_crypt_decode_mac_bp(bp, mac); ret = spa_do_crypt_mac_abd(B_FALSE, zio->io_spa, zio->io_bookmark.zb_objset, zio->io_abd, size, mac); } 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, zio->io_spa, zio->io_bookmark.zb_objset, bp, bp->blk_birth, size, data, zio->io_abd, iv, mac, salt, &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 != ENOENT || (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) { ret = SET_ERROR(EIO); if ((zio->io_flags & ZIO_FLAG_SPECULATIVE) == 0) { zfs_ereport_post(FM_EREPORT_ZFS_AUTHENTICATION, zio->io_spa, NULL, &zio->io_bookmark, zio, 0, 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; *zl = (*zl == NULL) ? list_head(cl) : list_next(cl, *zl); if (*zl == NULL) return (NULL); ASSERT((*zl)->zl_parent == pio); return ((*zl)->zl_child); } zio_t * zio_unique_parent(zio_t *cio) { zio_link_t *zl = NULL; zio_t *pio = zio_walk_parents(cio, &zl); VERIFY3P(zio_walk_parents(cio, &zl), ==, NULL); return (pio); } void zio_add_child(zio_t *pio, zio_t *cio) { zio_link_t *zl = kmem_cache_alloc(zio_link_cache, KM_SLEEP); /* * Logical I/Os can have logical, gang, or vdev children. * Gang I/Os can have gang or vdev children. * Vdev I/Os can only have vdev children. * The following ASSERT captures all of these constraints. */ - ASSERT(cio->io_child_type <= pio->io_child_type); + ASSERT3S(cio->io_child_type, <=, pio->io_child_type); zl->zl_parent = pio; zl->zl_child = cio; mutex_enter(&cio->io_lock); mutex_enter(&pio->io_lock); ASSERT(pio->io_state[ZIO_WAIT_DONE] == 0); for (int w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_children[cio->io_child_type][w] += !cio->io_state[w]; list_insert_head(&pio->io_child_list, zl); list_insert_head(&cio->io_parent_list, zl); pio->io_child_count++; cio->io_parent_count++; mutex_exit(&pio->io_lock); mutex_exit(&cio->io_lock); } static void zio_remove_child(zio_t *pio, zio_t *cio, zio_link_t *zl) { ASSERT(zl->zl_parent == pio); ASSERT(zl->zl_child == cio); mutex_enter(&cio->io_lock); mutex_enter(&pio->io_lock); list_remove(&pio->io_child_list, zl); list_remove(&cio->io_parent_list, zl); pio->io_child_count--; cio->io_parent_count--; mutex_exit(&pio->io_lock); mutex_exit(&cio->io_lock); kmem_cache_free(zio_link_cache, zl); } static boolean_t zio_wait_for_children(zio_t *zio, enum zio_child child, enum zio_wait_type wait) { uint64_t *countp = &zio->io_children[child][wait]; boolean_t waiting = B_FALSE; mutex_enter(&zio->io_lock); ASSERT(zio->io_stall == NULL); if (*countp != 0) { zio->io_stage >>= 1; ASSERT3U(zio->io_stage, !=, ZIO_STAGE_OPEN); zio->io_stall = countp; waiting = B_TRUE; } 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) { uint64_t *countp = &pio->io_children[zio->io_child_type][wait]; int *errorp = &pio->io_child_error[zio->io_child_type]; mutex_enter(&pio->io_lock); if (zio->io_error && !(zio->io_flags & ZIO_FLAG_DONT_PROPAGATE)) *errorp = zio_worst_error(*errorp, zio->io_error); pio->io_reexecute |= zio->io_reexecute; ASSERT3U(*countp, >, 0); (*countp)--; if (*countp == 0 && pio->io_stall == countp) { zio_taskq_type_t type = pio->io_stage < ZIO_STAGE_VDEV_IO_START ? ZIO_TASKQ_ISSUE : ZIO_TASKQ_INTERRUPT; pio->io_stall = NULL; mutex_exit(&pio->io_lock); /* * Dispatch the parent zio in its own taskq so that * the child can continue to make progress. This also * prevents overflowing the stack when we have deeply nested * parent-child relationships. */ zio_taskq_dispatch(pio, type, B_FALSE); } else { mutex_exit(&pio->io_lock); } } static void zio_inherit_child_errors(zio_t *zio, enum zio_child c) { if (zio->io_child_error[c] != 0 && zio->io_error == 0) zio->io_error = zio->io_child_error[c]; } int zio_bookmark_compare(const void *x1, const void *x2) { const zio_t *z1 = x1; const zio_t *z2 = x2; if (z1->io_bookmark.zb_objset < z2->io_bookmark.zb_objset) return (-1); if (z1->io_bookmark.zb_objset > z2->io_bookmark.zb_objset) return (1); if (z1->io_bookmark.zb_object < z2->io_bookmark.zb_object) return (-1); if (z1->io_bookmark.zb_object > z2->io_bookmark.zb_object) return (1); if (z1->io_bookmark.zb_level < z2->io_bookmark.zb_level) return (-1); if (z1->io_bookmark.zb_level > z2->io_bookmark.zb_level) return (1); if (z1->io_bookmark.zb_blkid < z2->io_bookmark.zb_blkid) return (-1); if (z1->io_bookmark.zb_blkid > z2->io_bookmark.zb_blkid) return (1); if (z1 < z2) return (-1); if (z1 > z2) return (1); return (0); } /* * ========================================================================== * Create the various types of I/O (read, write, free, etc) * ========================================================================== */ static zio_t * zio_create(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, abd_t *data, uint64_t lsize, uint64_t psize, zio_done_func_t *done, void *private, zio_type_t type, zio_priority_t priority, enum zio_flag flags, vdev_t *vd, uint64_t offset, const zbookmark_phys_t *zb, enum zio_stage stage, enum zio_stage pipeline) { zio_t *zio; ASSERT3U(psize, <=, SPA_MAXBLOCKSIZE); ASSERT(P2PHASE(psize, SPA_MINBLOCKSIZE) == 0); ASSERT(P2PHASE(offset, SPA_MINBLOCKSIZE) == 0); ASSERT(!vd || spa_config_held(spa, SCL_STATE_ALL, RW_READER)); ASSERT(!bp || !(flags & ZIO_FLAG_CONFIG_WRITER)); ASSERT(vd || stage == ZIO_STAGE_OPEN); IMPLY(lsize != psize, (flags & ZIO_FLAG_RAW_COMPRESS) != 0); zio = kmem_cache_alloc(zio_cache, KM_SLEEP); bzero(zio, sizeof (zio_t)); mutex_init(&zio->io_lock, NULL, MUTEX_NOLOCKDEP, NULL); cv_init(&zio->io_cv, NULL, CV_DEFAULT, NULL); list_create(&zio->io_parent_list, sizeof (zio_link_t), offsetof(zio_link_t, zl_parent_node)); list_create(&zio->io_child_list, sizeof (zio_link_t), offsetof(zio_link_t, zl_child_node)); metaslab_trace_init(&zio->io_alloc_list); if (vd != NULL) zio->io_child_type = ZIO_CHILD_VDEV; else if (flags & ZIO_FLAG_GANG_CHILD) zio->io_child_type = ZIO_CHILD_GANG; else if (flags & ZIO_FLAG_DDT_CHILD) zio->io_child_type = ZIO_CHILD_DDT; else zio->io_child_type = ZIO_CHILD_LOGICAL; if (bp != NULL) { zio->io_bp = (blkptr_t *)bp; zio->io_bp_copy = *bp; zio->io_bp_orig = *bp; if (type != ZIO_TYPE_WRITE || zio->io_child_type == ZIO_CHILD_DDT) zio->io_bp = &zio->io_bp_copy; /* so caller can free */ if (zio->io_child_type == ZIO_CHILD_LOGICAL) zio->io_logical = zio; if (zio->io_child_type > ZIO_CHILD_GANG && BP_IS_GANG(bp)) pipeline |= ZIO_GANG_STAGES; } zio->io_spa = spa; zio->io_txg = txg; zio->io_done = done; zio->io_private = private; zio->io_type = type; zio->io_priority = priority; zio->io_vd = vd; zio->io_offset = offset; zio->io_orig_abd = zio->io_abd = data; zio->io_orig_size = zio->io_size = psize; zio->io_lsize = lsize; zio->io_orig_flags = zio->io_flags = flags; zio->io_orig_stage = zio->io_stage = stage; zio->io_orig_pipeline = zio->io_pipeline = pipeline; zio->io_pipeline_trace = ZIO_STAGE_OPEN; zio->io_state[ZIO_WAIT_READY] = (stage >= ZIO_STAGE_READY); zio->io_state[ZIO_WAIT_DONE] = (stage >= ZIO_STAGE_DONE); if (zb != NULL) zio->io_bookmark = *zb; if (pio != NULL) { if (zio->io_logical == NULL) zio->io_logical = pio->io_logical; if (zio->io_child_type == ZIO_CHILD_GANG) zio->io_gang_leader = pio->io_gang_leader; zio_add_child(pio, zio); } taskq_init_ent(&zio->io_tqent); return (zio); } static void zio_destroy(zio_t *zio) { metaslab_trace_fini(&zio->io_alloc_list); list_destroy(&zio->io_parent_list); list_destroy(&zio->io_child_list); mutex_destroy(&zio->io_lock); cv_destroy(&zio->io_cv); kmem_cache_free(zio_cache, zio); } zio_t * zio_null(zio_t *pio, spa_t *spa, vdev_t *vd, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; zio = zio_create(pio, spa, 0, NULL, NULL, 0, 0, done, private, ZIO_TYPE_NULL, ZIO_PRIORITY_NOW, flags, vd, 0, NULL, ZIO_STAGE_OPEN, ZIO_INTERLOCK_PIPELINE); return (zio); } zio_t * zio_root(spa_t *spa, zio_done_func_t *done, void *private, enum zio_flag flags) { return (zio_null(NULL, spa, NULL, done, private, flags)); } void zfs_blkptr_verify(spa_t *spa, const blkptr_t *bp) { if (!DMU_OT_IS_VALID(BP_GET_TYPE(bp))) { zfs_panic_recover("blkptr at %p has invalid TYPE %llu", bp, (longlong_t)BP_GET_TYPE(bp)); } if (BP_GET_CHECKSUM(bp) >= ZIO_CHECKSUM_FUNCTIONS || BP_GET_CHECKSUM(bp) <= ZIO_CHECKSUM_ON) { zfs_panic_recover("blkptr at %p has invalid CHECKSUM %llu", bp, (longlong_t)BP_GET_CHECKSUM(bp)); } if (BP_GET_COMPRESS(bp) >= ZIO_COMPRESS_FUNCTIONS || BP_GET_COMPRESS(bp) <= ZIO_COMPRESS_ON) { zfs_panic_recover("blkptr at %p has invalid COMPRESS %llu", bp, (longlong_t)BP_GET_COMPRESS(bp)); } if (BP_GET_LSIZE(bp) > SPA_MAXBLOCKSIZE) { zfs_panic_recover("blkptr at %p has invalid LSIZE %llu", bp, (longlong_t)BP_GET_LSIZE(bp)); } if (BP_GET_PSIZE(bp) > SPA_MAXBLOCKSIZE) { zfs_panic_recover("blkptr at %p has invalid PSIZE %llu", bp, (longlong_t)BP_GET_PSIZE(bp)); } if (BP_IS_EMBEDDED(bp)) { if (BPE_GET_ETYPE(bp) > NUM_BP_EMBEDDED_TYPES) { zfs_panic_recover("blkptr at %p has invalid ETYPE %llu", bp, (longlong_t)BPE_GET_ETYPE(bp)); } } /* * Pool-specific checks. * * Note: it would be nice to verify that the blk_birth and * BP_PHYSICAL_BIRTH() are not too large. However, spa_freeze() * allows the birth time of log blocks (and dmu_sync()-ed blocks * that are in the log) to be arbitrarily large. */ for (int i = 0; i < BP_GET_NDVAS(bp); i++) { uint64_t vdevid = DVA_GET_VDEV(&bp->blk_dva[i]); if (vdevid >= spa->spa_root_vdev->vdev_children) { zfs_panic_recover("blkptr at %p DVA %u has invalid " "VDEV %llu", bp, i, (longlong_t)vdevid); continue; } vdev_t *vd = spa->spa_root_vdev->vdev_child[vdevid]; if (vd == NULL) { zfs_panic_recover("blkptr at %p DVA %u has invalid " "VDEV %llu", bp, i, (longlong_t)vdevid); continue; } if (vd->vdev_ops == &vdev_hole_ops) { zfs_panic_recover("blkptr at %p DVA %u has hole " "VDEV %llu", bp, i, (longlong_t)vdevid); continue; } if (vd->vdev_ops == &vdev_missing_ops) { /* * "missing" vdevs are valid during import, but we * don't have their detailed info (e.g. asize), so * we can't perform any more checks on them. */ continue; } uint64_t offset = DVA_GET_OFFSET(&bp->blk_dva[i]); uint64_t asize = DVA_GET_ASIZE(&bp->blk_dva[i]); if (BP_IS_GANG(bp)) asize = vdev_psize_to_asize(vd, SPA_GANGBLOCKSIZE); if (offset + asize > vd->vdev_asize) { zfs_panic_recover("blkptr at %p DVA %u has invalid " "OFFSET %llu", bp, i, (longlong_t)offset); } } } zio_t * zio_read(zio_t *pio, spa_t *spa, const blkptr_t *bp, abd_t *data, uint64_t size, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, const zbookmark_phys_t *zb) { zio_t *zio; zfs_blkptr_verify(spa, bp); zio = zio_create(pio, spa, BP_PHYSICAL_BIRTH(bp), bp, data, size, size, done, private, ZIO_TYPE_READ, priority, flags, NULL, 0, zb, ZIO_STAGE_OPEN, (flags & ZIO_FLAG_DDT_CHILD) ? ZIO_DDT_CHILD_READ_PIPELINE : ZIO_READ_PIPELINE); return (zio); } zio_t * zio_write(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, abd_t *data, uint64_t lsize, uint64_t psize, const zio_prop_t *zp, zio_done_func_t *ready, zio_done_func_t *children_ready, zio_done_func_t *physdone, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, const zbookmark_phys_t *zb) { zio_t *zio; ASSERT(zp->zp_checksum >= ZIO_CHECKSUM_OFF && zp->zp_checksum < ZIO_CHECKSUM_FUNCTIONS && zp->zp_compress >= ZIO_COMPRESS_OFF && zp->zp_compress < ZIO_COMPRESS_FUNCTIONS && DMU_OT_IS_VALID(zp->zp_type) && zp->zp_level < 32 && zp->zp_copies > 0 && zp->zp_copies <= spa_max_replication(spa)); zio = zio_create(pio, spa, txg, bp, data, lsize, psize, done, private, ZIO_TYPE_WRITE, priority, flags, NULL, 0, zb, ZIO_STAGE_OPEN, (flags & ZIO_FLAG_DDT_CHILD) ? ZIO_DDT_CHILD_WRITE_PIPELINE : ZIO_WRITE_PIPELINE); zio->io_ready = ready; zio->io_children_ready = children_ready; zio->io_physdone = physdone; zio->io_prop = *zp; /* * Data can be NULL if we are going to call zio_write_override() to * provide the already-allocated BP. But we may need the data to * verify a dedup hit (if requested). In this case, don't try to * dedup (just take the already-allocated BP verbatim). Encrypted * dedup blocks need data as well so we also disable dedup in this * case. */ if (data == NULL && (zio->io_prop.zp_dedup_verify || zio->io_prop.zp_encrypt)) { zio->io_prop.zp_dedup = zio->io_prop.zp_dedup_verify = B_FALSE; } return (zio); } zio_t * zio_rewrite(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, abd_t *data, uint64_t size, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, zbookmark_phys_t *zb) { zio_t *zio; zio = zio_create(pio, spa, txg, bp, data, size, size, done, private, ZIO_TYPE_WRITE, priority, flags | ZIO_FLAG_IO_REWRITE, NULL, 0, zb, ZIO_STAGE_OPEN, ZIO_REWRITE_PIPELINE); return (zio); } void zio_write_override(zio_t *zio, blkptr_t *bp, int copies, boolean_t nopwrite) { ASSERT(zio->io_type == ZIO_TYPE_WRITE); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(zio->io_stage == ZIO_STAGE_OPEN); ASSERT(zio->io_txg == spa_syncing_txg(zio->io_spa)); /* * We must reset the io_prop to match the values that existed * when the bp was first written by dmu_sync() keeping in mind * that nopwrite and dedup are mutually exclusive. */ zio->io_prop.zp_dedup = nopwrite ? B_FALSE : zio->io_prop.zp_dedup; zio->io_prop.zp_nopwrite = nopwrite; zio->io_prop.zp_copies = copies; zio->io_bp_override = bp; } void zio_free(spa_t *spa, uint64_t txg, const blkptr_t *bp) { /* * The check for EMBEDDED is a performance optimization. We * process the free here (by ignoring it) rather than * putting it on the list and then processing it in zio_free_sync(). */ if (BP_IS_EMBEDDED(bp)) return; metaslab_check_free(spa, bp); /* * Frees that are for the currently-syncing txg, are not going to be * deferred, and which will not need to do a read (i.e. not GANG or * DEDUP), can be processed immediately. Otherwise, put them on the * in-memory list for later processing. */ if (BP_IS_GANG(bp) || BP_GET_DEDUP(bp) || txg != spa->spa_syncing_txg || spa_sync_pass(spa) >= zfs_sync_pass_deferred_free) { bplist_append(&spa->spa_free_bplist[txg & TXG_MASK], bp); } else { VERIFY0(zio_wait(zio_free_sync(NULL, spa, txg, bp, 0))); } } zio_t * zio_free_sync(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, enum zio_flag flags) { zio_t *zio; enum zio_stage stage = ZIO_FREE_PIPELINE; ASSERT(!BP_IS_HOLE(bp)); ASSERT(spa_syncing_txg(spa) == txg); ASSERT(spa_sync_pass(spa) < zfs_sync_pass_deferred_free); if (BP_IS_EMBEDDED(bp)) return (zio_null(pio, spa, NULL, NULL, NULL, 0)); metaslab_check_free(spa, bp); arc_freed(spa, bp); dsl_scan_freed(spa, bp); /* * GANG and DEDUP blocks can induce a read (for the gang block header, * or the DDT), so issue them asynchronously so that this thread is * not tied up. */ if (BP_IS_GANG(bp) || BP_GET_DEDUP(bp)) stage |= ZIO_STAGE_ISSUE_ASYNC; zio = 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); return (zio); } zio_t * zio_claim(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; dprintf_bp(bp, "claiming in txg %llu", txg); if (BP_IS_EMBEDDED(bp)) return (zio_null(pio, spa, NULL, NULL, NULL, 0)); /* * A claim is an allocation of a specific block. Claims are needed * to support immediate writes in the intent log. The issue is that * immediate writes contain committed data, but in a txg that was * *not* committed. Upon opening the pool after an unclean shutdown, * the intent log claims all blocks that contain immediate write data * so that the SPA knows they're in use. * * All claims *must* be resolved in the first txg -- before the SPA * starts allocating blocks -- so that nothing is allocated twice. * If txg == 0 we just verify that the block is claimable. */ ASSERT3U(spa->spa_uberblock.ub_rootbp.blk_birth, <, spa_first_txg(spa)); ASSERT(txg == spa_first_txg(spa) || txg == 0); ASSERT(!BP_GET_DEDUP(bp) || !spa_writeable(spa)); /* zdb(1M) */ zio = zio_create(pio, spa, txg, bp, NULL, BP_GET_PSIZE(bp), BP_GET_PSIZE(bp), done, private, ZIO_TYPE_CLAIM, ZIO_PRIORITY_NOW, flags, NULL, 0, NULL, ZIO_STAGE_OPEN, ZIO_CLAIM_PIPELINE); ASSERT0(zio->io_queued_timestamp); return (zio); } zio_t * zio_ioctl(zio_t *pio, spa_t *spa, vdev_t *vd, int cmd, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; int c; if (vd->vdev_children == 0) { zio = zio_create(pio, spa, 0, NULL, NULL, 0, 0, done, private, ZIO_TYPE_IOCTL, ZIO_PRIORITY_NOW, flags, vd, 0, NULL, ZIO_STAGE_OPEN, ZIO_IOCTL_PIPELINE); zio->io_cmd = cmd; } else { zio = zio_null(pio, spa, NULL, NULL, NULL, flags); for (c = 0; c < vd->vdev_children; c++) zio_nowait(zio_ioctl(zio, spa, vd->vdev_child[c], cmd, done, private, flags)); } return (zio); } zio_t * zio_read_phys(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, abd_t *data, int checksum, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, boolean_t labels) { zio_t *zio; ASSERT(vd->vdev_children == 0); ASSERT(!labels || offset + size <= VDEV_LABEL_START_SIZE || offset >= vd->vdev_psize - VDEV_LABEL_END_SIZE); ASSERT3U(offset + size, <=, vd->vdev_psize); zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, size, done, private, ZIO_TYPE_READ, priority, flags | ZIO_FLAG_PHYSICAL, vd, offset, NULL, ZIO_STAGE_OPEN, ZIO_READ_PHYS_PIPELINE); zio->io_prop.zp_checksum = checksum; return (zio); } zio_t * zio_write_phys(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, abd_t *data, int checksum, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, boolean_t labels) { zio_t *zio; ASSERT(vd->vdev_children == 0); ASSERT(!labels || offset + size <= VDEV_LABEL_START_SIZE || offset >= vd->vdev_psize - VDEV_LABEL_END_SIZE); ASSERT3U(offset + size, <=, vd->vdev_psize); zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, size, done, private, ZIO_TYPE_WRITE, priority, flags | ZIO_FLAG_PHYSICAL, vd, offset, NULL, ZIO_STAGE_OPEN, ZIO_WRITE_PHYS_PIPELINE); zio->io_prop.zp_checksum = checksum; if (zio_checksum_table[checksum].ci_flags & ZCHECKSUM_FLAG_EMBEDDED) { /* * zec checksums are necessarily destructive -- they modify * the end of the write buffer to hold the verifier/checksum. * Therefore, we must make a local copy in case the data is * being written to multiple places in parallel. */ abd_t *wbuf = abd_alloc_sametype(data, size); abd_copy(wbuf, data, size); zio_push_transform(zio, wbuf, size, size, NULL); } return (zio); } /* * Create a child I/O to do some work for us. */ zio_t * zio_vdev_child_io(zio_t *pio, blkptr_t *bp, vdev_t *vd, uint64_t offset, abd_t *data, uint64_t size, int type, zio_priority_t priority, enum zio_flag flags, zio_done_func_t *done, void *private) { enum zio_stage pipeline = ZIO_VDEV_CHILD_PIPELINE; zio_t *zio; ASSERT(vd->vdev_parent == (pio->io_vd ? pio->io_vd : pio->io_spa->spa_root_vdev)); if (type == ZIO_TYPE_READ && bp != NULL) { /* * If we have the bp, then the child should perform the * checksum and the parent need not. This pushes error * detection as close to the leaves as possible and * eliminates redundant checksums in the interior nodes. */ pipeline |= ZIO_STAGE_CHECKSUM_VERIFY; pio->io_pipeline &= ~ZIO_STAGE_CHECKSUM_VERIFY; } if (vd->vdev_children == 0) offset += VDEV_LABEL_START_SIZE; flags |= ZIO_VDEV_CHILD_FLAGS(pio) | ZIO_FLAG_DONT_PROPAGATE; /* * If we've decided to do a repair, the write is not speculative -- * even if the original read was. */ if (flags & ZIO_FLAG_IO_REPAIR) flags &= ~ZIO_FLAG_SPECULATIVE; /* * If we're creating a child I/O that is not associated with a * top-level vdev, then the child zio is not an allocating I/O. * If this is a retried I/O then we ignore it since we will * have already processed the original allocating I/O. */ if (flags & ZIO_FLAG_IO_ALLOCATING && (vd != vd->vdev_top || (flags & ZIO_FLAG_IO_RETRY))) { ASSERTV(metaslab_class_t *mc = spa_normal_class(pio->io_spa)); ASSERT(mc->mc_alloc_throttle_enabled); ASSERT(type == ZIO_TYPE_WRITE); ASSERT(priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(!(flags & ZIO_FLAG_IO_REPAIR)); ASSERT(!(pio->io_flags & ZIO_FLAG_IO_REWRITE) || pio->io_child_type == ZIO_CHILD_GANG); flags &= ~ZIO_FLAG_IO_ALLOCATING; } zio = zio_create(pio, pio->io_spa, pio->io_txg, bp, data, size, size, done, private, type, priority, flags, vd, offset, &pio->io_bookmark, ZIO_STAGE_VDEV_IO_START >> 1, pipeline); ASSERT3U(zio->io_child_type, ==, ZIO_CHILD_VDEV); zio->io_physdone = pio->io_physdone; if (vd->vdev_ops->vdev_op_leaf && zio->io_logical != NULL) zio->io_logical->io_phys_children++; return (zio); } zio_t * zio_vdev_delegated_io(vdev_t *vd, uint64_t offset, abd_t *data, uint64_t size, int type, zio_priority_t priority, enum zio_flag flags, zio_done_func_t *done, void *private) { zio_t *zio; ASSERT(vd->vdev_ops->vdev_op_leaf); zio = zio_create(NULL, vd->vdev_spa, 0, NULL, data, size, size, done, private, type, priority, flags | ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_RETRY | ZIO_FLAG_DELEGATED, vd, offset, NULL, ZIO_STAGE_VDEV_IO_START >> 1, ZIO_VDEV_CHILD_PIPELINE); return (zio); } void zio_flush(zio_t *zio, vdev_t *vd) { zio_nowait(zio_ioctl(zio, zio->io_spa, vd, DKIOCFLUSHWRITECACHE, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_DONT_RETRY)); } void zio_shrink(zio_t *zio, uint64_t size) { - ASSERT(zio->io_executor == NULL); - ASSERT(zio->io_orig_size == zio->io_size); - ASSERT(size <= zio->io_size); + ASSERT3P(zio->io_executor, ==, NULL); + ASSERT3U(zio->io_orig_size, ==, zio->io_size); + ASSERT3U(size, <=, zio->io_size); /* * We don't shrink for raidz because of problems with the * reconstruction when reading back less than the block size. * Note, BP_IS_RAIDZ() assumes no compression. */ ASSERT(BP_GET_COMPRESS(zio->io_bp) == ZIO_COMPRESS_OFF); if (!BP_IS_RAIDZ(zio->io_bp)) { /* we are not doing a raw write */ ASSERT3U(zio->io_size, ==, zio->io_lsize); zio->io_orig_size = zio->io_size = zio->io_lsize = size; } } /* * ========================================================================== * Prepare to read and write logical blocks * ========================================================================== */ static int 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); 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 (!DMU_OT_IS_METADATA(BP_GET_TYPE(bp)) && BP_GET_LEVEL(bp) == 0) zio->io_flags |= ZIO_FLAG_DONT_CACHE; if (BP_GET_TYPE(bp) == DMU_OT_DDT_ZAP) zio->io_flags |= ZIO_FLAG_DONT_CACHE; if (BP_GET_DEDUP(bp) && zio->io_child_type == ZIO_CHILD_LOGICAL) zio->io_pipeline = ZIO_DDT_READ_PIPELINE; return (ZIO_PIPELINE_CONTINUE); } static int zio_write_bp_init(zio_t *zio) { if (!IO_IS_ALLOCATING(zio)) return (ZIO_PIPELINE_CONTINUE); ASSERT(zio->io_child_type != ZIO_CHILD_DDT); if (zio->io_bp_override) { blkptr_t *bp = zio->io_bp; zio_prop_t *zp = &zio->io_prop; ASSERT(bp->blk_birth != zio->io_txg); ASSERT(BP_GET_DEDUP(zio->io_bp_override) == 0); *bp = *zio->io_bp_override; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (BP_IS_EMBEDDED(bp)) return (ZIO_PIPELINE_CONTINUE); /* * 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_PIPELINE_CONTINUE); } ASSERT(!zp->zp_nopwrite); if (BP_IS_HOLE(bp) || !zp->zp_dedup) return (ZIO_PIPELINE_CONTINUE); 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_PIPELINE_CONTINUE); } /* * 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_PIPELINE_CONTINUE); } static int zio_write_compress(zio_t *zio) { spa_t *spa = zio->io_spa; zio_prop_t *zp = &zio->io_prop; enum zio_compress compress = zp->zp_compress; blkptr_t *bp = zio->io_bp; uint64_t lsize = zio->io_lsize; uint64_t psize = zio->io_size; int pass = 1; /* * If our children haven't all reached the ready stage, * wait for them and then repeat this pipeline stage. */ if (zio_wait_for_children(zio, ZIO_CHILD_GANG, ZIO_WAIT_READY) || zio_wait_for_children(zio, ZIO_CHILD_LOGICAL, ZIO_WAIT_READY)) return (ZIO_PIPELINE_STOP); if (!IO_IS_ALLOCATING(zio)) return (ZIO_PIPELINE_CONTINUE); if (zio->io_children_ready != NULL) { /* * Now that all our children are ready, run the callback * associated with this zio in case it wants to modify the * data to be written. */ ASSERT3U(zp->zp_level, >, 0); zio->io_children_ready(zio); } ASSERT(zio->io_child_type != ZIO_CHILD_DDT); ASSERT(zio->io_bp_override == NULL); if (!BP_IS_HOLE(bp) && bp->blk_birth == zio->io_txg) { /* * We're rewriting an existing block, which means we're * working on behalf of spa_sync(). For spa_sync() to * converge, it must eventually be the case that we don't * have to allocate new blocks. But compression changes * the blocksize, which forces a reallocate, and makes * convergence take longer. Therefore, after the first * few passes, stop compressing to ensure convergence. */ pass = spa_sync_pass(spa); ASSERT(zio->io_txg == spa_syncing_txg(spa)); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!BP_GET_DEDUP(bp)); if (pass >= zfs_sync_pass_dont_compress) compress = ZIO_COMPRESS_OFF; /* Make sure someone doesn't change their mind on overwrites */ ASSERT(BP_IS_EMBEDDED(bp) || MIN(zp->zp_copies + BP_IS_GANG(bp), spa_max_replication(spa)) == BP_GET_NDVAS(bp)); } /* If it's a compressed write that is not raw, compress the buffer. */ if (compress != ZIO_COMPRESS_OFF && !(zio->io_flags & ZIO_FLAG_RAW_COMPRESS)) { void *cbuf = zio_buf_alloc(lsize); psize = zio_compress_data(compress, zio->io_abd, cbuf, lsize); if (psize == 0 || psize == lsize) { compress = ZIO_COMPRESS_OFF; zio_buf_free(cbuf, lsize); } else if (!zp->zp_dedup && !zp->zp_encrypt && psize <= BPE_PAYLOAD_SIZE && zp->zp_level == 0 && !DMU_OT_HAS_FILL(zp->zp_type) && spa_feature_is_enabled(spa, SPA_FEATURE_EMBEDDED_DATA)) { encode_embedded_bp_compressed(bp, cbuf, compress, lsize, psize); BPE_SET_ETYPE(bp, BP_EMBEDDED_TYPE_DATA); BP_SET_TYPE(bp, zio->io_prop.zp_type); BP_SET_LEVEL(bp, zio->io_prop.zp_level); zio_buf_free(cbuf, lsize); bp->blk_birth = zio->io_txg; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; ASSERT(spa_feature_is_active(spa, SPA_FEATURE_EMBEDDED_DATA)); return (ZIO_PIPELINE_CONTINUE); } else { /* * Round up compressed size up to the ashift * of the smallest-ashift device, and zero the tail. * This ensures that the compressed size of the BP * (and thus compressratio property) are correct, * in that we charge for the padding used to fill out * the last sector. */ ASSERT3U(spa->spa_min_ashift, >=, SPA_MINBLOCKSHIFT); size_t rounded = (size_t)P2ROUNDUP(psize, 1ULL << spa->spa_min_ashift); if (rounded >= lsize) { compress = ZIO_COMPRESS_OFF; zio_buf_free(cbuf, lsize); psize = lsize; } else { abd_t *cdata = abd_get_from_buf(cbuf, lsize); abd_take_ownership_of_buf(cdata, B_TRUE); abd_zero_off(cdata, psize, rounded - psize); psize = rounded; zio_push_transform(zio, cdata, psize, lsize, NULL); } } /* * We were unable to handle this as an override bp, treat * it as a regular write I/O. */ zio->io_bp_override = NULL; *bp = zio->io_bp_orig; zio->io_pipeline = zio->io_orig_pipeline; } else { ASSERT3U(psize, !=, 0); } /* * The final pass of spa_sync() must be all rewrites, but the first * few passes offer a trade-off: allocating blocks defers convergence, * but newly allocated blocks are sequential, so they can be written * to disk faster. Therefore, we allow the first few passes of * spa_sync() to allocate new blocks, but force rewrites after that. * There should only be a handful of blocks after pass 1 in any case. */ if (!BP_IS_HOLE(bp) && bp->blk_birth == zio->io_txg && BP_GET_PSIZE(bp) == psize && pass >= zfs_sync_pass_rewrite) { ASSERT(psize != 0); enum zio_stage gang_stages = zio->io_pipeline & ZIO_GANG_STAGES; zio->io_pipeline = ZIO_REWRITE_PIPELINE | gang_stages; zio->io_flags |= ZIO_FLAG_IO_REWRITE; } else { BP_ZERO(bp); zio->io_pipeline = ZIO_WRITE_PIPELINE; } if (psize == 0) { if (zio->io_bp_orig.blk_birth != 0 && spa_feature_is_active(spa, SPA_FEATURE_HOLE_BIRTH)) { BP_SET_LSIZE(bp, lsize); BP_SET_TYPE(bp, zp->zp_type); BP_SET_LEVEL(bp, zp->zp_level); BP_SET_BIRTH(bp, zio->io_txg, 0); } zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; } else { ASSERT(zp->zp_checksum != ZIO_CHECKSUM_GANG_HEADER); BP_SET_LSIZE(bp, lsize); BP_SET_TYPE(bp, zp->zp_type); BP_SET_LEVEL(bp, zp->zp_level); BP_SET_PSIZE(bp, psize); BP_SET_COMPRESS(bp, compress); BP_SET_CHECKSUM(bp, zp->zp_checksum); BP_SET_DEDUP(bp, zp->zp_dedup); BP_SET_BYTEORDER(bp, ZFS_HOST_BYTEORDER); if (zp->zp_dedup) { ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); ASSERT(!zp->zp_encrypt || DMU_OT_IS_ENCRYPTED(zp->zp_type)); zio->io_pipeline = ZIO_DDT_WRITE_PIPELINE; } if (zp->zp_nopwrite) { ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); zio->io_pipeline |= ZIO_STAGE_NOP_WRITE; } } return (ZIO_PIPELINE_CONTINUE); } static int 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; } return (ZIO_PIPELINE_CONTINUE); } /* * ========================================================================== * Execute the I/O pipeline * ========================================================================== */ static void zio_taskq_dispatch(zio_t *zio, zio_taskq_type_t q, boolean_t cutinline) { spa_t *spa = zio->io_spa; zio_type_t t = zio->io_type; int flags = (cutinline ? TQ_FRONT : 0); /* * If we're a config writer or a probe, the normal issue and * interrupt threads may all be blocked waiting for the config lock. * In this case, select the otherwise-unused taskq for ZIO_TYPE_NULL. */ if (zio->io_flags & (ZIO_FLAG_CONFIG_WRITER | ZIO_FLAG_PROBE)) t = ZIO_TYPE_NULL; /* * A similar issue exists for the L2ARC write thread until L2ARC 2.0. */ if (t == ZIO_TYPE_WRITE && zio->io_vd && zio->io_vd->vdev_aux) t = ZIO_TYPE_NULL; /* * If this is a high priority I/O, then use the high priority taskq if * available. */ if (zio->io_priority == ZIO_PRIORITY_NOW && spa->spa_zio_taskq[t][q + 1].stqs_count != 0) q++; ASSERT3U(q, <, ZIO_TASKQ_TYPES); /* * NB: We are assuming that the zio can only be dispatched * to a single taskq at a time. It would be a grievous error * to dispatch the zio to another taskq at the same time. */ ASSERT(taskq_empty_ent(&zio->io_tqent)); spa_taskq_dispatch_ent(spa, t, q, (task_func_t *)zio_execute, zio, flags, &zio->io_tqent); } static boolean_t zio_taskq_member(zio_t *zio, zio_taskq_type_t q) { kthread_t *executor = zio->io_executor; spa_t *spa = zio->io_spa; 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 (taskq_member(tqs->stqs_taskq[i], executor)) return (B_TRUE); } } return (B_FALSE); } static int zio_issue_async(zio_t *zio) { zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_FALSE); return (ZIO_PIPELINE_STOP); } void zio_interrupt(zio_t *zio) { zio_taskq_dispatch(zio, ZIO_TASKQ_INTERRUPT, B_FALSE); } void zio_delay_interrupt(zio_t *zio) { /* * The timeout_generic() function isn't defined in userspace, so * rather than trying to implement the function, the zio delay * functionality has been disabled for userspace builds. */ #ifdef _KERNEL /* * If io_target_timestamp is zero, then no delay has been registered * for this IO, thus jump to the end of this function and "skip" the * delay; issuing it directly to the zio layer. */ if (zio->io_target_timestamp != 0) { hrtime_t now = gethrtime(); if (now >= zio->io_target_timestamp) { /* * This IO has already taken longer than the target * delay to complete, so we don't want to delay it * any longer; we "miss" the delay and issue it * directly to the zio layer. This is likely due to * the target latency being set to a value less than * the underlying hardware can satisfy (e.g. delay * set to 1ms, but the disks take 10ms to complete an * IO request). */ DTRACE_PROBE2(zio__delay__miss, zio_t *, zio, hrtime_t, now); zio_interrupt(zio); } else { taskqid_t tid; hrtime_t diff = zio->io_target_timestamp - now; clock_t expire_at_tick = ddi_get_lbolt() + NSEC_TO_TICK(diff); DTRACE_PROBE3(zio__delay__hit, zio_t *, zio, hrtime_t, now, hrtime_t, diff); if (NSEC_TO_TICK(diff) == 0) { /* Our delay is less than a jiffy - just spin */ zfs_sleep_until(zio->io_target_timestamp); } else { /* * Use taskq_dispatch_delay() in the place of * OpenZFS's timeout_generic(). */ tid = taskq_dispatch_delay(system_taskq, (task_func_t *)zio_interrupt, zio, TQ_NOSLEEP, expire_at_tick); if (tid == TASKQID_INVALID) { /* * Couldn't allocate a task. Just * finish the zio without a delay. */ zio_interrupt(zio); } } } return; } #endif DTRACE_PROBE1(zio__delay__skip, zio_t *, zio); zio_interrupt(zio); } /* * Execute the I/O pipeline until one of the following occurs: * (1) the I/O completes; (2) the pipeline stalls waiting for * dependent child I/Os; (3) the I/O issues, so we're waiting * for an I/O completion interrupt; (4) the I/O is delegated by * vdev-level caching or aggregation; (5) the I/O is deferred * due to vdev-level queueing; (6) the I/O is handed off to * another thread. In all cases, the pipeline stops whenever * there's no CPU work; it never burns a thread in cv_wait_io(). * * There's no locking on io_stage because there's no legitimate way * for multiple threads to be attempting to process the same I/O. */ static zio_pipe_stage_t *zio_pipeline[]; /* * zio_execute() is a wrapper around the static function * __zio_execute() so that we can force __zio_execute() to be * inlined. This reduces stack overhead which is important * because __zio_execute() is called recursively in several zio * code paths. zio_execute() itself cannot be inlined because * it is externally visible. */ void zio_execute(zio_t *zio) { fstrans_cookie_t cookie; cookie = spl_fstrans_mark(); __zio_execute(zio); spl_fstrans_unmark(cookie); } /* * Used to determine if in the current context the stack is sized large * enough to allow zio_execute() to be called recursively. A minimum * stack size of 16K is required to avoid needing to re-dispatch the zio. */ boolean_t zio_execute_stack_check(zio_t *zio) { #if !defined(HAVE_LARGE_STACKS) dsl_pool_t *dp = spa_get_dsl(zio->io_spa); /* Executing in txg_sync_thread() context. */ if (dp && curthread == dp->dp_tx.tx_sync_thread) return (B_TRUE); /* Pool initialization outside of zio_taskq context. */ if (dp && spa_is_initializing(dp->dp_spa) && !zio_taskq_member(zio, ZIO_TASKQ_ISSUE) && !zio_taskq_member(zio, ZIO_TASKQ_ISSUE_HIGH)) return (B_TRUE); #endif /* HAVE_LARGE_STACKS */ return (B_FALSE); } __attribute__((always_inline)) static inline void __zio_execute(zio_t *zio) { zio->io_executor = curthread; 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; int rv; 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; rv = zio_pipeline[highbit64(stage) - 1](zio); if (rv == ZIO_PIPELINE_STOP) return; ASSERT(rv == ZIO_PIPELINE_CONTINUE); } } /* * ========================================================================== * Initiate I/O, either sync or async * ========================================================================== */ int zio_wait(zio_t *zio) { int error; - ASSERT(zio->io_stage == ZIO_STAGE_OPEN); - ASSERT(zio->io_executor == NULL); + ASSERT3S(zio->io_stage, ==, ZIO_STAGE_OPEN); + ASSERT3P(zio->io_executor, ==, NULL); zio->io_waiter = curthread; ASSERT0(zio->io_queued_timestamp); zio->io_queued_timestamp = gethrtime(); __zio_execute(zio); mutex_enter(&zio->io_lock); while (zio->io_executor != NULL) cv_wait_io(&zio->io_cv, &zio->io_lock); mutex_exit(&zio->io_lock); error = zio->io_error; zio_destroy(zio); return (error); } void zio_nowait(zio_t *zio) { - ASSERT(zio->io_executor == NULL); + ASSERT3P(zio->io_executor, ==, NULL); if (zio->io_child_type == ZIO_CHILD_LOGICAL && zio_unique_parent(zio) == NULL) { zio_t *pio; /* * This is a logical async I/O with no parent to wait for it. * We add it to the spa_async_root_zio "Godfather" I/O which * will ensure they complete prior to unloading the pool. */ spa_t *spa = zio->io_spa; kpreempt_disable(); pio = spa->spa_async_zio_root[CPU_SEQID]; kpreempt_enable(); zio_add_child(pio, zio); } ASSERT0(zio->io_queued_timestamp); zio->io_queued_timestamp = gethrtime(); __zio_execute(zio); } /* * ========================================================================== - * Reexecute or suspend/resume failed I/O + * Reexecute, cancel, or suspend/resume failed I/O * ========================================================================== */ static void zio_reexecute(zio_t *pio) { zio_t *cio, *cio_next; ASSERT(pio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(pio->io_orig_stage == ZIO_STAGE_OPEN); ASSERT(pio->io_gang_leader == NULL); ASSERT(pio->io_gang_tree == NULL); pio->io_flags = pio->io_orig_flags; pio->io_stage = pio->io_orig_stage; pio->io_pipeline = pio->io_orig_pipeline; pio->io_reexecute = 0; pio->io_flags |= ZIO_FLAG_REEXECUTED; pio->io_pipeline_trace = 0; pio->io_error = 0; for (int w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_state[w] = 0; for (int c = 0; c < ZIO_CHILD_TYPES; c++) pio->io_child_error[c] = 0; if (IO_IS_ALLOCATING(pio)) BP_ZERO(pio->io_bp); /* * As we reexecute pio's children, new children could be created. * New children go to the head of pio's io_child_list, however, * so we will (correctly) not reexecute them. The key is that * the remainder of pio's io_child_list, from 'cio_next' onward, * cannot be affected by any side effects of reexecuting 'cio'. */ zio_link_t *zl = NULL; for (cio = zio_walk_children(pio, &zl); cio != NULL; cio = cio_next) { cio_next = zio_walk_children(pio, &zl); mutex_enter(&pio->io_lock); for (int w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_children[cio->io_child_type][w]++; mutex_exit(&pio->io_lock); zio_reexecute(cio); } /* * 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_cancel(zio_t *zio) +{ + /* + * Disallow cancellation of a zio that's already been issued. + */ + VERIFY3P(zio->io_executor, ==, NULL); + + zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; + zio->io_done = NULL; + + zio_nowait(zio); +} + void zio_suspend(spa_t *spa, zio_t *zio) { if (spa_get_failmode(spa) == ZIO_FAILURE_MODE_PANIC) fm_panic("Pool '%s' has encountered an uncorrectable I/O " "failure and the failure mode property for this pool " "is set to panic.", spa_name(spa)); cmn_err(CE_WARN, "Pool '%s' has encountered an uncorrectable I/O " "failure and has been suspended.\n", spa_name(spa)); zfs_ereport_post(FM_EREPORT_ZFS_IO_FAILURE, spa, NULL, NULL, NULL, 0, 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 = B_TRUE; if (zio != NULL) { ASSERT(!(zio->io_flags & ZIO_FLAG_GODFATHER)); ASSERT(zio != spa->spa_suspend_zio_root); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(zio_unique_parent(zio) == NULL); ASSERT(zio->io_stage == ZIO_STAGE_DONE); zio_add_child(spa->spa_suspend_zio_root, zio); } mutex_exit(&spa->spa_suspend_lock); } int zio_resume(spa_t *spa) { zio_t *pio; /* * Reexecute all previously suspended i/o. */ mutex_enter(&spa->spa_suspend_lock); spa->spa_suspended = B_FALSE; cv_broadcast(&spa->spa_suspend_cv); pio = spa->spa_suspend_zio_root; spa->spa_suspend_zio_root = NULL; mutex_exit(&spa->spa_suspend_lock); if (pio == NULL) return (0); zio_reexecute(pio); return (zio_wait(pio)); } void zio_resume_wait(spa_t *spa) { mutex_enter(&spa->spa_suspend_lock); while (spa_suspended(spa)) cv_wait(&spa->spa_suspend_cv, &spa->spa_suspend_lock); mutex_exit(&spa->spa_suspend_lock); } /* * ========================================================================== * Gang blocks. * * A gang block is a collection of small blocks that looks to the DMU * like one large block. When zio_dva_allocate() cannot find a block * of the requested size, due to either severe fragmentation or the pool * being nearly full, it calls zio_write_gang_block() to construct the * block from smaller fragments. * * A gang block consists of a gang header (zio_gbh_phys_t) and up to * three (SPA_GBH_NBLKPTRS) gang members. The gang header is just like * an indirect block: it's an array of block pointers. It consumes * only one sector and hence is allocatable regardless of fragmentation. * The gang header's bps point to its gang members, which hold the data. * * Gang blocks are self-checksumming, using the bp's * as the verifier to ensure uniqueness of the SHA256 checksum. * Critically, the gang block bp's blk_cksum is the checksum of the data, * not the gang header. This ensures that data block signatures (needed for * deduplication) are independent of how the block is physically stored. * * Gang blocks can be nested: a gang member may itself be a gang block. * Thus every gang block is a tree in which root and all interior nodes are * gang headers, and the leaves are normal blocks that contain user data. * The root of the gang tree is called the gang leader. * * To perform any operation (read, rewrite, free, claim) on a gang block, * zio_gang_assemble() first assembles the gang tree (minus data leaves) * in the io_gang_tree field of the original logical i/o by recursively * reading the gang leader and all gang headers below it. This yields * an in-core tree containing the contents of every gang header and the * bps for every constituent of the gang block. * * With the gang tree now assembled, zio_gang_issue() just walks the gang tree * and invokes a callback on each bp. To free a gang block, zio_gang_issue() * calls zio_free_gang() -- a trivial wrapper around zio_free() -- for each bp. * zio_claim_gang() provides a similarly trivial wrapper for zio_claim(). * zio_read_gang() is a wrapper around zio_read() that omits reading gang * headers, since we already have those in io_gang_tree. zio_rewrite_gang() * performs a zio_rewrite() of the data or, for gang headers, a zio_rewrite() * of the gang header plus zio_checksum_compute() of the data to update the * gang header's blk_cksum as described above. * * The two-phase assemble/issue model solves the problem of partial failure -- * what if you'd freed part of a gang block but then couldn't read the * gang header for another part? Assembling the entire gang tree first * ensures that all the necessary gang header I/O has succeeded before * starting the actual work of free, claim, or write. Once the gang tree * is assembled, free and claim are in-memory operations that cannot fail. * * In the event that a gang write fails, zio_dva_unallocate() walks the * gang tree to immediately free (i.e. insert back into the space map) * everything we've allocated. This ensures that we don't get ENOSPC * errors during repeated suspend/resume cycles due to a flaky device. * * Gang rewrites only happen during sync-to-convergence. If we can't assemble * the gang tree, we won't modify the block, so we can safely defer the free * (knowing that the block is still intact). If we *can* assemble the gang * tree, then even if some of the rewrites fail, zio_dva_unallocate() will free * each constituent bp and we can allocate a new block on the next sync pass. * * In all cases, the gang tree allows complete recovery from partial failure. * ========================================================================== */ static void zio_gang_issue_func_done(zio_t *zio) { abd_put(zio->io_abd); } static zio_t * zio_read_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, abd_t *data, uint64_t offset) { if (gn != NULL) return (pio); return (zio_read(pio, pio->io_spa, bp, abd_get_offset(data, offset), BP_GET_PSIZE(bp), zio_gang_issue_func_done, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark)); } static zio_t * zio_rewrite_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, abd_t *data, uint64_t offset) { zio_t *zio; if (gn != NULL) { abd_t *gbh_abd = abd_get_from_buf(gn->gn_gbh, SPA_GANGBLOCKSIZE); zio = zio_rewrite(pio, pio->io_spa, pio->io_txg, bp, gbh_abd, SPA_GANGBLOCKSIZE, zio_gang_issue_func_done, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); /* * As we rewrite each gang header, the pipeline will compute * a new gang block header checksum for it; but no one will * compute a new data checksum, so we do that here. The one * exception is the gang leader: the pipeline already computed * its data checksum because that stage precedes gang assembly. * (Presently, nothing actually uses interior data checksums; * this is just good hygiene.) */ if (gn != pio->io_gang_leader->io_gang_tree) { abd_t *buf = abd_get_offset(data, offset); zio_checksum_compute(zio, BP_GET_CHECKSUM(bp), buf, BP_GET_PSIZE(bp)); abd_put(buf); } /* * If we are here to damage data for testing purposes, * leave the GBH alone so that we can detect the damage. */ if (pio->io_gang_leader->io_flags & ZIO_FLAG_INDUCE_DAMAGE) zio->io_pipeline &= ~ZIO_VDEV_IO_STAGES; } else { zio = zio_rewrite(pio, pio->io_spa, pio->io_txg, bp, abd_get_offset(data, offset), BP_GET_PSIZE(bp), zio_gang_issue_func_done, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); } return (zio); } /* ARGSUSED */ static zio_t * zio_free_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, abd_t *data, uint64_t offset) { return (zio_free_sync(pio, pio->io_spa, pio->io_txg, bp, ZIO_GANG_CHILD_FLAGS(pio))); } /* ARGSUSED */ static zio_t * zio_claim_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, abd_t *data, uint64_t offset) { return (zio_claim(pio, pio->io_spa, pio->io_txg, bp, NULL, NULL, ZIO_GANG_CHILD_FLAGS(pio))); } static zio_gang_issue_func_t *zio_gang_issue_func[ZIO_TYPES] = { NULL, zio_read_gang, zio_rewrite_gang, zio_free_gang, zio_claim_gang, NULL }; static void zio_gang_tree_assemble_done(zio_t *zio); static zio_gang_node_t * zio_gang_node_alloc(zio_gang_node_t **gnpp) { zio_gang_node_t *gn; ASSERT(*gnpp == NULL); gn = kmem_zalloc(sizeof (*gn), KM_SLEEP); gn->gn_gbh = zio_buf_alloc(SPA_GANGBLOCKSIZE); *gnpp = gn; return (gn); } static void zio_gang_node_free(zio_gang_node_t **gnpp) { zio_gang_node_t *gn = *gnpp; for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) ASSERT(gn->gn_child[g] == NULL); zio_buf_free(gn->gn_gbh, SPA_GANGBLOCKSIZE); kmem_free(gn, sizeof (*gn)); *gnpp = NULL; } static void zio_gang_tree_free(zio_gang_node_t **gnpp) { zio_gang_node_t *gn = *gnpp; if (gn == NULL) return; for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) zio_gang_tree_free(&gn->gn_child[g]); zio_gang_node_free(gnpp); } static void zio_gang_tree_assemble(zio_t *gio, blkptr_t *bp, zio_gang_node_t **gnpp) { zio_gang_node_t *gn = zio_gang_node_alloc(gnpp); abd_t *gbh_abd = abd_get_from_buf(gn->gn_gbh, SPA_GANGBLOCKSIZE); ASSERT(gio->io_gang_leader == gio); ASSERT(BP_IS_GANG(bp)); zio_nowait(zio_read(gio, gio->io_spa, bp, gbh_abd, SPA_GANGBLOCKSIZE, zio_gang_tree_assemble_done, gn, gio->io_priority, ZIO_GANG_CHILD_FLAGS(gio), &gio->io_bookmark)); } static void zio_gang_tree_assemble_done(zio_t *zio) { zio_t *gio = zio->io_gang_leader; zio_gang_node_t *gn = zio->io_private; blkptr_t *bp = zio->io_bp; ASSERT(gio == zio_unique_parent(zio)); ASSERT(zio->io_child_count == 0); if (zio->io_error) return; /* this ABD was created from a linear buf in zio_gang_tree_assemble */ if (BP_SHOULD_BYTESWAP(bp)) byteswap_uint64_array(abd_to_buf(zio->io_abd), zio->io_size); ASSERT3P(abd_to_buf(zio->io_abd), ==, gn->gn_gbh); ASSERT(zio->io_size == SPA_GANGBLOCKSIZE); ASSERT(gn->gn_gbh->zg_tail.zec_magic == ZEC_MAGIC); abd_put(zio->io_abd); for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) { blkptr_t *gbp = &gn->gn_gbh->zg_blkptr[g]; if (!BP_IS_GANG(gbp)) continue; zio_gang_tree_assemble(gio, gbp, &gn->gn_child[g]); } } static void zio_gang_tree_issue(zio_t *pio, zio_gang_node_t *gn, blkptr_t *bp, abd_t *data, uint64_t offset) { zio_t *gio = pio->io_gang_leader; zio_t *zio; ASSERT(BP_IS_GANG(bp) == !!gn); ASSERT(BP_GET_CHECKSUM(bp) == BP_GET_CHECKSUM(gio->io_bp)); ASSERT(BP_GET_LSIZE(bp) == BP_GET_PSIZE(bp) || gn == gio->io_gang_tree); /* * If you're a gang header, your data is in gn->gn_gbh. * If you're a gang member, your data is in 'data' and gn == NULL. */ zio = zio_gang_issue_func[gio->io_type](pio, bp, gn, data, offset); if (gn != NULL) { ASSERT(gn->gn_gbh->zg_tail.zec_magic == ZEC_MAGIC); for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) { blkptr_t *gbp = &gn->gn_gbh->zg_blkptr[g]; if (BP_IS_HOLE(gbp)) continue; zio_gang_tree_issue(zio, gn->gn_child[g], gbp, data, offset); offset += BP_GET_PSIZE(gbp); } } if (gn == gio->io_gang_tree) ASSERT3U(gio->io_size, ==, offset); if (zio != pio) zio_nowait(zio); } static int 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_PIPELINE_CONTINUE); } static int zio_gang_issue(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (zio_wait_for_children(zio, ZIO_CHILD_GANG, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); 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_PIPELINE_CONTINUE); } 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; ASSERTV(zio_t *gio = zio->io_gang_leader); if (BP_IS_HOLE(zio->io_bp)) return; ASSERT(BP_IS_HOLE(&zio->io_bp_orig)); ASSERT(zio->io_child_type == ZIO_CHILD_GANG); ASSERT3U(zio->io_prop.zp_copies, ==, gio->io_prop.zp_copies); ASSERT3U(zio->io_prop.zp_copies, <=, BP_GET_NDVAS(zio->io_bp)); ASSERT3U(pio->io_prop.zp_copies, <=, BP_GET_NDVAS(pio->io_bp)); ASSERT3U(BP_GET_NDVAS(zio->io_bp), <=, BP_GET_NDVAS(pio->io_bp)); mutex_enter(&pio->io_lock); for (int d = 0; d < BP_GET_NDVAS(zio->io_bp); d++) { ASSERT(DVA_GET_GANG(&pdva[d])); asize = DVA_GET_ASIZE(&pdva[d]); asize += DVA_GET_ASIZE(&cdva[d]); DVA_SET_ASIZE(&pdva[d], asize); } mutex_exit(&pio->io_lock); } static void zio_write_gang_done(zio_t *zio) { abd_put(zio->io_abd); } static int zio_write_gang_block(zio_t *pio) { spa_t *spa = pio->io_spa; metaslab_class_t *mc = spa_normal_class(spa); blkptr_t *bp = pio->io_bp; zio_t *gio = pio->io_gang_leader; zio_t *zio; zio_gang_node_t *gn, **gnpp; zio_gbh_phys_t *gbh; abd_t *gbh_abd; uint64_t txg = pio->io_txg; uint64_t resid = pio->io_size; uint64_t lsize; int copies = gio->io_prop.zp_copies; int gbh_copies; zio_prop_t zp; int error; /* * encrypted blocks need DVA[2] free so encrypted gang headers can't * have a third copy. */ gbh_copies = MIN(copies + 1, spa_max_replication(spa)); if (gio->io_prop.zp_encrypt && gbh_copies >= SPA_DVAS_PER_BP) gbh_copies = SPA_DVAS_PER_BP - 1; int flags = METASLAB_HINTBP_FAVOR | METASLAB_GANG_HEADER; if (pio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(pio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(!(pio->io_flags & ZIO_FLAG_NODATA)); flags |= METASLAB_ASYNC_ALLOC; VERIFY(refcount_held(&mc->mc_alloc_slots, pio)); /* * The logical zio has already placed a reservation for * 'copies' allocation slots but gang blocks may require * additional copies. These additional copies * (i.e. gbh_copies - copies) are guaranteed to succeed * since metaslab_class_throttle_reserve() always allows * additional reservations for gang blocks. */ VERIFY(metaslab_class_throttle_reserve(mc, gbh_copies - copies, pio, flags)); } error = metaslab_alloc(spa, mc, SPA_GANGBLOCKSIZE, bp, gbh_copies, txg, pio == gio ? NULL : gio->io_bp, flags, &pio->io_alloc_list, pio); if (error) { if (pio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(pio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(!(pio->io_flags & ZIO_FLAG_NODATA)); /* * If we failed to allocate the gang block header then * we remove any additional allocation reservations that * we placed here. The original reservation will * be removed when the logical I/O goes to the ready * stage. */ metaslab_class_throttle_unreserve(mc, gbh_copies - copies, pio); } pio->io_error = error; return (ZIO_PIPELINE_CONTINUE); } if (pio == gio) { gnpp = &gio->io_gang_tree; } else { gnpp = pio->io_private; ASSERT(pio->io_ready == zio_write_gang_member_ready); } gn = zio_gang_node_alloc(gnpp); gbh = gn->gn_gbh; bzero(gbh, SPA_GANGBLOCKSIZE); gbh_abd = abd_get_from_buf(gbh, SPA_GANGBLOCKSIZE); /* * Create the gang header. */ zio = zio_rewrite(pio, spa, txg, bp, gbh_abd, SPA_GANGBLOCKSIZE, zio_write_gang_done, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); /* * Create and nowait the gang children. */ for (int g = 0; resid != 0; resid -= lsize, g++) { lsize = P2ROUNDUP(resid / (SPA_GBH_NBLKPTRS - g), SPA_MINBLOCKSIZE); ASSERT(lsize >= SPA_MINBLOCKSIZE && lsize <= resid); zp.zp_checksum = gio->io_prop.zp_checksum; zp.zp_compress = ZIO_COMPRESS_OFF; zp.zp_type = DMU_OT_NONE; zp.zp_level = 0; zp.zp_copies = gio->io_prop.zp_copies; zp.zp_dedup = B_FALSE; zp.zp_dedup_verify = B_FALSE; zp.zp_nopwrite = B_FALSE; zp.zp_encrypt = gio->io_prop.zp_encrypt; zp.zp_byteorder = gio->io_prop.zp_byteorder; bzero(zp.zp_salt, ZIO_DATA_SALT_LEN); bzero(zp.zp_iv, ZIO_DATA_IV_LEN); bzero(zp.zp_mac, ZIO_DATA_MAC_LEN); zio_t *cio = zio_write(zio, spa, txg, &gbh->zg_blkptr[g], abd_get_offset(pio->io_abd, pio->io_size - resid), lsize, lsize, &zp, zio_write_gang_member_ready, NULL, NULL, zio_write_gang_done, &gn->gn_child[g], pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); if (pio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(pio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(!(pio->io_flags & ZIO_FLAG_NODATA)); /* * Gang children won't throttle but we should * account for their work, so reserve an allocation * slot for them here. */ VERIFY(metaslab_class_throttle_reserve(mc, zp.zp_copies, cio, flags)); } zio_nowait(cio); } /* * Set pio's pipeline to just wait for zio to finish. */ pio->io_pipeline = ZIO_INTERLOCK_PIPELINE; /* * We didn't allocate this bp, so make sure it doesn't get unmarked. */ pio->io_flags &= ~ZIO_FLAG_FASTWRITE; zio_nowait(zio); return (ZIO_PIPELINE_CONTINUE); } /* * 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 int zio_nop_write(zio_t *zio) { blkptr_t *bp = zio->io_bp; blkptr_t *bp_orig = &zio->io_bp_orig; zio_prop_t *zp = &zio->io_prop; ASSERT(BP_GET_LEVEL(bp) == 0); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); ASSERT(zp->zp_nopwrite); ASSERT(!zp->zp_dedup); ASSERT(zio->io_bp_override == NULL); ASSERT(IO_IS_ALLOCATING(zio)); /* * Check to see if the original bp and the new bp have matching * characteristics (i.e. same checksum, compression algorithms, etc). * If they don't then just continue with the pipeline which will * allocate a new bp. */ if (BP_IS_HOLE(bp_orig) || !(zio_checksum_table[BP_GET_CHECKSUM(bp)].ci_flags & ZCHECKSUM_FLAG_NOPWRITE) || BP_IS_ENCRYPTED(bp) || BP_IS_ENCRYPTED(bp_orig) || BP_GET_CHECKSUM(bp) != BP_GET_CHECKSUM(bp_orig) || BP_GET_COMPRESS(bp) != BP_GET_COMPRESS(bp_orig) || BP_GET_DEDUP(bp) != BP_GET_DEDUP(bp_orig) || zp->zp_copies != BP_GET_NDVAS(bp_orig)) return (ZIO_PIPELINE_CONTINUE); /* * If the checksums match then reset the pipeline so that we * avoid allocating a new bp and issuing any I/O. */ if (ZIO_CHECKSUM_EQUAL(bp->blk_cksum, bp_orig->blk_cksum)) { ASSERT(zio_checksum_table[zp->zp_checksum].ci_flags & ZCHECKSUM_FLAG_NOPWRITE); ASSERT3U(BP_GET_PSIZE(bp), ==, BP_GET_PSIZE(bp_orig)); ASSERT3U(BP_GET_LSIZE(bp), ==, BP_GET_LSIZE(bp_orig)); ASSERT(zp->zp_compress != ZIO_COMPRESS_OFF); ASSERT(bcmp(&bp->blk_prop, &bp_orig->blk_prop, sizeof (uint64_t)) == 0); *bp = *bp_orig; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; zio->io_flags |= ZIO_FLAG_NOPWRITE; } return (ZIO_PIPELINE_CONTINUE); } /* * ========================================================================== * Dedup * ========================================================================== */ static void zio_ddt_child_read_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp; zio_t *pio = zio_unique_parent(zio); mutex_enter(&pio->io_lock); ddp = ddt_phys_select(dde, bp); if (zio->io_error == 0) ddt_phys_clear(ddp); /* this ddp doesn't need repair */ if (zio->io_error == 0 && dde->dde_repair_abd == NULL) dde->dde_repair_abd = zio->io_abd; else abd_free(zio->io_abd); mutex_exit(&pio->io_lock); } static int zio_ddt_read_start(zio_t *zio) { blkptr_t *bp = zio->io_bp; ASSERT(BP_GET_DEDUP(bp)); ASSERT(BP_GET_PSIZE(bp) == zio->io_size); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); if (zio->io_child_error[ZIO_CHILD_DDT]) { ddt_t *ddt = ddt_select(zio->io_spa, bp); ddt_entry_t *dde = ddt_repair_start(ddt, bp); ddt_phys_t *ddp = dde->dde_phys; ddt_phys_t *ddp_self = ddt_phys_select(dde, bp); blkptr_t blk; ASSERT(zio->io_vsd == NULL); zio->io_vsd = dde; if (ddp_self == NULL) return (ZIO_PIPELINE_CONTINUE); for (int p = 0; p < DDT_PHYS_TYPES; p++, ddp++) { if (ddp->ddp_phys_birth == 0 || ddp == ddp_self) continue; ddt_bp_create(ddt->ddt_checksum, &dde->dde_key, ddp, &blk); zio_nowait(zio_read(zio, zio->io_spa, &blk, abd_alloc_for_io(zio->io_size, B_TRUE), zio->io_size, zio_ddt_child_read_done, dde, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio) | ZIO_FLAG_DONT_PROPAGATE, &zio->io_bookmark)); } return (ZIO_PIPELINE_CONTINUE); } 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_PIPELINE_CONTINUE); } static int zio_ddt_read_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (zio_wait_for_children(zio, ZIO_CHILD_DDT, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); 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_PIPELINE_CONTINUE); } if (dde == NULL) { zio->io_stage = ZIO_STAGE_DDT_READ_START >> 1; zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_FALSE); return (ZIO_PIPELINE_STOP); } if (dde->dde_repair_abd != NULL) { abd_copy(zio->io_abd, dde->dde_repair_abd, zio->io_size); zio->io_child_error[ZIO_CHILD_DDT] = 0; } ddt_repair_done(ddt, dde); zio->io_vsd = NULL; } ASSERT(zio->io_vsd == NULL); return (ZIO_PIPELINE_CONTINUE); } static boolean_t zio_ddt_collision(zio_t *zio, ddt_t *ddt, ddt_entry_t *dde) { spa_t *spa = zio->io_spa; boolean_t do_raw = !!(zio->io_flags & ZIO_FLAG_RAW); ASSERT(!(zio->io_bp_override && do_raw)); /* * Note: we compare the original data, not the transformed data, * because when zio->io_bp is an override bp, we will not have * pushed the I/O transforms. That's an important optimization * because otherwise we'd compress/encrypt all dmu_sync() data twice. * However, we should never get a raw, override zio so in these * cases we can compare the io_abd directly. This is useful because * it allows us to do dedup verification even if we don't have access * to the original data (for instance, if the encryption keys aren't * loaded). */ for (int p = DDT_PHYS_SINGLE; p <= DDT_PHYS_TRIPLE; p++) { zio_t *lio = dde->dde_lead_zio[p]; if (lio != NULL && do_raw) { return (lio->io_size != zio->io_size || abd_cmp(zio->io_abd, lio->io_abd) != 0); } else if (lio != NULL) { return (lio->io_orig_size != zio->io_orig_size || abd_cmp(zio->io_orig_abd, lio->io_orig_abd) != 0); } } for (int p = DDT_PHYS_SINGLE; p <= DDT_PHYS_TRIPLE; p++) { ddt_phys_t *ddp = &dde->dde_phys[p]; if (ddp->ddp_phys_birth != 0 && do_raw) { blkptr_t blk = *zio->io_bp; uint64_t psize; abd_t *tmpabd; int error; ddt_bp_fill(ddp, &blk, ddp->ddp_phys_birth); psize = BP_GET_PSIZE(&blk); if (psize != zio->io_size) return (B_TRUE); ddt_exit(ddt); tmpabd = abd_alloc_for_io(psize, B_TRUE); error = zio_wait(zio_read(NULL, spa, &blk, tmpabd, psize, NULL, NULL, ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_RAW, &zio->io_bookmark)); if (error == 0) { if (abd_cmp(tmpabd, zio->io_abd) != 0) error = SET_ERROR(ENOENT); } abd_free(tmpabd); ddt_enter(ddt); return (error != 0); } else if (ddp->ddp_phys_birth != 0) { arc_buf_t *abuf = NULL; arc_flags_t aflags = ARC_FLAG_WAIT; blkptr_t blk = *zio->io_bp; int error; ddt_bp_fill(ddp, &blk, ddp->ddp_phys_birth); if (BP_GET_LSIZE(&blk) != zio->io_orig_size) return (B_TRUE); ddt_exit(ddt); error = arc_read(NULL, spa, &blk, arc_getbuf_func, &abuf, ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &aflags, &zio->io_bookmark); if (error == 0) { if (abd_cmp_buf(zio->io_orig_abd, abuf->b_data, zio->io_orig_size) != 0) error = SET_ERROR(ENOENT); arc_buf_destroy(abuf, &abuf); } ddt_enter(ddt); return (error != 0); } } return (B_FALSE); } static void zio_ddt_child_write_ready(zio_t *zio) { int p = zio->io_prop.zp_copies; ddt_t *ddt = ddt_select(zio->io_spa, zio->io_bp); ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp = &dde->dde_phys[p]; zio_t *pio; if (zio->io_error) return; ddt_enter(ddt); ASSERT(dde->dde_lead_zio[p] == zio); ddt_phys_fill(ddp, zio->io_bp); zio_link_t *zl = NULL; while ((pio = zio_walk_parents(zio, &zl)) != NULL) ddt_bp_fill(ddp, pio->io_bp, zio->io_txg); ddt_exit(ddt); } static void zio_ddt_child_write_done(zio_t *zio) { int p = zio->io_prop.zp_copies; ddt_t *ddt = ddt_select(zio->io_spa, zio->io_bp); ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp = &dde->dde_phys[p]; ddt_enter(ddt); ASSERT(ddp->ddp_refcnt == 0); ASSERT(dde->dde_lead_zio[p] == zio); dde->dde_lead_zio[p] = NULL; if (zio->io_error == 0) { zio_link_t *zl = NULL; while (zio_walk_parents(zio, &zl) != NULL) ddt_phys_addref(ddp); } else { ddt_phys_clear(ddp); } ddt_exit(ddt); } static void zio_ddt_ditto_write_done(zio_t *zio) { int p = DDT_PHYS_DITTO; ASSERTV(zio_prop_t *zp = &zio->io_prop); blkptr_t *bp = zio->io_bp; ddt_t *ddt = ddt_select(zio->io_spa, bp); ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp = &dde->dde_phys[p]; ddt_key_t *ddk = &dde->dde_key; ddt_enter(ddt); ASSERT(ddp->ddp_refcnt == 0); ASSERT(dde->dde_lead_zio[p] == zio); dde->dde_lead_zio[p] = NULL; if (zio->io_error == 0) { ASSERT(ZIO_CHECKSUM_EQUAL(bp->blk_cksum, ddk->ddk_cksum)); ASSERT(zp->zp_copies < SPA_DVAS_PER_BP); ASSERT(zp->zp_copies == BP_GET_NDVAS(bp) - BP_IS_GANG(bp)); if (ddp->ddp_phys_birth != 0) ddt_phys_free(ddt, ddk, ddp, zio->io_txg); ddt_phys_fill(ddp, bp); } ddt_exit(ddt); } static int zio_ddt_write(zio_t *zio) { spa_t *spa = zio->io_spa; blkptr_t *bp = zio->io_bp; uint64_t txg = zio->io_txg; zio_prop_t *zp = &zio->io_prop; int p = zp->zp_copies; int ditto_copies; zio_t *cio = NULL; zio_t *dio = NULL; ddt_t *ddt = ddt_select(spa, bp); ddt_entry_t *dde; ddt_phys_t *ddp; ASSERT(BP_GET_DEDUP(bp)); ASSERT(BP_GET_CHECKSUM(bp) == zp->zp_checksum); ASSERT(BP_IS_HOLE(bp) || zio->io_bp_override); ASSERT(!(zio->io_bp_override && (zio->io_flags & ZIO_FLAG_RAW))); ddt_enter(ddt); dde = ddt_lookup(ddt, bp, B_TRUE); ddp = &dde->dde_phys[p]; if (zp->zp_dedup_verify && zio_ddt_collision(zio, ddt, dde)) { /* * If we're using a weak checksum, upgrade to a strong checksum * and try again. If we're already using a strong checksum, * we can't resolve it, so just convert to an ordinary write. * (And automatically e-mail a paper to Nature?) */ if (!(zio_checksum_table[zp->zp_checksum].ci_flags & ZCHECKSUM_FLAG_DEDUP)) { zp->zp_checksum = spa_dedup_checksum(spa); zio_pop_transforms(zio); zio->io_stage = ZIO_STAGE_OPEN; BP_ZERO(bp); } else { zp->zp_dedup = B_FALSE; } zio->io_pipeline = ZIO_WRITE_PIPELINE; ddt_exit(ddt); return (ZIO_PIPELINE_CONTINUE); } ditto_copies = ddt_ditto_copies_needed(ddt, dde, ddp); ASSERT(ditto_copies < SPA_DVAS_PER_BP); if (ditto_copies > ddt_ditto_copies_present(dde) && dde->dde_lead_zio[DDT_PHYS_DITTO] == NULL) { zio_prop_t czp = *zp; czp.zp_copies = ditto_copies; /* * If we arrived here with an override bp, we won't have run * the transform stack, so we won't have the data we need to * generate a child i/o. So, toss the override bp and restart. * This is safe, because using the override bp is just an * optimization; and it's rare, so the cost doesn't matter. */ if (zio->io_bp_override) { zio_pop_transforms(zio); zio->io_stage = ZIO_STAGE_OPEN; zio->io_pipeline = ZIO_WRITE_PIPELINE; zio->io_bp_override = NULL; BP_ZERO(bp); ddt_exit(ddt); return (ZIO_PIPELINE_CONTINUE); } dio = zio_write(zio, spa, txg, bp, zio->io_orig_abd, zio->io_orig_size, zio->io_orig_size, &czp, NULL, NULL, NULL, zio_ddt_ditto_write_done, dde, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark); zio_push_transform(dio, zio->io_abd, zio->io_size, 0, NULL); dde->dde_lead_zio[DDT_PHYS_DITTO] = dio; } if (ddp->ddp_phys_birth != 0 || dde->dde_lead_zio[p] != NULL) { if (ddp->ddp_phys_birth != 0) ddt_bp_fill(ddp, bp, txg); if (dde->dde_lead_zio[p] != NULL) zio_add_child(zio, dde->dde_lead_zio[p]); else ddt_phys_addref(ddp); } else if (zio->io_bp_override) { ASSERT(bp->blk_birth == txg); ASSERT(BP_EQUAL(bp, zio->io_bp_override)); ddt_phys_fill(ddp, bp); ddt_phys_addref(ddp); } else { cio = zio_write(zio, spa, txg, bp, zio->io_orig_abd, zio->io_orig_size, zio->io_orig_size, zp, zio_ddt_child_write_ready, NULL, NULL, zio_ddt_child_write_done, dde, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark); zio_push_transform(cio, zio->io_abd, zio->io_size, 0, NULL); dde->dde_lead_zio[p] = cio; } ddt_exit(ddt); if (cio) zio_nowait(cio); if (dio) zio_nowait(dio); return (ZIO_PIPELINE_CONTINUE); } ddt_entry_t *freedde; /* for debugging */ static int zio_ddt_free(zio_t *zio) { spa_t *spa = zio->io_spa; blkptr_t *bp = zio->io_bp; ddt_t *ddt = ddt_select(spa, bp); ddt_entry_t *dde; ddt_phys_t *ddp; ASSERT(BP_GET_DEDUP(bp)); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ddt_enter(ddt); freedde = dde = ddt_lookup(ddt, bp, B_TRUE); if (dde) { ddp = ddt_phys_select(dde, bp); if (ddp) ddt_phys_decref(ddp); } ddt_exit(ddt); return (ZIO_PIPELINE_CONTINUE); } /* * ========================================================================== * Allocate and free blocks * ========================================================================== */ static zio_t * zio_io_to_allocate(spa_t *spa) { zio_t *zio; ASSERT(MUTEX_HELD(&spa->spa_alloc_lock)); zio = avl_first(&spa->spa_alloc_tree); if (zio == NULL) return (NULL); ASSERT(IO_IS_ALLOCATING(zio)); /* * Try to place a reservation for this zio. If we're unable to * reserve then we throttle. */ if (!metaslab_class_throttle_reserve(spa_normal_class(spa), zio->io_prop.zp_copies, zio, 0)) { return (NULL); } avl_remove(&spa->spa_alloc_tree, zio); ASSERT3U(zio->io_stage, <, ZIO_STAGE_DVA_ALLOCATE); return (zio); } static int zio_dva_throttle(zio_t *zio) { spa_t *spa = zio->io_spa; zio_t *nio; if (zio->io_priority == ZIO_PRIORITY_SYNC_WRITE || !spa_normal_class(zio->io_spa)->mc_alloc_throttle_enabled || zio->io_child_type == ZIO_CHILD_GANG || zio->io_flags & ZIO_FLAG_NODATA) { return (ZIO_PIPELINE_CONTINUE); } ASSERT(zio->io_child_type > ZIO_CHILD_GANG); ASSERT3U(zio->io_queued_timestamp, >, 0); ASSERT(zio->io_stage == ZIO_STAGE_DVA_THROTTLE); mutex_enter(&spa->spa_alloc_lock); ASSERT(zio->io_type == ZIO_TYPE_WRITE); avl_add(&spa->spa_alloc_tree, zio); nio = zio_io_to_allocate(zio->io_spa); mutex_exit(&spa->spa_alloc_lock); if (nio == zio) return (ZIO_PIPELINE_CONTINUE); if (nio != NULL) { ASSERT(nio->io_stage == ZIO_STAGE_DVA_THROTTLE); /* * We are passing control to a new zio so make sure that * it is processed by a different thread. We do this to * avoid stack overflows that can occur when parents are * throttled and children are making progress. We allow * it to go to the head of the taskq since it's already * been waiting. */ zio_taskq_dispatch(nio, ZIO_TASKQ_ISSUE, B_TRUE); } return (ZIO_PIPELINE_STOP); } void zio_allocate_dispatch(spa_t *spa) { zio_t *zio; mutex_enter(&spa->spa_alloc_lock); zio = zio_io_to_allocate(spa); mutex_exit(&spa->spa_alloc_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); } static int zio_dva_allocate(zio_t *zio) { spa_t *spa = zio->io_spa; metaslab_class_t *mc = spa_normal_class(spa); blkptr_t *bp = zio->io_bp; int error; int flags = 0; if (zio->io_gang_leader == NULL) { ASSERT(zio->io_child_type > ZIO_CHILD_GANG); zio->io_gang_leader = zio; } ASSERT(BP_IS_HOLE(bp)); ASSERT0(BP_GET_NDVAS(bp)); ASSERT3U(zio->io_prop.zp_copies, >, 0); ASSERT3U(zio->io_prop.zp_copies, <=, spa_max_replication(spa)); ASSERT3U(zio->io_size, ==, BP_GET_PSIZE(bp)); flags |= (zio->io_flags & ZIO_FLAG_FASTWRITE) ? METASLAB_FASTWRITE : 0; if (zio->io_flags & ZIO_FLAG_NODATA) flags |= METASLAB_DONT_THROTTLE; if (zio->io_flags & ZIO_FLAG_GANG_CHILD) flags |= METASLAB_GANG_CHILD; if (zio->io_priority == ZIO_PRIORITY_ASYNC_WRITE) flags |= METASLAB_ASYNC_ALLOC; error = metaslab_alloc(spa, mc, zio->io_size, bp, zio->io_prop.zp_copies, zio->io_txg, NULL, flags, &zio->io_alloc_list, zio); if (error != 0) { spa_dbgmsg(spa, "%s: metaslab allocation failure: zio %p, " "size %llu, error %d", spa_name(spa), zio, zio->io_size, error); if (error == ENOSPC && zio->io_size > SPA_MINBLOCKSIZE) return (zio_write_gang_block(zio)); zio->io_error = error; } return (ZIO_PIPELINE_CONTINUE); } static int zio_dva_free(zio_t *zio) { metaslab_free(zio->io_spa, zio->io_bp, zio->io_txg, B_FALSE); return (ZIO_PIPELINE_CONTINUE); } static int 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_PIPELINE_CONTINUE); } /* * Undo an allocation. This is used by zio_done() when an I/O fails * and we want to give back the block we just allocated. * This handles both normal blocks and gang blocks. */ static void zio_dva_unallocate(zio_t *zio, zio_gang_node_t *gn, blkptr_t *bp) { ASSERT(bp->blk_birth == zio->io_txg || BP_IS_HOLE(bp)); ASSERT(zio->io_bp_override == NULL); if (!BP_IS_HOLE(bp)) metaslab_free(zio->io_spa, bp, bp->blk_birth, B_TRUE); if (gn != NULL) { for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) { zio_dva_unallocate(zio, gn->gn_child[g], &gn->gn_gbh->zg_blkptr[g]); } } } /* * Try to allocate an intent log block. Return 0 on success, errno on failure. */ int zio_alloc_zil(spa_t *spa, objset_t *os, uint64_t txg, blkptr_t *new_bp, uint64_t size, boolean_t *slog) { int error = 1; zio_alloc_list_t io_alloc_list; ASSERT(txg > spa_syncing_txg(spa)); metaslab_trace_init(&io_alloc_list); error = metaslab_alloc(spa, spa_log_class(spa), size, new_bp, 1, txg, NULL, METASLAB_FASTWRITE, &io_alloc_list, NULL); if (error == 0) { *slog = TRUE; } else { error = metaslab_alloc(spa, spa_normal_class(spa), size, new_bp, 1, txg, NULL, METASLAB_FASTWRITE, &io_alloc_list, NULL); if (error == 0) *slog = FALSE; } metaslab_trace_fini(&io_alloc_list); if (error == 0) { BP_SET_LSIZE(new_bp, size); BP_SET_PSIZE(new_bp, size); BP_SET_COMPRESS(new_bp, ZIO_COMPRESS_OFF); BP_SET_CHECKSUM(new_bp, spa_version(spa) >= SPA_VERSION_SLIM_ZIL ? ZIO_CHECKSUM_ZILOG2 : ZIO_CHECKSUM_ZILOG); BP_SET_TYPE(new_bp, DMU_OT_INTENT_LOG); BP_SET_LEVEL(new_bp, 0); BP_SET_DEDUP(new_bp, 0); BP_SET_BYTEORDER(new_bp, ZFS_HOST_BYTEORDER); /* * encrypted blocks will require an IV and salt. We generate * these now since we will not be rewriting the bp at * rewrite time. */ if (os->os_encrypted) { uint8_t iv[ZIO_DATA_IV_LEN]; uint8_t salt[ZIO_DATA_SALT_LEN]; BP_SET_CRYPT(new_bp, B_TRUE); VERIFY0(spa_crypt_get_salt(spa, dmu_objset_id(os), salt)); VERIFY0(zio_crypt_generate_iv(iv)); zio_crypt_encode_params_bp(new_bp, salt, iv); } + } else { + zfs_dbgmsg("%s: zil block allocation failure: " + "size %llu, error %d", spa_name(spa), size, error); } return (error); } /* * Free an intent log block. */ void zio_free_zil(spa_t *spa, uint64_t txg, blkptr_t *bp) { ASSERT(BP_GET_TYPE(bp) == DMU_OT_INTENT_LOG); ASSERT(!BP_IS_GANG(bp)); zio_free(spa, txg, bp); } /* * ========================================================================== * 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 int 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 (ZIO_PIPELINE_STOP); } ASSERT3P(zio->io_logical, !=, zio); 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 with nested replication. * For example, given a mirror of mirrors, (A+B)+(C+D), if only * A is out of date, we'll read from C+D, 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. * The same logic applies to any form of nested replication: * ditto + mirror, RAID-Z + replacing, etc. This covers them all. */ if ((zio->io_flags & ZIO_FLAG_IO_REPAIR) && !(zio->io_flags & ZIO_FLAG_SELF_HEAL) && zio->io_txg != 0 && /* not a delegated i/o */ !vdev_dtl_contains(vd, DTL_PARTIAL, zio->io_txg, 1)) { ASSERT(zio->io_type == ZIO_TYPE_WRITE); zio_vdev_io_bypass(zio); return (ZIO_PIPELINE_CONTINUE); } if (vd->vdev_ops->vdev_op_leaf && (zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE)) { if (zio->io_type == ZIO_TYPE_READ && vdev_cache_read(zio)) return (ZIO_PIPELINE_CONTINUE); if ((zio = vdev_queue_io(zio)) == NULL) return (ZIO_PIPELINE_STOP); if (!vdev_accessible(vd, zio)) { zio->io_error = SET_ERROR(ENXIO); zio_interrupt(zio); return (ZIO_PIPELINE_STOP); } zio->io_delay = gethrtime(); } vd->vdev_ops->vdev_op_io_start(zio); return (ZIO_PIPELINE_STOP); } static int 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, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); ASSERT(zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE); if (zio->io_delay) zio->io_delay = gethrtime() - zio->io_delay; if (vd != NULL && vd->vdev_ops->vdev_op_leaf) { vdev_queue_io_done(zio); if (zio->io_type == ZIO_TYPE_WRITE) vdev_cache_write(zio); if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_device_injections(vd, zio, EIO, EILSEQ); if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_label_injection(zio, EIO); if (zio->io_error) { if (!vdev_accessible(vd, zio)) { zio->io_error = SET_ERROR(ENXIO); } else { unexpected_error = B_TRUE; } } } ops->vdev_op_io_done(zio); if (unexpected_error) VERIFY(vdev_probe(vd, zio) == NULL); return (ZIO_PIPELINE_CONTINUE); } /* * For non-raidz ZIOs, we can just copy aside the bad data read from the * disk, and use that to finish the checksum ereport later. */ static void zio_vsd_default_cksum_finish(zio_cksum_report_t *zcr, const abd_t *good_buf) { /* no processing needed */ zfs_ereport_finish_checksum(zcr, good_buf, zcr->zcr_cbdata, B_FALSE); } /*ARGSUSED*/ void zio_vsd_default_cksum_report(zio_t *zio, zio_cksum_report_t *zcr, void *ignored) { void *abd = abd_alloc_sametype(zio->io_abd, zio->io_size); abd_copy(abd, zio->io_abd, zio->io_size); zcr->zcr_cbinfo = zio->io_size; zcr->zcr_cbdata = abd; zcr->zcr_finish = zio_vsd_default_cksum_finish; zcr->zcr_free = zio_abd_free; } static int zio_vdev_io_assess(zio_t *zio) { vdev_t *vd = zio->io_vd; if (zio_wait_for_children(zio, ZIO_CHILD_VDEV, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); if (vd == NULL && !(zio->io_flags & ZIO_FLAG_CONFIG_WRITER)) spa_config_exit(zio->io_spa, SCL_ZIO, zio); if (zio->io_vsd != NULL) { zio->io_vsd_ops->vsd_free(zio); zio->io_vsd = NULL; } if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_fault_injection(zio, EIO); /* * If the I/O failed, determine whether we should attempt to retry it. * * On retry, we cut in line in the issue queue, since we don't want * compression/checksumming/etc. work to prevent our (cheap) IO reissue. */ if (zio->io_error && vd == NULL && !(zio->io_flags & (ZIO_FLAG_DONT_RETRY | ZIO_FLAG_IO_RETRY))) { ASSERT(!(zio->io_flags & ZIO_FLAG_DONT_QUEUE)); /* not a leaf */ ASSERT(!(zio->io_flags & ZIO_FLAG_IO_BYPASS)); /* not a leaf */ zio->io_error = 0; zio->io_flags |= ZIO_FLAG_IO_RETRY | ZIO_FLAG_DONT_CACHE | ZIO_FLAG_DONT_AGGREGATE; zio->io_stage = ZIO_STAGE_VDEV_IO_START >> 1; zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, zio_requeue_io_start_cut_in_line); return (ZIO_PIPELINE_STOP); } /* * If we got an error on a leaf device, convert it to ENXIO * if the device is not accessible at all. */ if (zio->io_error && vd != NULL && vd->vdev_ops->vdev_op_leaf && !vdev_accessible(vd, zio)) zio->io_error = SET_ERROR(ENXIO); /* * If we can't write to an interior vdev (mirror or RAID-Z), * set vdev_cant_write so that we stop trying to allocate from it. */ if (zio->io_error == ENXIO && zio->io_type == ZIO_TYPE_WRITE && vd != NULL && !vd->vdev_ops->vdev_op_leaf) { vd->vdev_cant_write = B_TRUE; } /* * If a cache flush returns ENOTSUP or ENOTTY, we know that no future * attempts will ever succeed. In this case we set a persistent bit so * that we don't bother with it in the future. */ if ((zio->io_error == ENOTSUP || zio->io_error == ENOTTY) && zio->io_type == ZIO_TYPE_IOCTL && zio->io_cmd == DKIOCFLUSHWRITECACHE && vd != NULL) vd->vdev_nowritecache = B_TRUE; if (zio->io_error) zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (vd != NULL && vd->vdev_ops->vdev_op_leaf && zio->io_physdone != NULL) { ASSERT(!(zio->io_flags & ZIO_FLAG_DELEGATED)); ASSERT(zio->io_child_type == ZIO_CHILD_VDEV); zio->io_physdone(zio->io_logical); } return (ZIO_PIPELINE_CONTINUE); } 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 int 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); 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_PIPELINE_CONTINUE); /* only ZIL blocks are re-encrypted on rewrite */ if (!IO_IS_ALLOCATING(zio) && ot != DMU_OT_INTENT_LOG) return (ZIO_PIPELINE_CONTINUE); if (!(zp->zp_encrypt || BP_IS_ENCRYPTED(bp))) { BP_SET_CRYPT(bp, B_FALSE); return (ZIO_PIPELINE_CONTINUE); } /* if we are doing raw encryption set the provided encryption params */ if (zio->io_flags & ZIO_FLAG_RAW_ENCRYPT) { 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); if (DMU_OT_IS_ENCRYPTED(ot)) zio_crypt_encode_params_bp(bp, zp->zp_salt, zp->zp_iv); return (ZIO_PIPELINE_CONTINUE); } /* 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_PIPELINE_CONTINUE); } /* * 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, zio->io_bookmark.zb_objset, zio->io_abd, psize, BP_SHOULD_BYTESWAP(bp))); return (ZIO_PIPELINE_CONTINUE); } /* 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, zio->io_bookmark.zb_objset, zio->io_abd, psize, mac)); zio_crypt_encode_mac_bp(bp, mac); return (ZIO_PIPELINE_CONTINUE); } /* * 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 consitutes 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.zb_objset, bp, zio->io_txg, psize, zio->io_abd, eabd, iv, mac, salt, &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_PIPELINE_CONTINUE); } /* * ========================================================================== * Generate and verify checksums * ========================================================================== */ static int 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_PIPELINE_CONTINUE); 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_PIPELINE_CONTINUE); } static int 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_PIPELINE_CONTINUE); ASSERT(zio->io_prop.zp_checksum == ZIO_CHECKSUM_LABEL); } if ((error = zio_checksum_error(zio, &info)) != 0) { zio->io_error = error; if (error == ECKSUM && !(zio->io_flags & ZIO_FLAG_SPECULATIVE)) { zfs_ereport_start_checksum(zio->io_spa, zio->io_vd, &zio->io_bookmark, zio, zio->io_offset, zio->io_size, NULL, &info); } } return (ZIO_PIPELINE_CONTINUE); } /* * Called by RAID-Z to ensure we don't compute the checksum twice. */ void zio_checksum_verified(zio_t *zio) { zio->io_pipeline &= ~ZIO_STAGE_CHECKSUM_VERIFY; } /* * ========================================================================== * Error rank. Error are ranked in the order 0, ENXIO, ECKSUM, EIO, other. * An error of 0 indicates success. ENXIO indicates whole-device failure, * which may be transient (e.g. unplugged) or permament. 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 int zio_ready(zio_t *zio) { blkptr_t *bp = zio->io_bp; zio_t *pio, *pio_next; zio_link_t *zl = NULL; if (zio_wait_for_children(zio, ZIO_CHILD_GANG, ZIO_WAIT_READY) || zio_wait_for_children(zio, ZIO_CHILD_DDT, ZIO_WAIT_READY)) return (ZIO_PIPELINE_STOP); if (zio->io_ready) { ASSERT(IO_IS_ALLOCATING(zio)); ASSERT(bp->blk_birth == zio->io_txg || BP_IS_HOLE(bp) || (zio->io_flags & ZIO_FLAG_NOPWRITE)); ASSERT(zio->io_children[ZIO_CHILD_GANG][ZIO_WAIT_READY] == 0); zio->io_ready(zio); } if (bp != NULL && bp != &zio->io_bp_copy) zio->io_bp_copy = *bp; if (zio->io_error != 0) { zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (zio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(IO_IS_ALLOCATING(zio)); ASSERT(zio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); /* * We were unable to allocate anything, unreserve and * issue the next I/O to allocate. */ metaslab_class_throttle_unreserve( spa_normal_class(zio->io_spa), zio->io_prop.zp_copies, zio); zio_allocate_dispatch(zio->io_spa); } } 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); } if (zio->io_flags & ZIO_FLAG_NODATA) { if (BP_IS_GANG(bp)) { zio->io_flags &= ~ZIO_FLAG_NODATA; } else { ASSERT((uintptr_t)zio->io_abd < SPA_MAXBLOCKSIZE); zio->io_pipeline &= ~ZIO_VDEV_IO_STAGES; } } if (zio_injection_enabled && zio->io_spa->spa_syncing_txg == zio->io_txg) zio_handle_ignored_writes(zio); return (ZIO_PIPELINE_CONTINUE); } /* * Update the allocation throttle accounting. */ static void zio_dva_throttle_done(zio_t *zio) { ASSERTV(zio_t *lio = zio->io_logical); zio_t *pio = zio_unique_parent(zio); vdev_t *vd = zio->io_vd; int flags = METASLAB_ASYNC_ALLOC; ASSERT3P(zio->io_bp, !=, NULL); ASSERT3U(zio->io_type, ==, ZIO_TYPE_WRITE); ASSERT3U(zio->io_priority, ==, ZIO_PRIORITY_ASYNC_WRITE); ASSERT3U(zio->io_child_type, ==, ZIO_CHILD_VDEV); ASSERT(vd != NULL); ASSERT3P(vd, ==, vd->vdev_top); ASSERT(zio_injection_enabled || !(zio->io_flags & ZIO_FLAG_IO_RETRY)); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REPAIR)); ASSERT(zio->io_flags & ZIO_FLAG_IO_ALLOCATING); ASSERT(!(lio->io_flags & ZIO_FLAG_IO_REWRITE)); ASSERT(!(lio->io_orig_flags & ZIO_FLAG_NODATA)); /* * Parents of gang children can have two flavors -- ones that * allocated the gang header (will have ZIO_FLAG_IO_REWRITE set) * and ones that allocated the constituent blocks. The allocation * throttle needs to know the allocating parent zio so we must find * it here. */ if (pio->io_child_type == ZIO_CHILD_GANG) { /* * If our parent is a rewrite gang child then our grandparent * would have been the one that performed the allocation. */ if (pio->io_flags & ZIO_FLAG_IO_REWRITE) pio = zio_unique_parent(pio); flags |= METASLAB_GANG_CHILD; } ASSERT(IO_IS_ALLOCATING(pio)); ASSERT3P(zio, !=, zio->io_logical); ASSERT(zio->io_logical != NULL); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REPAIR)); ASSERT0(zio->io_flags & ZIO_FLAG_NOPWRITE); mutex_enter(&pio->io_lock); metaslab_group_alloc_decrement(zio->io_spa, vd->vdev_id, pio, flags); mutex_exit(&pio->io_lock); metaslab_class_throttle_unreserve(spa_normal_class(zio->io_spa), 1, pio); /* * Call into the pipeline to see if there is more work that * needs to be done. If there is work to be done it will be * dispatched to another taskq thread. */ zio_allocate_dispatch(zio->io_spa); } static int zio_done(zio_t *zio) { /* * Always attempt to keep stack usage minimal here since * we can be called recurisvely 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_VDEV, ZIO_WAIT_DONE) || zio_wait_for_children(zio, ZIO_CHILD_GANG, ZIO_WAIT_DONE) || zio_wait_for_children(zio, ZIO_CHILD_DDT, ZIO_WAIT_DONE) || zio_wait_for_children(zio, ZIO_CHILD_LOGICAL, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); /* * If the allocation throttle is enabled, then update the accounting. * We only track child I/Os that are part of an allocating async * write. We must do this since the allocation is performed * by the logical I/O but the actual write is done by child I/Os. */ if (zio->io_flags & ZIO_FLAG_IO_ALLOCATING && zio->io_child_type == ZIO_CHILD_VDEV) { ASSERT(spa_normal_class( zio->io_spa)->mc_alloc_throttle_enabled); zio_dva_throttle_done(zio); } /* * If the allocation throttle is enabled, verify that * we have decremented the refcounts for every I/O that was throttled. */ if (zio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(zio->io_type == ZIO_TYPE_WRITE); ASSERT(zio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(zio->io_bp != NULL); metaslab_group_alloc_verify(zio->io_spa, zio->io_bp, zio); VERIFY(refcount_not_held( &(spa_normal_class(zio->io_spa)->mc_alloc_slots), zio)); } for (int c = 0; c < ZIO_CHILD_TYPES; c++) for (int w = 0; w < ZIO_WAIT_TYPES; w++) ASSERT(zio->io_children[c][w] == 0); if (zio->io_bp != NULL && !BP_IS_EMBEDDED(zio->io_bp)) { ASSERT(zio->io_bp->blk_pad[0] == 0); ASSERT(zio->io_bp->blk_pad[1] == 0); ASSERT(bcmp(zio->io_bp, &zio->io_bp_copy, sizeof (blkptr_t)) == 0 || (zio->io_bp == zio_unique_parent(zio)->io_bp)); if (zio->io_type == ZIO_TYPE_WRITE && !BP_IS_HOLE(zio->io_bp) && zio->io_bp_override == NULL && !(zio->io_flags & ZIO_FLAG_IO_REPAIR)) { ASSERT(!BP_SHOULD_BYTESWAP(zio->io_bp)); ASSERT3U(zio->io_prop.zp_copies, <=, BP_GET_NDVAS(zio->io_bp)); ASSERT(BP_COUNT_GANG(zio->io_bp) == 0 || (BP_COUNT_GANG(zio->io_bp) == BP_GET_NDVAS(zio->io_bp))); } if (zio->io_flags & ZIO_FLAG_NOPWRITE) VERIFY(BP_EQUAL(zio->io_bp, &zio->io_bp_orig)); } /* * If there were child vdev/gang/ddt errors, they apply to us now. */ zio_inherit_child_errors(zio, ZIO_CHILD_VDEV); zio_inherit_child_errors(zio, ZIO_CHILD_GANG); zio_inherit_child_errors(zio, ZIO_CHILD_DDT); /* * If the I/O on the transformed data was successful, generate any * checksum reports now while we still have the transformed data. */ if (zio->io_error == 0) { while (zio->io_cksum_report != NULL) { zio_cksum_report_t *zcr = zio->io_cksum_report; uint64_t align = zcr->zcr_align; uint64_t asize = P2ROUNDUP(psize, align); abd_t *adata = zio->io_abd; if (asize != psize) { adata = abd_alloc(asize, B_TRUE); abd_copy(adata, zio->io_abd, psize); abd_zero_off(adata, psize, asize - psize); } zio->io_cksum_report = zcr->zcr_next; zcr->zcr_next = NULL; zcr->zcr_finish(zcr, adata); zfs_ereport_free_checksum(zcr); if (asize != psize) abd_free(adata); } } zio_pop_transforms(zio); /* note: may set zio->io_error */ vdev_stat_update(zio, psize); /* * If this I/O is attached to a particular vdev is slow, exceeding * 30 seconds to complete, post an error described the I/O delay. * We ignore these errors if the device is currently unavailable. */ if (zio->io_delay >= MSEC2NSEC(zio_delay_max)) { if (zio->io_vd != NULL && !vdev_is_dead(zio->io_vd)) zfs_ereport_post(FM_EREPORT_ZFS_DELAY, zio->io_spa, zio->io_vd, &zio->io_bookmark, zio, 0, 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)) zfs_ereport_post(FM_EREPORT_ZFS_IO, zio->io_spa, zio->io_vd, &zio->io_bookmark, zio, 0, 0); if ((zio->io_error == EIO || !(zio->io_flags & (ZIO_FLAG_SPECULATIVE | ZIO_FLAG_DONT_PROPAGATE))) && zio == zio->io_logical) { /* * For logical I/O requests, tell the SPA to log the * error and generate a logical data ereport. */ spa_log_error(zio->io_spa, &zio->io_bookmark); zfs_ereport_post(FM_EREPORT_ZFS_DATA, zio->io_spa, NULL, &zio->io_bookmark, zio, 0, 0); } } if (zio->io_error && zio == zio->io_logical) { /* * Determine whether zio should be reexecuted. This will * propagate all the way to the root via zio_notify_parent(). */ ASSERT(zio->io_vd == NULL && zio->io_bp != NULL); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); if (IO_IS_ALLOCATING(zio) && !(zio->io_flags & ZIO_FLAG_CANFAIL)) { if (zio->io_error != ENOSPC) zio->io_reexecute |= ZIO_REEXECUTE_NOW; else zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; } if ((zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_FREE) && !(zio->io_flags & ZIO_FLAG_SCAN_THREAD) && zio->io_error == ENXIO && spa_load_state(zio->io_spa) == SPA_LOAD_NONE && spa_get_failmode(zio->io_spa) != ZIO_FAILURE_MODE_CONTINUE) zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; if (!(zio->io_flags & ZIO_FLAG_CANFAIL) && !zio->io_reexecute) zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; /* * Here is a possibly good place to attempt to do * either combinatorial reconstruction or error correction * based on checksums. It also might be a good place * to send out preliminary ereports before we suspend * processing. */ } /* * If there were logical child errors, they apply to us now. * We defer this until now to avoid conflating logical child * errors with errors that happened to the zio itself when * updating vdev stats and reporting FMA events above. */ zio_inherit_child_errors(zio, ZIO_CHILD_LOGICAL); if ((zio->io_error || zio->io_reexecute) && IO_IS_ALLOCATING(zio) && zio->io_gang_leader == zio && !(zio->io_flags & (ZIO_FLAG_IO_REWRITE | ZIO_FLAG_NOPWRITE))) zio_dva_unallocate(zio, zio->io_gang_tree, zio->io_bp); zio_gang_tree_free(&zio->io_gang_tree); /* * Godfather I/Os should never suspend. */ if ((zio->io_flags & ZIO_FLAG_GODFATHER) && (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND)) zio->io_reexecute &= ~ZIO_REEXECUTE_SUSPEND; if (zio->io_reexecute) { /* * This is a logical I/O that wants to reexecute. * * Reexecute is top-down. When an i/o fails, if it's not * the root, it simply notifies its parent and sticks around. * The parent, seeing that it still has children in zio_done(), * does the same. This percolates all the way up to the root. * The root i/o will reexecute or suspend the entire tree. * * This approach ensures that zio_reexecute() honors * all the original i/o dependency relationships, e.g. * parents not executing until children are ready. */ ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); zio->io_gang_leader = NULL; mutex_enter(&zio->io_lock); zio->io_state[ZIO_WAIT_DONE] = 1; mutex_exit(&zio->io_lock); /* * "The Godfather" I/O monitors its children but is * not a true parent to them. It will track them through * the pipeline but severs its ties whenever they get into * trouble (e.g. suspended). This allows "The Godfather" * I/O to return status without blocking. */ zl = NULL; for (pio = zio_walk_parents(zio, &zl); pio != NULL; pio = pio_next) { zio_link_t *remove_zl = zl; pio_next = zio_walk_parents(zio, &zl); if ((pio->io_flags & ZIO_FLAG_GODFATHER) && (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND)) { zio_remove_child(pio, zio, remove_zl); zio_notify_parent(pio, zio, ZIO_WAIT_DONE); } } 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; zio_notify_parent(pio, zio, ZIO_WAIT_DONE); } else if (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND) { /* * We'd fail again if we reexecuted now, so suspend * until conditions improve (e.g. device comes online). */ zio_suspend(zio->io_spa, zio); } else { /* * Reexecution is potentially a huge amount of work. * Hand it off to the otherwise-unused claim taskq. */ ASSERT(taskq_empty_ent(&zio->io_tqent)); spa_taskq_dispatch_ent(zio->io_spa, ZIO_TYPE_CLAIM, ZIO_TASKQ_ISSUE, (task_func_t *)zio_reexecute, zio, 0, &zio->io_tqent); } return (ZIO_PIPELINE_STOP); } ASSERT(zio->io_child_count == 0); ASSERT(zio->io_reexecute == 0); ASSERT(zio->io_error == 0 || (zio->io_flags & ZIO_FLAG_CANFAIL)); /* * Report any checksum errors, since the I/O is complete. */ while (zio->io_cksum_report != NULL) { zio_cksum_report_t *zcr = zio->io_cksum_report; zio->io_cksum_report = zcr->zcr_next; zcr->zcr_next = NULL; zcr->zcr_finish(zcr, NULL); zfs_ereport_free_checksum(zcr); } if (zio->io_flags & ZIO_FLAG_FASTWRITE && zio->io_bp && !BP_IS_HOLE(zio->io_bp) && !BP_IS_EMBEDDED(zio->io_bp) && !(zio->io_flags & ZIO_FLAG_NOPWRITE)) { metaslab_fastwrite_unmark(zio->io_spa, zio->io_bp); } /* * It is the responsibility of the done callback to ensure that this * particular zio is no longer discoverable for adoption, and as * such, cannot acquire any new parents. */ if (zio->io_done) zio->io_done(zio); mutex_enter(&zio->io_lock); zio->io_state[ZIO_WAIT_DONE] = 1; mutex_exit(&zio->io_lock); 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); } 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 (ZIO_PIPELINE_STOP); } /* * ========================================================================== * I/O pipeline definition * ========================================================================== */ static zio_pipe_stage_t *zio_pipeline[] = { NULL, zio_read_bp_init, zio_write_bp_init, zio_free_bp_init, zio_issue_async, zio_write_compress, zio_encrypt, zio_checksum_generate, zio_nop_write, zio_ddt_read_start, zio_ddt_read_done, zio_ddt_write, zio_ddt_free, zio_gang_assemble, zio_gang_issue, zio_dva_throttle, zio_dva_allocate, zio_dva_free, zio_dva_claim, zio_ready, zio_vdev_io_start, zio_vdev_io_done, zio_vdev_io_assess, zio_checksum_verify, zio_done }; /* * Compare two zbookmark_phys_t's to see which we would reach first in a * pre-order traversal of the object tree. * * This is simple in every case aside from the meta-dnode object. For all other * objects, we traverse them in order (object 1 before object 2, and so on). * However, all of these objects are traversed while traversing object 0, since * the data it points to is the list of objects. Thus, we need to convert to a * canonical representation so we can compare meta-dnode bookmarks to * non-meta-dnode bookmarks. * * We do this by calculating "equivalents" for each field of the zbookmark. * zbookmarks outside of the meta-dnode use their own object and level, and * calculate the level 0 equivalent (the first L0 blkid that is contained in the * blocks this bookmark refers to) by multiplying their blkid by their span * (the number of L0 blocks contained within one block at their level). * zbookmarks inside the meta-dnode calculate their object equivalent * (which is L0equiv * dnodes per data block), use 0 for their L0equiv, and use * level + 1<<31 (any value larger than a level could ever be) for their level. * This causes them to always compare before a bookmark in their object * equivalent, compare appropriately to bookmarks in other objects, and to * compare appropriately to other bookmarks in the meta-dnode. */ int zbookmark_compare(uint16_t dbss1, uint8_t ibs1, uint16_t dbss2, uint8_t ibs2, const zbookmark_phys_t *zb1, const zbookmark_phys_t *zb2) { /* * These variables represent the "equivalent" values for the zbookmark, * after converting zbookmarks inside the meta dnode to their * normal-object equivalents. */ uint64_t zb1obj, zb2obj; uint64_t zb1L0, zb2L0; uint64_t zb1level, zb2level; if (zb1->zb_object == zb2->zb_object && zb1->zb_level == zb2->zb_level && zb1->zb_blkid == zb2->zb_blkid) return (0); /* * BP_SPANB calculates the span in blocks. */ zb1L0 = (zb1->zb_blkid) * BP_SPANB(ibs1, zb1->zb_level); zb2L0 = (zb2->zb_blkid) * BP_SPANB(ibs2, zb2->zb_level); if (zb1->zb_object == DMU_META_DNODE_OBJECT) { zb1obj = zb1L0 * (dbss1 << (SPA_MINBLOCKSHIFT - DNODE_SHIFT)); zb1L0 = 0; zb1level = zb1->zb_level + COMPARE_META_LEVEL; } else { zb1obj = zb1->zb_object; zb1level = zb1->zb_level; } if (zb2->zb_object == DMU_META_DNODE_OBJECT) { zb2obj = zb2L0 * (dbss2 << (SPA_MINBLOCKSHIFT - DNODE_SHIFT)); zb2L0 = 0; zb2level = zb2->zb_level + COMPARE_META_LEVEL; } else { zb2obj = zb2->zb_object; zb2level = zb2->zb_level; } /* Now that we have a canonical representation, do the comparison. */ if (zb1obj != zb2obj) return (zb1obj < zb2obj ? -1 : 1); else if (zb1L0 != zb2L0) return (zb1L0 < zb2L0 ? -1 : 1); else if (zb1level != zb2level) return (zb1level > zb2level ? -1 : 1); /* * This can (theoretically) happen if the bookmarks have the same object * and level, but different blkids, if the block sizes are not the same. * There is presently no way to change the indirect block sizes */ return (0); } /* * This function checks the following: given that last_block is the place that * our traversal stopped last time, does that guarantee that we've visited * every node under subtree_root? Therefore, we can't just use the raw output * of zbookmark_compare. We have to pass in a modified version of * subtree_root; by incrementing the block id, and then checking whether * last_block is before or equal to that, we can tell whether or not having * visited last_block implies that all of subtree_root's children have been * visited. */ boolean_t zbookmark_subtree_completed(const dnode_phys_t *dnp, const zbookmark_phys_t *subtree_root, const zbookmark_phys_t *last_block) { zbookmark_phys_t mod_zb = *subtree_root; mod_zb.zb_blkid++; ASSERT(last_block->zb_level == 0); /* The objset_phys_t isn't before anything. */ if (dnp == NULL) return (B_FALSE); /* * We pass in 1ULL << (DNODE_BLOCK_SHIFT - SPA_MINBLOCKSHIFT) for the * data block size in sectors, because that variable is only used if * the bookmark refers to a block in the meta-dnode. Since we don't * know without examining it what object it refers to, and there's no * harm in passing in this value in other cases, we always pass it in. * * We pass in 0 for the indirect block size shift because zb2 must be * level 0. The indirect block size is only used to calculate the span * of the bookmark, but since the bookmark must be level 0, the span is * always 1, so the math works out. * * If you make changes to how the zbookmark_compare code works, be sure * to make sure that this code still works afterwards. */ return (zbookmark_compare(dnp->dn_datablkszsec, dnp->dn_indblkshift, 1ULL << (DNODE_BLOCK_SHIFT - SPA_MINBLOCKSHIFT), 0, &mod_zb, last_block) <= 0); } #if defined(_KERNEL) && defined(HAVE_SPL) 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); module_param(zio_delay_max, int, 0644); MODULE_PARM_DESC(zio_delay_max, "Max zio millisec delay before posting event"); module_param(zio_requeue_io_start_cut_in_line, int, 0644); MODULE_PARM_DESC(zio_requeue_io_start_cut_in_line, "Prioritize requeued I/O"); module_param(zfs_sync_pass_deferred_free, int, 0644); MODULE_PARM_DESC(zfs_sync_pass_deferred_free, "Defer frees starting in this pass"); module_param(zfs_sync_pass_dont_compress, int, 0644); MODULE_PARM_DESC(zfs_sync_pass_dont_compress, "Don't compress starting in this pass"); module_param(zfs_sync_pass_rewrite, int, 0644); MODULE_PARM_DESC(zfs_sync_pass_rewrite, "Rewrite new bps starting in this pass"); module_param(zio_dva_throttle_enabled, int, 0644); MODULE_PARM_DESC(zio_dva_throttle_enabled, "Throttle block allocations in the ZIO pipeline"); #endif diff --git a/module/zfs/zvol.c b/module/zfs/zvol.c index 5cd190c09bdf..6ea822467b61 100644 --- a/module/zfs/zvol.c +++ b/module/zfs/zvol.c @@ -1,2746 +1,2747 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (C) 2008-2010 Lawrence Livermore National Security, LLC. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). * Rewritten for Linux by Brian Behlendorf . * LLNL-CODE-403049. * * ZFS volume emulation driver. * * Makes a DMU object look like a volume of arbitrary size, up to 2^64 bytes. * Volumes are accessed through the symbolic links named: * * /dev// * * Volumes are persistent through reboot and module load. No user command * needs to be run before opening and using a device. * * Copyright 2014 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2016 Actifio, Inc. All rights reserved. * Copyright (c) 2012, 2017 by Delphix. All rights reserved. */ /* * Note on locking of zvol state structures. * * These structures are used to maintain internal state used to emulate block * devices on top of zvols. In particular, management of device minor number * operations - create, remove, rename, and set_snapdev - involves access to * these structures. The zvol_state_lock is primarily used to protect the * zvol_state_list. The zv->zv_state_lock is used to protect the contents * of the zvol_state_t structures, as well as to make sure that when the * time comes to remove the structure from the list, it is not in use, and * therefore, it can be taken off zvol_state_list and freed. * * The zv_suspend_lock was introduced to allow for suspending I/O to a zvol, * e.g. for the duration of receive and rollback operations. This lock can be * held for significant periods of time. Given that it is undesirable to hold * mutexes for long periods of time, the following lock ordering applies: * - take zvol_state_lock if necessary, to protect zvol_state_list * - take zv_suspend_lock if necessary, by the code path in question * - take zv_state_lock to protect zvol_state_t * * The minor operations are issued to spa->spa_zvol_taskq queues, that are * single-threaded (to preserve order of minor operations), and are executed * through the zvol_task_cb that dispatches the specific operations. Therefore, * these operations are serialized per pool. Consequently, we can be certain * that for a given zvol, there is only one operation at a time in progress. * That is why one can be sure that first, zvol_state_t for a given zvol is * allocated and placed on zvol_state_list, and then other minor operations * for this zvol are going to proceed in the order of issue. * * It is also worth keeping in mind that once add_disk() is called, the zvol is * announced to the world, and zvol_open()/zvol_release() can be called at any * time. Incidentally, add_disk() itself calls zvol_open()->zvol_first_open() * and zvol_release()->zvol_last_close() directly as well. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include unsigned int zvol_inhibit_dev = 0; unsigned int zvol_major = ZVOL_MAJOR; unsigned int zvol_threads = 32; unsigned int zvol_request_sync = 0; unsigned int zvol_prefetch_bytes = (128 * 1024); unsigned long zvol_max_discard_blocks = 16384; unsigned int zvol_volmode = ZFS_VOLMODE_GEOM; static taskq_t *zvol_taskq; static kmutex_t zvol_state_lock; static list_t zvol_state_list; #define ZVOL_HT_SIZE 1024 static struct hlist_head *zvol_htable; #define ZVOL_HT_HEAD(hash) (&zvol_htable[(hash) & (ZVOL_HT_SIZE-1)]) static struct ida zvol_ida; /* * The in-core state of each volume. */ struct zvol_state { char zv_name[MAXNAMELEN]; /* name */ uint64_t zv_volsize; /* advertised space */ uint64_t zv_volblocksize; /* volume block size */ objset_t *zv_objset; /* objset handle */ uint32_t zv_flags; /* ZVOL_* flags */ uint32_t zv_open_count; /* open counts */ uint32_t zv_changed; /* disk changed */ zilog_t *zv_zilog; /* ZIL handle */ zfs_rlock_t zv_range_lock; /* range lock */ dnode_t *zv_dn; /* dnode hold */ dev_t zv_dev; /* device id */ struct gendisk *zv_disk; /* generic disk */ struct request_queue *zv_queue; /* request queue */ list_node_t zv_next; /* next zvol_state_t linkage */ uint64_t zv_hash; /* name hash */ struct hlist_node zv_hlink; /* hash link */ kmutex_t zv_state_lock; /* protects zvol_state_t */ atomic_t zv_suspend_ref; /* refcount for suspend */ krwlock_t zv_suspend_lock; /* suspend lock */ }; typedef enum { ZVOL_ASYNC_CREATE_MINORS, ZVOL_ASYNC_REMOVE_MINORS, ZVOL_ASYNC_RENAME_MINORS, ZVOL_ASYNC_SET_SNAPDEV, ZVOL_ASYNC_SET_VOLMODE, ZVOL_ASYNC_MAX } zvol_async_op_t; typedef struct { zvol_async_op_t op; char pool[MAXNAMELEN]; char name1[MAXNAMELEN]; char name2[MAXNAMELEN]; zprop_source_t source; uint64_t value; } zvol_task_t; #define ZVOL_RDONLY 0x1 static uint64_t zvol_name_hash(const char *name) { int i; uint64_t crc = -1ULL; uint8_t *p = (uint8_t *)name; ASSERT(zfs_crc64_table[128] == ZFS_CRC64_POLY); for (i = 0; i < MAXNAMELEN - 1 && *p; i++, p++) { crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ (*p)) & 0xFF]; } return (crc); } /* * Find a zvol_state_t given the full major+minor dev_t. If found, * return with zv_state_lock taken, otherwise, return (NULL) without * taking zv_state_lock. */ static zvol_state_t * zvol_find_by_dev(dev_t dev) { zvol_state_t *zv; mutex_enter(&zvol_state_lock); for (zv = list_head(&zvol_state_list); zv != NULL; zv = list_next(&zvol_state_list, zv)) { mutex_enter(&zv->zv_state_lock); if (zv->zv_dev == dev) { mutex_exit(&zvol_state_lock); return (zv); } mutex_exit(&zv->zv_state_lock); } mutex_exit(&zvol_state_lock); return (NULL); } /* * Find a zvol_state_t given the name and hash generated by zvol_name_hash. * If found, return with zv_suspend_lock and zv_state_lock taken, otherwise, * return (NULL) without the taking locks. The zv_suspend_lock is always taken * before zv_state_lock. The mode argument indicates the mode (including none) * for zv_suspend_lock to be taken. */ static zvol_state_t * zvol_find_by_name_hash(const char *name, uint64_t hash, int mode) { zvol_state_t *zv; struct hlist_node *p = NULL; mutex_enter(&zvol_state_lock); hlist_for_each(p, ZVOL_HT_HEAD(hash)) { zv = hlist_entry(p, zvol_state_t, zv_hlink); mutex_enter(&zv->zv_state_lock); if (zv->zv_hash == hash && strncmp(zv->zv_name, name, MAXNAMELEN) == 0) { /* * this is the right zvol, take the locks in the * right order */ if (mode != RW_NONE && !rw_tryenter(&zv->zv_suspend_lock, mode)) { mutex_exit(&zv->zv_state_lock); rw_enter(&zv->zv_suspend_lock, mode); mutex_enter(&zv->zv_state_lock); /* * zvol cannot be renamed as we continue * to hold zvol_state_lock */ ASSERT(zv->zv_hash == hash && strncmp(zv->zv_name, name, MAXNAMELEN) == 0); } mutex_exit(&zvol_state_lock); return (zv); } mutex_exit(&zv->zv_state_lock); } mutex_exit(&zvol_state_lock); return (NULL); } /* * Find a zvol_state_t given the name. * If found, return with zv_suspend_lock and zv_state_lock taken, otherwise, * return (NULL) without the taking locks. The zv_suspend_lock is always taken * before zv_state_lock. The mode argument indicates the mode (including none) * for zv_suspend_lock to be taken. */ static zvol_state_t * zvol_find_by_name(const char *name, int mode) { return (zvol_find_by_name_hash(name, zvol_name_hash(name), mode)); } /* * Given a path, return TRUE if path is a ZVOL. */ boolean_t zvol_is_zvol(const char *device) { struct block_device *bdev; unsigned int major; bdev = vdev_lookup_bdev(device); if (IS_ERR(bdev)) return (B_FALSE); major = MAJOR(bdev->bd_dev); bdput(bdev); if (major == zvol_major) return (B_TRUE); return (B_FALSE); } /* * ZFS_IOC_CREATE callback handles dmu zvol and zap object creation. */ void zvol_create_cb(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx) { zfs_creat_t *zct = arg; nvlist_t *nvprops = zct->zct_props; int error; uint64_t volblocksize, volsize; VERIFY(nvlist_lookup_uint64(nvprops, zfs_prop_to_name(ZFS_PROP_VOLSIZE), &volsize) == 0); if (nvlist_lookup_uint64(nvprops, zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE), &volblocksize) != 0) volblocksize = zfs_prop_default_numeric(ZFS_PROP_VOLBLOCKSIZE); /* * These properties must be removed from the list so the generic * property setting step won't apply to them. */ VERIFY(nvlist_remove_all(nvprops, zfs_prop_to_name(ZFS_PROP_VOLSIZE)) == 0); (void) nvlist_remove_all(nvprops, zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE)); error = dmu_object_claim(os, ZVOL_OBJ, DMU_OT_ZVOL, volblocksize, DMU_OT_NONE, 0, tx); ASSERT(error == 0); error = zap_create_claim(os, ZVOL_ZAP_OBJ, DMU_OT_ZVOL_PROP, DMU_OT_NONE, 0, tx); ASSERT(error == 0); error = zap_update(os, ZVOL_ZAP_OBJ, "size", 8, 1, &volsize, tx); ASSERT(error == 0); } /* * ZFS_IOC_OBJSET_STATS entry point. */ int zvol_get_stats(objset_t *os, nvlist_t *nv) { int error; dmu_object_info_t *doi; uint64_t val; error = zap_lookup(os, ZVOL_ZAP_OBJ, "size", 8, 1, &val); if (error) return (SET_ERROR(error)); dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_VOLSIZE, val); doi = kmem_alloc(sizeof (dmu_object_info_t), KM_SLEEP); error = dmu_object_info(os, ZVOL_OBJ, doi); if (error == 0) { dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_VOLBLOCKSIZE, doi->doi_data_block_size); } kmem_free(doi, sizeof (dmu_object_info_t)); return (SET_ERROR(error)); } static void zvol_size_changed(zvol_state_t *zv, uint64_t volsize) { struct block_device *bdev; ASSERT(MUTEX_HELD(&zv->zv_state_lock)); bdev = bdget_disk(zv->zv_disk, 0); if (bdev == NULL) return; set_capacity(zv->zv_disk, volsize >> 9); zv->zv_volsize = volsize; check_disk_size_change(zv->zv_disk, bdev); bdput(bdev); } /* * Sanity check volume size. */ int zvol_check_volsize(uint64_t volsize, uint64_t blocksize) { if (volsize == 0) return (SET_ERROR(EINVAL)); if (volsize % blocksize != 0) return (SET_ERROR(EINVAL)); #ifdef _ILP32 if (volsize - 1 > SPEC_MAXOFFSET_T) return (SET_ERROR(EOVERFLOW)); #endif return (0); } /* * Ensure the zap is flushed then inform the VFS of the capacity change. */ static int zvol_update_volsize(uint64_t volsize, objset_t *os) { dmu_tx_t *tx; int error; uint64_t txg; tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, ZVOL_ZAP_OBJ, TRUE, NULL); dmu_tx_mark_netfree(tx); error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); return (SET_ERROR(error)); } txg = dmu_tx_get_txg(tx); error = zap_update(os, ZVOL_ZAP_OBJ, "size", 8, 1, &volsize, tx); dmu_tx_commit(tx); txg_wait_synced(dmu_objset_pool(os), txg); if (error == 0) error = dmu_free_long_range(os, ZVOL_OBJ, volsize, DMU_OBJECT_END); return (error); } static int zvol_update_live_volsize(zvol_state_t *zv, uint64_t volsize) { zvol_size_changed(zv, volsize); /* * We should post a event here describing the expansion. However, * the zfs_ereport_post() interface doesn't nicely support posting * events for zvols, it assumes events relate to vdevs or zios. */ return (0); } /* * Set ZFS_PROP_VOLSIZE set entry point. */ int zvol_set_volsize(const char *name, uint64_t volsize) { zvol_state_t *zv = NULL; objset_t *os = NULL; int error; dmu_object_info_t *doi; uint64_t readonly; boolean_t owned = B_FALSE; error = dsl_prop_get_integer(name, zfs_prop_to_name(ZFS_PROP_READONLY), &readonly, NULL); if (error != 0) return (SET_ERROR(error)); if (readonly) return (SET_ERROR(EROFS)); zv = zvol_find_by_name(name, RW_READER); ASSERT(zv == NULL || (MUTEX_HELD(&zv->zv_state_lock) && RW_READ_HELD(&zv->zv_suspend_lock))); if (zv == NULL || zv->zv_objset == NULL) { if (zv != NULL) rw_exit(&zv->zv_suspend_lock); if ((error = dmu_objset_own(name, DMU_OST_ZVOL, B_FALSE, B_TRUE, FTAG, &os)) != 0) { if (zv != NULL) mutex_exit(&zv->zv_state_lock); return (SET_ERROR(error)); } owned = B_TRUE; if (zv != NULL) zv->zv_objset = os; } else { os = zv->zv_objset; } doi = kmem_alloc(sizeof (dmu_object_info_t), KM_SLEEP); if ((error = dmu_object_info(os, ZVOL_OBJ, doi)) || (error = zvol_check_volsize(volsize, doi->doi_data_block_size))) goto out; error = zvol_update_volsize(volsize, os); if (error == 0 && zv != NULL) error = zvol_update_live_volsize(zv, volsize); out: kmem_free(doi, sizeof (dmu_object_info_t)); if (owned) { dmu_objset_disown(os, B_TRUE, FTAG); if (zv != NULL) zv->zv_objset = NULL; } else { rw_exit(&zv->zv_suspend_lock); } if (zv != NULL) mutex_exit(&zv->zv_state_lock); return (SET_ERROR(error)); } /* * Sanity check volume block size. */ int zvol_check_volblocksize(const char *name, uint64_t volblocksize) { /* Record sizes above 128k need the feature to be enabled */ if (volblocksize > SPA_OLD_MAXBLOCKSIZE) { spa_t *spa; int error; if ((error = spa_open(name, &spa, FTAG)) != 0) return (error); if (!spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_BLOCKS)) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } /* * We don't allow setting the property above 1MB, * unless the tunable has been changed. */ if (volblocksize > zfs_max_recordsize) return (SET_ERROR(EDOM)); spa_close(spa, FTAG); } if (volblocksize < SPA_MINBLOCKSIZE || volblocksize > SPA_MAXBLOCKSIZE || !ISP2(volblocksize)) return (SET_ERROR(EDOM)); return (0); } /* * Set ZFS_PROP_VOLBLOCKSIZE set entry point. */ int zvol_set_volblocksize(const char *name, uint64_t volblocksize) { zvol_state_t *zv; dmu_tx_t *tx; int error; zv = zvol_find_by_name(name, RW_READER); if (zv == NULL) return (SET_ERROR(ENXIO)); ASSERT(MUTEX_HELD(&zv->zv_state_lock) && RW_READ_HELD(&zv->zv_suspend_lock)); if (zv->zv_flags & ZVOL_RDONLY) { mutex_exit(&zv->zv_state_lock); rw_exit(&zv->zv_suspend_lock); return (SET_ERROR(EROFS)); } tx = dmu_tx_create(zv->zv_objset); dmu_tx_hold_bonus(tx, ZVOL_OBJ); error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); } else { error = dmu_object_set_blocksize(zv->zv_objset, ZVOL_OBJ, volblocksize, 0, tx); if (error == ENOTSUP) error = SET_ERROR(EBUSY); dmu_tx_commit(tx); if (error == 0) zv->zv_volblocksize = volblocksize; } mutex_exit(&zv->zv_state_lock); rw_exit(&zv->zv_suspend_lock); return (SET_ERROR(error)); } /* * Replay a TX_TRUNCATE ZIL transaction if asked. TX_TRUNCATE is how we * implement DKIOCFREE/free-long-range. */ static int zvol_replay_truncate(void *arg1, void *arg2, boolean_t byteswap) { zvol_state_t *zv = arg1; lr_truncate_t *lr = arg2; uint64_t offset, length; if (byteswap) byteswap_uint64_array(lr, sizeof (*lr)); offset = lr->lr_offset; length = lr->lr_length; return (dmu_free_long_range(zv->zv_objset, ZVOL_OBJ, offset, length)); } /* * Replay a TX_WRITE ZIL transaction that didn't get committed * after a system failure */ static int zvol_replay_write(void *arg1, void *arg2, boolean_t byteswap) { zvol_state_t *zv = arg1; lr_write_t *lr = arg2; objset_t *os = zv->zv_objset; char *data = (char *)(lr + 1); /* data follows lr_write_t */ uint64_t offset, length; dmu_tx_t *tx; int error; if (byteswap) byteswap_uint64_array(lr, sizeof (*lr)); offset = lr->lr_offset; length = lr->lr_length; /* If it's a dmu_sync() block, write the whole block */ if (lr->lr_common.lrc_reclen == sizeof (lr_write_t)) { uint64_t blocksize = BP_GET_LSIZE(&lr->lr_blkptr); if (length < blocksize) { offset -= offset % blocksize; length = blocksize; } } tx = dmu_tx_create(os); dmu_tx_hold_write(tx, ZVOL_OBJ, offset, length); error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); } else { dmu_write(os, ZVOL_OBJ, offset, length, data, tx); dmu_tx_commit(tx); } return (error); } static int zvol_replay_err(void *arg1, void *arg2, boolean_t byteswap) { return (SET_ERROR(ENOTSUP)); } /* * Callback vectors for replaying records. * Only TX_WRITE and TX_TRUNCATE are needed for zvol. */ zil_replay_func_t *zvol_replay_vector[TX_MAX_TYPE] = { zvol_replay_err, /* no such transaction type */ zvol_replay_err, /* TX_CREATE */ zvol_replay_err, /* TX_MKDIR */ zvol_replay_err, /* TX_MKXATTR */ zvol_replay_err, /* TX_SYMLINK */ zvol_replay_err, /* TX_REMOVE */ zvol_replay_err, /* TX_RMDIR */ zvol_replay_err, /* TX_LINK */ zvol_replay_err, /* TX_RENAME */ zvol_replay_write, /* TX_WRITE */ zvol_replay_truncate, /* TX_TRUNCATE */ zvol_replay_err, /* TX_SETATTR */ zvol_replay_err, /* TX_ACL */ zvol_replay_err, /* TX_CREATE_ATTR */ zvol_replay_err, /* TX_CREATE_ACL_ATTR */ zvol_replay_err, /* TX_MKDIR_ACL */ zvol_replay_err, /* TX_MKDIR_ATTR */ zvol_replay_err, /* TX_MKDIR_ACL_ATTR */ zvol_replay_err, /* TX_WRITE2 */ }; /* * zvol_log_write() handles synchronous writes using TX_WRITE ZIL transactions. * * We store data in the log buffers if it's small enough. * Otherwise we will later flush the data out via dmu_sync(). */ ssize_t zvol_immediate_write_sz = 32768; static void zvol_log_write(zvol_state_t *zv, dmu_tx_t *tx, uint64_t offset, uint64_t size, int sync) { uint32_t blocksize = zv->zv_volblocksize; zilog_t *zilog = zv->zv_zilog; itx_wr_state_t write_state; if (zil_replaying(zilog, tx)) return; if (zilog->zl_logbias == ZFS_LOGBIAS_THROUGHPUT) write_state = WR_INDIRECT; else if (!spa_has_slogs(zilog->zl_spa) && size >= blocksize && blocksize > zvol_immediate_write_sz) write_state = WR_INDIRECT; else if (sync) write_state = WR_COPIED; else write_state = WR_NEED_COPY; while (size) { itx_t *itx; lr_write_t *lr; itx_wr_state_t wr_state = write_state; ssize_t len = size; if (wr_state == WR_COPIED && size > ZIL_MAX_COPIED_DATA) wr_state = WR_NEED_COPY; else if (wr_state == WR_INDIRECT) len = MIN(blocksize - P2PHASE(offset, blocksize), size); itx = zil_itx_create(TX_WRITE, sizeof (*lr) + (wr_state == WR_COPIED ? len : 0)); lr = (lr_write_t *)&itx->itx_lr; if (wr_state == WR_COPIED && dmu_read_by_dnode(zv->zv_dn, offset, len, lr+1, DMU_READ_NO_PREFETCH) != 0) { zil_itx_destroy(itx); itx = zil_itx_create(TX_WRITE, sizeof (*lr)); lr = (lr_write_t *)&itx->itx_lr; wr_state = WR_NEED_COPY; } itx->itx_wr_state = wr_state; lr->lr_foid = ZVOL_OBJ; lr->lr_offset = offset; lr->lr_length = len; lr->lr_blkoff = 0; BP_ZERO(&lr->lr_blkptr); itx->itx_private = zv; itx->itx_sync = sync; (void) zil_itx_assign(zilog, itx, tx); offset += len; size -= len; } } typedef struct zv_request { zvol_state_t *zv; struct bio *bio; rl_t *rl; } zv_request_t; static void uio_from_bio(uio_t *uio, struct bio *bio) { uio->uio_bvec = &bio->bi_io_vec[BIO_BI_IDX(bio)]; uio->uio_skip = BIO_BI_SKIP(bio); uio->uio_resid = BIO_BI_SIZE(bio); uio->uio_iovcnt = bio->bi_vcnt - BIO_BI_IDX(bio); uio->uio_loffset = BIO_BI_SECTOR(bio) << 9; uio->uio_limit = MAXOFFSET_T; uio->uio_segflg = UIO_BVEC; } static void zvol_write(void *arg) { zv_request_t *zvr = arg; struct bio *bio = zvr->bio; uio_t uio; zvol_state_t *zv = zvr->zv; uint64_t volsize = zv->zv_volsize; boolean_t sync; int error = 0; unsigned long start_jif; uio_from_bio(&uio, bio); ASSERT(zv && zv->zv_open_count > 0); start_jif = jiffies; blk_generic_start_io_acct(zv->zv_queue, WRITE, bio_sectors(bio), &zv->zv_disk->part0); sync = bio_is_fua(bio) || zv->zv_objset->os_sync == ZFS_SYNC_ALWAYS; while (uio.uio_resid > 0 && uio.uio_loffset < volsize) { uint64_t bytes = MIN(uio.uio_resid, DMU_MAX_ACCESS >> 1); uint64_t off = uio.uio_loffset; dmu_tx_t *tx = dmu_tx_create(zv->zv_objset); if (bytes > volsize - off) /* don't write past the end */ bytes = volsize - off; dmu_tx_hold_write(tx, ZVOL_OBJ, off, bytes); /* This will only fail for ENOSPC */ error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); break; } error = dmu_write_uio_dnode(zv->zv_dn, &uio, bytes, tx); if (error == 0) zvol_log_write(zv, tx, off, bytes, sync); dmu_tx_commit(tx); if (error) break; } zfs_range_unlock(zvr->rl); if (sync) zil_commit(zv->zv_zilog, ZVOL_OBJ); rw_exit(&zv->zv_suspend_lock); blk_generic_end_io_acct(zv->zv_queue, WRITE, &zv->zv_disk->part0, start_jif); BIO_END_IO(bio, -error); kmem_free(zvr, sizeof (zv_request_t)); } /* * Log a DKIOCFREE/free-long-range to the ZIL with TX_TRUNCATE. */ static void zvol_log_truncate(zvol_state_t *zv, dmu_tx_t *tx, uint64_t off, uint64_t len, boolean_t sync) { itx_t *itx; lr_truncate_t *lr; zilog_t *zilog = zv->zv_zilog; if (zil_replaying(zilog, tx)) return; itx = zil_itx_create(TX_TRUNCATE, sizeof (*lr)); lr = (lr_truncate_t *)&itx->itx_lr; lr->lr_foid = ZVOL_OBJ; lr->lr_offset = off; lr->lr_length = len; itx->itx_sync = sync; zil_itx_assign(zilog, itx, tx); } static void zvol_discard(void *arg) { zv_request_t *zvr = arg; struct bio *bio = zvr->bio; zvol_state_t *zv = zvr->zv; uint64_t start = BIO_BI_SECTOR(bio) << 9; uint64_t size = BIO_BI_SIZE(bio); uint64_t end = start + size; boolean_t sync; int error = 0; dmu_tx_t *tx; unsigned long start_jif; ASSERT(zv && zv->zv_open_count > 0); start_jif = jiffies; blk_generic_start_io_acct(zv->zv_queue, WRITE, bio_sectors(bio), &zv->zv_disk->part0); sync = bio_is_fua(bio) || zv->zv_objset->os_sync == ZFS_SYNC_ALWAYS; if (end > zv->zv_volsize) { error = SET_ERROR(EIO); goto unlock; } /* * Align the request to volume block boundaries when a secure erase is * not required. This will prevent dnode_free_range() from zeroing out * the unaligned parts which is slow (read-modify-write) and useless * since we are not freeing any space by doing so. */ if (!bio_is_secure_erase(bio)) { start = P2ROUNDUP(start, zv->zv_volblocksize); end = P2ALIGN(end, zv->zv_volblocksize); size = end - start; } if (start >= end) goto unlock; tx = dmu_tx_create(zv->zv_objset); dmu_tx_mark_netfree(tx); error = dmu_tx_assign(tx, TXG_WAIT); if (error != 0) { dmu_tx_abort(tx); } else { zvol_log_truncate(zv, tx, start, size, B_TRUE); dmu_tx_commit(tx); error = dmu_free_long_range(zv->zv_objset, ZVOL_OBJ, start, size); } unlock: zfs_range_unlock(zvr->rl); if (error == 0 && sync) zil_commit(zv->zv_zilog, ZVOL_OBJ); rw_exit(&zv->zv_suspend_lock); blk_generic_end_io_acct(zv->zv_queue, WRITE, &zv->zv_disk->part0, start_jif); BIO_END_IO(bio, -error); kmem_free(zvr, sizeof (zv_request_t)); } static void zvol_read(void *arg) { zv_request_t *zvr = arg; struct bio *bio = zvr->bio; uio_t uio; zvol_state_t *zv = zvr->zv; uint64_t volsize = zv->zv_volsize; int error = 0; unsigned long start_jif; uio_from_bio(&uio, bio); ASSERT(zv && zv->zv_open_count > 0); start_jif = jiffies; blk_generic_start_io_acct(zv->zv_queue, READ, bio_sectors(bio), &zv->zv_disk->part0); while (uio.uio_resid > 0 && uio.uio_loffset < volsize) { uint64_t bytes = MIN(uio.uio_resid, DMU_MAX_ACCESS >> 1); /* don't read past the end */ if (bytes > volsize - uio.uio_loffset) bytes = volsize - uio.uio_loffset; error = dmu_read_uio_dnode(zv->zv_dn, &uio, bytes); if (error) { /* convert checksum errors into IO errors */ if (error == ECKSUM) error = SET_ERROR(EIO); break; } } zfs_range_unlock(zvr->rl); rw_exit(&zv->zv_suspend_lock); blk_generic_end_io_acct(zv->zv_queue, READ, &zv->zv_disk->part0, start_jif); BIO_END_IO(bio, -error); kmem_free(zvr, sizeof (zv_request_t)); } static MAKE_REQUEST_FN_RET zvol_request(struct request_queue *q, struct bio *bio) { zvol_state_t *zv = q->queuedata; fstrans_cookie_t cookie = spl_fstrans_mark(); uint64_t offset = BIO_BI_SECTOR(bio) << 9; uint64_t size = BIO_BI_SIZE(bio); int rw = bio_data_dir(bio); zv_request_t *zvr; if (bio_has_data(bio) && offset + size > zv->zv_volsize) { printk(KERN_INFO "%s: bad access: offset=%llu, size=%lu\n", zv->zv_disk->disk_name, (long long unsigned)offset, (long unsigned)size); BIO_END_IO(bio, -SET_ERROR(EIO)); goto out; } if (rw == WRITE) { boolean_t need_sync = B_FALSE; if (unlikely(zv->zv_flags & ZVOL_RDONLY)) { BIO_END_IO(bio, -SET_ERROR(EROFS)); goto out; } /* * To be released in the I/O function. See the comment on * zfs_range_lock below. */ rw_enter(&zv->zv_suspend_lock, RW_READER); /* bio marked as FLUSH need to flush before write */ if (bio_is_flush(bio)) zil_commit(zv->zv_zilog, ZVOL_OBJ); /* Some requests are just for flush and nothing else. */ if (size == 0) { rw_exit(&zv->zv_suspend_lock); BIO_END_IO(bio, 0); goto out; } zvr = kmem_alloc(sizeof (zv_request_t), KM_SLEEP); zvr->zv = zv; zvr->bio = bio; /* * To be released in the I/O function. Since the I/O functions * are asynchronous, we take it here synchronously to make * sure overlapped I/Os are properly ordered. */ zvr->rl = zfs_range_lock(&zv->zv_range_lock, offset, size, RL_WRITER); /* * Sync writes and discards execute zil_commit() which may need * to take a RL_READER lock on the whole block being modified * via its zillog->zl_get_data(): to avoid circular dependency * issues with taskq threads execute these requests * synchronously here in zvol_request(). */ need_sync = bio_is_fua(bio) || zv->zv_objset->os_sync == ZFS_SYNC_ALWAYS; if (bio_is_discard(bio) || bio_is_secure_erase(bio)) { if (zvol_request_sync || need_sync || taskq_dispatch(zvol_taskq, zvol_discard, zvr, TQ_SLEEP) == TASKQID_INVALID) zvol_discard(zvr); } else { if (zvol_request_sync || need_sync || taskq_dispatch(zvol_taskq, zvol_write, zvr, TQ_SLEEP) == TASKQID_INVALID) zvol_write(zvr); } } else { zvr = kmem_alloc(sizeof (zv_request_t), KM_SLEEP); zvr->zv = zv; zvr->bio = bio; rw_enter(&zv->zv_suspend_lock, RW_READER); zvr->rl = zfs_range_lock(&zv->zv_range_lock, offset, size, RL_READER); if (zvol_request_sync || taskq_dispatch(zvol_taskq, zvol_read, zvr, TQ_SLEEP) == TASKQID_INVALID) zvol_read(zvr); } out: spl_fstrans_unmark(cookie); #ifdef HAVE_MAKE_REQUEST_FN_RET_INT return (0); #elif defined(HAVE_MAKE_REQUEST_FN_RET_QC) return (BLK_QC_T_NONE); #endif } static void zvol_get_done(zgd_t *zgd, int error) { if (zgd->zgd_db) dmu_buf_rele(zgd->zgd_db, zgd); zfs_range_unlock(zgd->zgd_rl); if (error == 0 && zgd->zgd_bp) - zil_add_block(zgd->zgd_zilog, zgd->zgd_bp); + zil_lwb_add_block(zgd->zgd_lwb, zgd->zgd_bp); kmem_free(zgd, sizeof (zgd_t)); } /* * Get data to generate a TX_WRITE intent log record. */ static int -zvol_get_data(void *arg, lr_write_t *lr, char *buf, zio_t *zio) +zvol_get_data(void *arg, lr_write_t *lr, char *buf, struct lwb *lwb, zio_t *zio) { zvol_state_t *zv = arg; uint64_t offset = lr->lr_offset; uint64_t size = lr->lr_length; dmu_buf_t *db; zgd_t *zgd; int error; - ASSERT(zio != NULL); - ASSERT(size != 0); + ASSERT3P(lwb, !=, NULL); + ASSERT3P(zio, !=, NULL); + ASSERT3U(size, !=, 0); zgd = (zgd_t *)kmem_zalloc(sizeof (zgd_t), KM_SLEEP); - zgd->zgd_zilog = zv->zv_zilog; + zgd->zgd_lwb = lwb; /* * 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_rl = zfs_range_lock(&zv->zv_range_lock, offset, size, RL_READER); error = dmu_read_by_dnode(zv->zv_dn, offset, size, buf, DMU_READ_NO_PREFETCH); } else { /* indirect write */ /* * 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. Contrarily to zfs_get_data we need not re-check * blocksize after we get the lock because it cannot be changed. */ size = zv->zv_volblocksize; offset = P2ALIGN_TYPED(offset, size, uint64_t); zgd->zgd_rl = zfs_range_lock(&zv->zv_range_lock, offset, size, RL_READER); error = dmu_buf_hold_by_dnode(zv->zv_dn, offset, zgd, &db, DMU_READ_NO_PREFETCH); if (error == 0) { blkptr_t *bp = &lr->lr_blkptr; zgd->zgd_db = db; zgd->zgd_bp = bp; ASSERT(db != NULL); ASSERT(db->db_offset == offset); ASSERT(db->db_size == size); error = dmu_sync(zio, lr->lr_common.lrc_txg, zvol_get_done, zgd); if (error == 0) return (0); } } zvol_get_done(zgd, error); return (SET_ERROR(error)); } /* * The zvol_state_t's are inserted into zvol_state_list and zvol_htable. */ static void zvol_insert(zvol_state_t *zv) { ASSERT(MUTEX_HELD(&zvol_state_lock)); ASSERT3U(MINOR(zv->zv_dev) & ZVOL_MINOR_MASK, ==, 0); list_insert_head(&zvol_state_list, zv); hlist_add_head(&zv->zv_hlink, ZVOL_HT_HEAD(zv->zv_hash)); } /* * Simply remove the zvol from to list of zvols. */ static void zvol_remove(zvol_state_t *zv) { ASSERT(MUTEX_HELD(&zvol_state_lock)); list_remove(&zvol_state_list, zv); hlist_del(&zv->zv_hlink); } /* * Setup zv after we just own the zv->objset */ static int zvol_setup_zv(zvol_state_t *zv) { uint64_t volsize; int error; uint64_t ro; objset_t *os = zv->zv_objset; ASSERT(MUTEX_HELD(&zv->zv_state_lock) && RW_LOCK_HELD(&zv->zv_suspend_lock)); error = dsl_prop_get_integer(zv->zv_name, "readonly", &ro, NULL); if (error) return (SET_ERROR(error)); error = zap_lookup(os, ZVOL_ZAP_OBJ, "size", 8, 1, &volsize); if (error) return (SET_ERROR(error)); error = dnode_hold(os, ZVOL_OBJ, FTAG, &zv->zv_dn); if (error) return (SET_ERROR(error)); set_capacity(zv->zv_disk, volsize >> 9); zv->zv_volsize = volsize; zv->zv_zilog = zil_open(os, zvol_get_data); if (ro || dmu_objset_is_snapshot(os) || !spa_writeable(dmu_objset_spa(os))) { set_disk_ro(zv->zv_disk, 1); zv->zv_flags |= ZVOL_RDONLY; } else { set_disk_ro(zv->zv_disk, 0); zv->zv_flags &= ~ZVOL_RDONLY; } return (0); } /* * Shutdown every zv_objset related stuff except zv_objset itself. * The is the reverse of zvol_setup_zv. */ static void zvol_shutdown_zv(zvol_state_t *zv) { ASSERT(MUTEX_HELD(&zv->zv_state_lock) && RW_LOCK_HELD(&zv->zv_suspend_lock)); zil_close(zv->zv_zilog); zv->zv_zilog = NULL; dnode_rele(zv->zv_dn, FTAG); zv->zv_dn = NULL; /* * Evict cached data */ if (dsl_dataset_is_dirty(dmu_objset_ds(zv->zv_objset)) && !(zv->zv_flags & ZVOL_RDONLY)) txg_wait_synced(dmu_objset_pool(zv->zv_objset), 0); (void) dmu_objset_evict_dbufs(zv->zv_objset); } /* * return the proper tag for rollback and recv */ void * zvol_tag(zvol_state_t *zv) { ASSERT(RW_WRITE_HELD(&zv->zv_suspend_lock)); return (zv->zv_open_count > 0 ? zv : NULL); } /* * Suspend the zvol for recv and rollback. */ zvol_state_t * zvol_suspend(const char *name) { zvol_state_t *zv; zv = zvol_find_by_name(name, RW_WRITER); if (zv == NULL) return (NULL); /* block all I/O, release in zvol_resume. */ ASSERT(MUTEX_HELD(&zv->zv_state_lock) && RW_WRITE_HELD(&zv->zv_suspend_lock)); atomic_inc(&zv->zv_suspend_ref); if (zv->zv_open_count > 0) zvol_shutdown_zv(zv); /* * do not hold zv_state_lock across suspend/resume to * avoid locking up zvol lookups */ mutex_exit(&zv->zv_state_lock); /* zv_suspend_lock is released in zvol_resume() */ return (zv); } int zvol_resume(zvol_state_t *zv) { int error = 0; ASSERT(RW_WRITE_HELD(&zv->zv_suspend_lock)); mutex_enter(&zv->zv_state_lock); if (zv->zv_open_count > 0) { VERIFY0(dmu_objset_hold(zv->zv_name, zv, &zv->zv_objset)); VERIFY3P(zv->zv_objset->os_dsl_dataset->ds_owner, ==, zv); VERIFY(dsl_dataset_long_held(zv->zv_objset->os_dsl_dataset)); dmu_objset_rele(zv->zv_objset, zv); error = zvol_setup_zv(zv); } mutex_exit(&zv->zv_state_lock); rw_exit(&zv->zv_suspend_lock); /* * We need this because we don't hold zvol_state_lock while releasing * zv_suspend_lock. zvol_remove_minors_impl thus cannot check * zv_suspend_lock to determine it is safe to free because rwlock is * not inherent atomic. */ atomic_dec(&zv->zv_suspend_ref); return (SET_ERROR(error)); } static int zvol_first_open(zvol_state_t *zv) { objset_t *os; int error, locked = 0; ASSERT(RW_READ_HELD(&zv->zv_suspend_lock)); ASSERT(MUTEX_HELD(&zv->zv_state_lock)); /* * In all other cases the spa_namespace_lock is taken before the * bdev->bd_mutex lock. But in this case the Linux __blkdev_get() * function calls fops->open() with the bdev->bd_mutex lock held. * This deadlock can be easily observed with zvols used as vdevs. * * To avoid a potential lock inversion deadlock we preemptively * try to take the spa_namespace_lock(). Normally it will not * be contended and this is safe because spa_open_common() handles * the case where the caller already holds the spa_namespace_lock. * * When it is contended we risk a lock inversion if we were to * block waiting for the lock. Luckily, the __blkdev_get() * function allows us to return -ERESTARTSYS which will result in * bdev->bd_mutex being dropped, reacquired, and fops->open() being * called again. This process can be repeated safely until both * locks are acquired. */ if (!mutex_owned(&spa_namespace_lock)) { locked = mutex_tryenter(&spa_namespace_lock); if (!locked) return (-SET_ERROR(ERESTARTSYS)); } /* lie and say we're read-only */ error = dmu_objset_own(zv->zv_name, DMU_OST_ZVOL, 1, 1, zv, &os); if (error) goto out_mutex; zv->zv_objset = os; error = zvol_setup_zv(zv); if (error) { dmu_objset_disown(os, 1, zv); zv->zv_objset = NULL; } out_mutex: if (locked) mutex_exit(&spa_namespace_lock); return (SET_ERROR(-error)); } static void zvol_last_close(zvol_state_t *zv) { ASSERT(RW_READ_HELD(&zv->zv_suspend_lock)); ASSERT(MUTEX_HELD(&zv->zv_state_lock)); zvol_shutdown_zv(zv); dmu_objset_disown(zv->zv_objset, 1, zv); zv->zv_objset = NULL; } static int zvol_open(struct block_device *bdev, fmode_t flag) { zvol_state_t *zv; int error = 0; boolean_t drop_suspend = B_TRUE; ASSERT(!MUTEX_HELD(&zvol_state_lock)); mutex_enter(&zvol_state_lock); /* * Obtain a copy of private_data under the zvol_state_lock to make * sure that either the result of zvol free code path setting * bdev->bd_disk->private_data to NULL is observed, or zvol_free() * is not called on this zv because of the positive zv_open_count. */ zv = bdev->bd_disk->private_data; if (zv == NULL) { mutex_exit(&zvol_state_lock); return (SET_ERROR(-ENXIO)); } mutex_enter(&zv->zv_state_lock); /* * make sure zvol is not suspended during first open * (hold zv_suspend_lock) and respect proper lock acquisition * ordering - zv_suspend_lock before zv_state_lock */ if (zv->zv_open_count == 0) { if (!rw_tryenter(&zv->zv_suspend_lock, RW_READER)) { mutex_exit(&zv->zv_state_lock); rw_enter(&zv->zv_suspend_lock, RW_READER); mutex_enter(&zv->zv_state_lock); /* check to see if zv_suspend_lock is needed */ if (zv->zv_open_count != 0) { rw_exit(&zv->zv_suspend_lock); drop_suspend = B_FALSE; } } } else { drop_suspend = B_FALSE; } mutex_exit(&zvol_state_lock); ASSERT(MUTEX_HELD(&zv->zv_state_lock)); ASSERT(zv->zv_open_count != 0 || RW_READ_HELD(&zv->zv_suspend_lock)); if (zv->zv_open_count == 0) { error = zvol_first_open(zv); if (error) goto out_mutex; } if ((flag & FMODE_WRITE) && (zv->zv_flags & ZVOL_RDONLY)) { error = -EROFS; goto out_open_count; } zv->zv_open_count++; check_disk_change(bdev); out_open_count: if (zv->zv_open_count == 0) zvol_last_close(zv); out_mutex: mutex_exit(&zv->zv_state_lock); if (drop_suspend) rw_exit(&zv->zv_suspend_lock); if (error == -ERESTARTSYS) schedule(); return (SET_ERROR(error)); } #ifdef HAVE_BLOCK_DEVICE_OPERATIONS_RELEASE_VOID static void #else static int #endif zvol_release(struct gendisk *disk, fmode_t mode) { zvol_state_t *zv; boolean_t drop_suspend = B_TRUE; ASSERT(!MUTEX_HELD(&zvol_state_lock)); mutex_enter(&zvol_state_lock); zv = disk->private_data; mutex_enter(&zv->zv_state_lock); ASSERT(zv->zv_open_count > 0); /* * make sure zvol is not suspended during last close * (hold zv_suspend_lock) and respect proper lock acquisition * ordering - zv_suspend_lock before zv_state_lock */ if (zv->zv_open_count == 1) { if (!rw_tryenter(&zv->zv_suspend_lock, RW_READER)) { mutex_exit(&zv->zv_state_lock); rw_enter(&zv->zv_suspend_lock, RW_READER); mutex_enter(&zv->zv_state_lock); /* check to see if zv_suspend_lock is needed */ if (zv->zv_open_count != 1) { rw_exit(&zv->zv_suspend_lock); drop_suspend = B_FALSE; } } } else { drop_suspend = B_FALSE; } mutex_exit(&zvol_state_lock); ASSERT(MUTEX_HELD(&zv->zv_state_lock)); ASSERT(zv->zv_open_count != 1 || RW_READ_HELD(&zv->zv_suspend_lock)); zv->zv_open_count--; if (zv->zv_open_count == 0) zvol_last_close(zv); mutex_exit(&zv->zv_state_lock); if (drop_suspend) rw_exit(&zv->zv_suspend_lock); #ifndef HAVE_BLOCK_DEVICE_OPERATIONS_RELEASE_VOID return (0); #endif } static int zvol_ioctl(struct block_device *bdev, fmode_t mode, unsigned int cmd, unsigned long arg) { zvol_state_t *zv = bdev->bd_disk->private_data; int error = 0; ASSERT(zv && zv->zv_open_count > 0); switch (cmd) { case BLKFLSBUF: fsync_bdev(bdev); invalidate_bdev(bdev); rw_enter(&zv->zv_suspend_lock, RW_READER); if (dsl_dataset_is_dirty(dmu_objset_ds(zv->zv_objset)) && !(zv->zv_flags & ZVOL_RDONLY)) txg_wait_synced(dmu_objset_pool(zv->zv_objset), 0); rw_exit(&zv->zv_suspend_lock); break; case BLKZNAME: mutex_enter(&zv->zv_state_lock); error = copy_to_user((void *)arg, zv->zv_name, MAXNAMELEN); mutex_exit(&zv->zv_state_lock); break; default: error = -ENOTTY; break; } return (SET_ERROR(error)); } #ifdef CONFIG_COMPAT static int zvol_compat_ioctl(struct block_device *bdev, fmode_t mode, unsigned cmd, unsigned long arg) { return (zvol_ioctl(bdev, mode, cmd, arg)); } #else #define zvol_compat_ioctl NULL #endif static int zvol_media_changed(struct gendisk *disk) { zvol_state_t *zv = disk->private_data; ASSERT(zv && zv->zv_open_count > 0); return (zv->zv_changed); } static int zvol_revalidate_disk(struct gendisk *disk) { zvol_state_t *zv = disk->private_data; ASSERT(zv && zv->zv_open_count > 0); zv->zv_changed = 0; set_capacity(zv->zv_disk, zv->zv_volsize >> 9); return (0); } /* * Provide a simple virtual geometry for legacy compatibility. For devices * smaller than 1 MiB a small head and sector count is used to allow very * tiny devices. For devices over 1 Mib a standard head and sector count * is used to keep the cylinders count reasonable. */ static int zvol_getgeo(struct block_device *bdev, struct hd_geometry *geo) { zvol_state_t *zv = bdev->bd_disk->private_data; sector_t sectors; ASSERT(zv && zv->zv_open_count > 0); sectors = get_capacity(zv->zv_disk); if (sectors > 2048) { geo->heads = 16; geo->sectors = 63; } else { geo->heads = 2; geo->sectors = 4; } geo->start = 0; geo->cylinders = sectors / (geo->heads * geo->sectors); return (0); } static struct kobject * zvol_probe(dev_t dev, int *part, void *arg) { zvol_state_t *zv; struct kobject *kobj; zv = zvol_find_by_dev(dev); kobj = zv ? get_disk(zv->zv_disk) : NULL; ASSERT(zv == NULL || MUTEX_HELD(&zv->zv_state_lock)); if (zv) mutex_exit(&zv->zv_state_lock); return (kobj); } #ifdef HAVE_BDEV_BLOCK_DEVICE_OPERATIONS static struct block_device_operations zvol_ops = { .open = zvol_open, .release = zvol_release, .ioctl = zvol_ioctl, .compat_ioctl = zvol_compat_ioctl, .media_changed = zvol_media_changed, .revalidate_disk = zvol_revalidate_disk, .getgeo = zvol_getgeo, .owner = THIS_MODULE, }; #else /* HAVE_BDEV_BLOCK_DEVICE_OPERATIONS */ static int zvol_open_by_inode(struct inode *inode, struct file *file) { return (zvol_open(inode->i_bdev, file->f_mode)); } static int zvol_release_by_inode(struct inode *inode, struct file *file) { return (zvol_release(inode->i_bdev->bd_disk, file->f_mode)); } static int zvol_ioctl_by_inode(struct inode *inode, struct file *file, unsigned int cmd, unsigned long arg) { if (file == NULL || inode == NULL) return (SET_ERROR(-EINVAL)); return (zvol_ioctl(inode->i_bdev, file->f_mode, cmd, arg)); } #ifdef CONFIG_COMPAT static long zvol_compat_ioctl_by_inode(struct file *file, unsigned int cmd, unsigned long arg) { if (file == NULL) return (SET_ERROR(-EINVAL)); return (zvol_compat_ioctl(file->f_dentry->d_inode->i_bdev, file->f_mode, cmd, arg)); } #else #define zvol_compat_ioctl_by_inode NULL #endif static struct block_device_operations zvol_ops = { .open = zvol_open_by_inode, .release = zvol_release_by_inode, .ioctl = zvol_ioctl_by_inode, .compat_ioctl = zvol_compat_ioctl_by_inode, .media_changed = zvol_media_changed, .revalidate_disk = zvol_revalidate_disk, .getgeo = zvol_getgeo, .owner = THIS_MODULE, }; #endif /* HAVE_BDEV_BLOCK_DEVICE_OPERATIONS */ /* * Allocate memory for a new zvol_state_t and setup the required * request queue and generic disk structures for the block device. */ static zvol_state_t * zvol_alloc(dev_t dev, const char *name) { zvol_state_t *zv; uint64_t volmode; if (dsl_prop_get_integer(name, "volmode", &volmode, NULL) != 0) return (NULL); if (volmode == ZFS_VOLMODE_DEFAULT) volmode = zvol_volmode; if (volmode == ZFS_VOLMODE_NONE) return (NULL); zv = kmem_zalloc(sizeof (zvol_state_t), KM_SLEEP); list_link_init(&zv->zv_next); mutex_init(&zv->zv_state_lock, NULL, MUTEX_DEFAULT, NULL); zv->zv_queue = blk_alloc_queue(GFP_ATOMIC); if (zv->zv_queue == NULL) goto out_kmem; blk_queue_make_request(zv->zv_queue, zvol_request); blk_queue_set_write_cache(zv->zv_queue, B_TRUE, B_TRUE); /* Limit read-ahead to a single page to prevent over-prefetching. */ blk_queue_set_read_ahead(zv->zv_queue, 1); /* Disable write merging in favor of the ZIO pipeline. */ queue_flag_set_unlocked(QUEUE_FLAG_NOMERGES, zv->zv_queue); zv->zv_disk = alloc_disk(ZVOL_MINORS); if (zv->zv_disk == NULL) goto out_queue; zv->zv_queue->queuedata = zv; zv->zv_dev = dev; zv->zv_open_count = 0; strlcpy(zv->zv_name, name, MAXNAMELEN); zfs_rlock_init(&zv->zv_range_lock); rw_init(&zv->zv_suspend_lock, NULL, RW_DEFAULT, NULL); zv->zv_disk->major = zvol_major; if (volmode == ZFS_VOLMODE_DEV) { /* * ZFS_VOLMODE_DEV disable partitioning on ZVOL devices: set * gendisk->minors = 1 as noted in include/linux/genhd.h. * Also disable extended partition numbers (GENHD_FL_EXT_DEVT) * and suppresses partition scanning (GENHD_FL_NO_PART_SCAN) * setting gendisk->flags accordingly. */ zv->zv_disk->minors = 1; #if defined(GENHD_FL_EXT_DEVT) zv->zv_disk->flags &= ~GENHD_FL_EXT_DEVT; #endif #if defined(GENHD_FL_NO_PART_SCAN) zv->zv_disk->flags |= GENHD_FL_NO_PART_SCAN; #endif } zv->zv_disk->first_minor = (dev & MINORMASK); zv->zv_disk->fops = &zvol_ops; zv->zv_disk->private_data = zv; zv->zv_disk->queue = zv->zv_queue; snprintf(zv->zv_disk->disk_name, DISK_NAME_LEN, "%s%d", ZVOL_DEV_NAME, (dev & MINORMASK)); return (zv); out_queue: blk_cleanup_queue(zv->zv_queue); out_kmem: kmem_free(zv, sizeof (zvol_state_t)); return (NULL); } /* * Cleanup then free a zvol_state_t which was created by zvol_alloc(). * At this time, the structure is not opened by anyone, is taken off * the zvol_state_list, and has its private data set to NULL. * The zvol_state_lock is dropped. */ static void zvol_free(void *arg) { zvol_state_t *zv = arg; ASSERT(!MUTEX_HELD(&zvol_state_lock)); ASSERT(!RW_LOCK_HELD(&zv->zv_suspend_lock)); ASSERT(!MUTEX_HELD(&zv->zv_state_lock)); ASSERT(zv->zv_open_count == 0); ASSERT(zv->zv_disk->private_data == NULL); rw_destroy(&zv->zv_suspend_lock); zfs_rlock_destroy(&zv->zv_range_lock); del_gendisk(zv->zv_disk); blk_cleanup_queue(zv->zv_queue); put_disk(zv->zv_disk); ida_simple_remove(&zvol_ida, MINOR(zv->zv_dev) >> ZVOL_MINOR_BITS); mutex_destroy(&zv->zv_state_lock); kmem_free(zv, sizeof (zvol_state_t)); } /* * Create a block device minor node and setup the linkage between it * and the specified volume. Once this function returns the block * device is live and ready for use. */ static int zvol_create_minor_impl(const char *name) { zvol_state_t *zv; objset_t *os; dmu_object_info_t *doi; uint64_t volsize; uint64_t len; unsigned minor = 0; int error = 0; int idx; uint64_t hash = zvol_name_hash(name); if (zvol_inhibit_dev) return (0); idx = ida_simple_get(&zvol_ida, 0, 0, kmem_flags_convert(KM_SLEEP)); if (idx < 0) return (SET_ERROR(-idx)); minor = idx << ZVOL_MINOR_BITS; zv = zvol_find_by_name_hash(name, hash, RW_NONE); if (zv) { ASSERT(MUTEX_HELD(&zv->zv_state_lock)); mutex_exit(&zv->zv_state_lock); ida_simple_remove(&zvol_ida, idx); return (SET_ERROR(EEXIST)); } doi = kmem_alloc(sizeof (dmu_object_info_t), KM_SLEEP); error = dmu_objset_own(name, DMU_OST_ZVOL, B_TRUE, B_TRUE, FTAG, &os); if (error) goto out_doi; error = dmu_object_info(os, ZVOL_OBJ, doi); if (error) goto out_dmu_objset_disown; error = zap_lookup(os, ZVOL_ZAP_OBJ, "size", 8, 1, &volsize); if (error) goto out_dmu_objset_disown; zv = zvol_alloc(MKDEV(zvol_major, minor), name); if (zv == NULL) { error = SET_ERROR(EAGAIN); goto out_dmu_objset_disown; } zv->zv_hash = hash; if (dmu_objset_is_snapshot(os)) zv->zv_flags |= ZVOL_RDONLY; zv->zv_volblocksize = doi->doi_data_block_size; zv->zv_volsize = volsize; zv->zv_objset = os; set_capacity(zv->zv_disk, zv->zv_volsize >> 9); blk_queue_max_hw_sectors(zv->zv_queue, (DMU_MAX_ACCESS / 4) >> 9); blk_queue_max_segments(zv->zv_queue, UINT16_MAX); blk_queue_max_segment_size(zv->zv_queue, UINT_MAX); blk_queue_physical_block_size(zv->zv_queue, zv->zv_volblocksize); blk_queue_io_opt(zv->zv_queue, zv->zv_volblocksize); blk_queue_max_discard_sectors(zv->zv_queue, (zvol_max_discard_blocks * zv->zv_volblocksize) >> 9); blk_queue_discard_granularity(zv->zv_queue, zv->zv_volblocksize); queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, zv->zv_queue); #ifdef QUEUE_FLAG_NONROT queue_flag_set_unlocked(QUEUE_FLAG_NONROT, zv->zv_queue); #endif #ifdef QUEUE_FLAG_ADD_RANDOM queue_flag_clear_unlocked(QUEUE_FLAG_ADD_RANDOM, zv->zv_queue); #endif if (spa_writeable(dmu_objset_spa(os))) { if (zil_replay_disable) zil_destroy(dmu_objset_zil(os), B_FALSE); else zil_replay(os, zv, zvol_replay_vector); } /* * When udev detects the addition of the device it will immediately * invoke blkid(8) to determine the type of content on the device. * Prefetching the blocks commonly scanned by blkid(8) will speed * up this process. */ len = MIN(MAX(zvol_prefetch_bytes, 0), SPA_MAXBLOCKSIZE); if (len > 0) { dmu_prefetch(os, ZVOL_OBJ, 0, 0, len, ZIO_PRIORITY_SYNC_READ); dmu_prefetch(os, ZVOL_OBJ, 0, volsize - len, len, ZIO_PRIORITY_SYNC_READ); } zv->zv_objset = NULL; out_dmu_objset_disown: dmu_objset_disown(os, B_TRUE, FTAG); out_doi: kmem_free(doi, sizeof (dmu_object_info_t)); if (error == 0) { mutex_enter(&zvol_state_lock); zvol_insert(zv); mutex_exit(&zvol_state_lock); add_disk(zv->zv_disk); } else { ida_simple_remove(&zvol_ida, idx); } return (SET_ERROR(error)); } /* * Rename a block device minor mode for the specified volume. */ static void zvol_rename_minor(zvol_state_t *zv, const char *newname) { int readonly = get_disk_ro(zv->zv_disk); ASSERT(MUTEX_HELD(&zvol_state_lock)); ASSERT(MUTEX_HELD(&zv->zv_state_lock)); strlcpy(zv->zv_name, newname, sizeof (zv->zv_name)); /* move to new hashtable entry */ zv->zv_hash = zvol_name_hash(zv->zv_name); hlist_del(&zv->zv_hlink); hlist_add_head(&zv->zv_hlink, ZVOL_HT_HEAD(zv->zv_hash)); /* * The block device's read-only state is briefly changed causing * a KOBJ_CHANGE uevent to be issued. This ensures udev detects * the name change and fixes the symlinks. This does not change * ZVOL_RDONLY in zv->zv_flags so the actual read-only state never * changes. This would normally be done using kobject_uevent() but * that is a GPL-only symbol which is why we need this workaround. */ set_disk_ro(zv->zv_disk, !readonly); set_disk_ro(zv->zv_disk, readonly); } typedef struct minors_job { list_t *list; list_node_t link; /* input */ char *name; /* output */ int error; } minors_job_t; /* * Prefetch zvol dnodes for the minors_job */ static void zvol_prefetch_minors_impl(void *arg) { minors_job_t *job = arg; char *dsname = job->name; objset_t *os = NULL; job->error = dmu_objset_own(dsname, DMU_OST_ZVOL, B_TRUE, B_TRUE, FTAG, &os); if (job->error == 0) { dmu_prefetch(os, ZVOL_OBJ, 0, 0, 0, ZIO_PRIORITY_SYNC_READ); dmu_objset_disown(os, B_TRUE, FTAG); } } /* * Mask errors to continue dmu_objset_find() traversal */ static int zvol_create_snap_minor_cb(const char *dsname, void *arg) { minors_job_t *j = arg; list_t *minors_list = j->list; const char *name = j->name; ASSERT0(MUTEX_HELD(&spa_namespace_lock)); /* skip the designated dataset */ if (name && strcmp(dsname, name) == 0) return (0); /* at this point, the dsname should name a snapshot */ if (strchr(dsname, '@') == 0) { dprintf("zvol_create_snap_minor_cb(): " "%s is not a shapshot name\n", dsname); } else { minors_job_t *job; char *n = strdup(dsname); if (n == NULL) return (0); job = kmem_alloc(sizeof (minors_job_t), KM_SLEEP); job->name = n; job->list = minors_list; job->error = 0; list_insert_tail(minors_list, job); /* don't care if dispatch fails, because job->error is 0 */ taskq_dispatch(system_taskq, zvol_prefetch_minors_impl, job, TQ_SLEEP); } return (0); } /* * Mask errors to continue dmu_objset_find() traversal */ static int zvol_create_minors_cb(const char *dsname, void *arg) { uint64_t snapdev; int error; list_t *minors_list = arg; ASSERT0(MUTEX_HELD(&spa_namespace_lock)); error = dsl_prop_get_integer(dsname, "snapdev", &snapdev, NULL); if (error) return (0); /* * Given the name and the 'snapdev' property, create device minor nodes * with the linkages to zvols/snapshots as needed. * If the name represents a zvol, create a minor node for the zvol, then * check if its snapshots are 'visible', and if so, iterate over the * snapshots and create device minor nodes for those. */ if (strchr(dsname, '@') == 0) { minors_job_t *job; char *n = strdup(dsname); if (n == NULL) return (0); job = kmem_alloc(sizeof (minors_job_t), KM_SLEEP); job->name = n; job->list = minors_list; job->error = 0; list_insert_tail(minors_list, job); /* don't care if dispatch fails, because job->error is 0 */ taskq_dispatch(system_taskq, zvol_prefetch_minors_impl, job, TQ_SLEEP); if (snapdev == ZFS_SNAPDEV_VISIBLE) { /* * traverse snapshots only, do not traverse children, * and skip the 'dsname' */ error = dmu_objset_find((char *)dsname, zvol_create_snap_minor_cb, (void *)job, DS_FIND_SNAPSHOTS); } } else { dprintf("zvol_create_minors_cb(): %s is not a zvol name\n", dsname); } return (0); } /* * Create minors for the specified dataset, including children and snapshots. * Pay attention to the 'snapdev' property and iterate over the snapshots * only if they are 'visible'. This approach allows one to assure that the * snapshot metadata is read from disk only if it is needed. * * The name can represent a dataset to be recursively scanned for zvols and * their snapshots, or a single zvol snapshot. If the name represents a * dataset, the scan is performed in two nested stages: * - scan the dataset for zvols, and * - for each zvol, create a minor node, then check if the zvol's snapshots * are 'visible', and only then iterate over the snapshots if needed * * If the name represents a snapshot, a check is performed if the snapshot is * 'visible' (which also verifies that the parent is a zvol), and if so, * a minor node for that snapshot is created. */ static int zvol_create_minors_impl(const char *name) { int error = 0; fstrans_cookie_t cookie; char *atp, *parent; list_t minors_list; minors_job_t *job; if (zvol_inhibit_dev) return (0); /* * This is the list for prefetch jobs. Whenever we found a match * during dmu_objset_find, we insert a minors_job to the list and do * taskq_dispatch to parallel prefetch zvol dnodes. Note we don't need * any lock because all list operation is done on the current thread. * * We will use this list to do zvol_create_minor_impl after prefetch * so we don't have to traverse using dmu_objset_find again. */ list_create(&minors_list, sizeof (minors_job_t), offsetof(minors_job_t, link)); parent = kmem_alloc(MAXPATHLEN, KM_SLEEP); (void) strlcpy(parent, name, MAXPATHLEN); if ((atp = strrchr(parent, '@')) != NULL) { uint64_t snapdev; *atp = '\0'; error = dsl_prop_get_integer(parent, "snapdev", &snapdev, NULL); if (error == 0 && snapdev == ZFS_SNAPDEV_VISIBLE) error = zvol_create_minor_impl(name); } else { cookie = spl_fstrans_mark(); error = dmu_objset_find(parent, zvol_create_minors_cb, &minors_list, DS_FIND_CHILDREN); spl_fstrans_unmark(cookie); } kmem_free(parent, MAXPATHLEN); taskq_wait_outstanding(system_taskq, 0); /* * Prefetch is completed, we can do zvol_create_minor_impl * sequentially. */ while ((job = list_head(&minors_list)) != NULL) { list_remove(&minors_list, job); if (!job->error) zvol_create_minor_impl(job->name); strfree(job->name); kmem_free(job, sizeof (minors_job_t)); } list_destroy(&minors_list); return (SET_ERROR(error)); } /* * Remove minors for specified dataset including children and snapshots. */ static void zvol_remove_minors_impl(const char *name) { zvol_state_t *zv, *zv_next; int namelen = ((name) ? strlen(name) : 0); taskqid_t t, tid = TASKQID_INVALID; list_t free_list; if (zvol_inhibit_dev) return; list_create(&free_list, sizeof (zvol_state_t), offsetof(zvol_state_t, zv_next)); mutex_enter(&zvol_state_lock); for (zv = list_head(&zvol_state_list); zv != NULL; zv = zv_next) { zv_next = list_next(&zvol_state_list, zv); mutex_enter(&zv->zv_state_lock); if (name == NULL || strcmp(zv->zv_name, name) == 0 || (strncmp(zv->zv_name, name, namelen) == 0 && (zv->zv_name[namelen] == '/' || zv->zv_name[namelen] == '@'))) { /* * By holding zv_state_lock here, we guarantee that no * one is currently using this zv */ /* If in use, leave alone */ if (zv->zv_open_count > 0 || atomic_read(&zv->zv_suspend_ref)) { mutex_exit(&zv->zv_state_lock); continue; } zvol_remove(zv); /* * clear this while holding zvol_state_lock so * zvol_open won't open it */ zv->zv_disk->private_data = NULL; /* Drop zv_state_lock before zvol_free() */ mutex_exit(&zv->zv_state_lock); /* try parallel zv_free, if failed do it in place */ t = taskq_dispatch(system_taskq, zvol_free, zv, TQ_SLEEP); if (t == TASKQID_INVALID) list_insert_head(&free_list, zv); else tid = t; } else { mutex_exit(&zv->zv_state_lock); } } mutex_exit(&zvol_state_lock); /* * Drop zvol_state_lock before calling zvol_free() */ while ((zv = list_head(&free_list)) != NULL) { list_remove(&free_list, zv); zvol_free(zv); } if (tid != TASKQID_INVALID) taskq_wait_outstanding(system_taskq, tid); } /* Remove minor for this specific volume only */ static void zvol_remove_minor_impl(const char *name) { zvol_state_t *zv = NULL, *zv_next; if (zvol_inhibit_dev) return; mutex_enter(&zvol_state_lock); for (zv = list_head(&zvol_state_list); zv != NULL; zv = zv_next) { zv_next = list_next(&zvol_state_list, zv); mutex_enter(&zv->zv_state_lock); if (strcmp(zv->zv_name, name) == 0) { /* * By holding zv_state_lock here, we guarantee that no * one is currently using this zv */ /* If in use, leave alone */ if (zv->zv_open_count > 0 || atomic_read(&zv->zv_suspend_ref)) { mutex_exit(&zv->zv_state_lock); continue; } zvol_remove(zv); /* clear this so zvol_open won't open it */ zv->zv_disk->private_data = NULL; mutex_exit(&zv->zv_state_lock); break; } else { mutex_exit(&zv->zv_state_lock); } } /* Drop zvol_state_lock before calling zvol_free() */ mutex_exit(&zvol_state_lock); if (zv != NULL) zvol_free(zv); } /* * Rename minors for specified dataset including children and snapshots. */ static void zvol_rename_minors_impl(const char *oldname, const char *newname) { zvol_state_t *zv, *zv_next; int oldnamelen, newnamelen; if (zvol_inhibit_dev) return; oldnamelen = strlen(oldname); newnamelen = strlen(newname); mutex_enter(&zvol_state_lock); for (zv = list_head(&zvol_state_list); zv != NULL; zv = zv_next) { zv_next = list_next(&zvol_state_list, zv); mutex_enter(&zv->zv_state_lock); /* If in use, leave alone */ if (zv->zv_open_count > 0) { mutex_exit(&zv->zv_state_lock); continue; } if (strcmp(zv->zv_name, oldname) == 0) { zvol_rename_minor(zv, newname); } else if (strncmp(zv->zv_name, oldname, oldnamelen) == 0 && (zv->zv_name[oldnamelen] == '/' || zv->zv_name[oldnamelen] == '@')) { char *name = kmem_asprintf("%s%c%s", newname, zv->zv_name[oldnamelen], zv->zv_name + oldnamelen + 1); zvol_rename_minor(zv, name); kmem_free(name, strlen(name + 1)); } mutex_exit(&zv->zv_state_lock); } mutex_exit(&zvol_state_lock); } typedef struct zvol_snapdev_cb_arg { uint64_t snapdev; } zvol_snapdev_cb_arg_t; static int zvol_set_snapdev_cb(const char *dsname, void *param) { zvol_snapdev_cb_arg_t *arg = param; if (strchr(dsname, '@') == NULL) return (0); switch (arg->snapdev) { case ZFS_SNAPDEV_VISIBLE: (void) zvol_create_minor_impl(dsname); break; case ZFS_SNAPDEV_HIDDEN: (void) zvol_remove_minor_impl(dsname); break; } return (0); } static void zvol_set_snapdev_impl(char *name, uint64_t snapdev) { zvol_snapdev_cb_arg_t arg = {snapdev}; fstrans_cookie_t cookie = spl_fstrans_mark(); /* * The zvol_set_snapdev_sync() sets snapdev appropriately * in the dataset hierarchy. Here, we only scan snapshots. */ dmu_objset_find(name, zvol_set_snapdev_cb, &arg, DS_FIND_SNAPSHOTS); spl_fstrans_unmark(cookie); } typedef struct zvol_volmode_cb_arg { uint64_t volmode; } zvol_volmode_cb_arg_t; static void zvol_set_volmode_impl(char *name, uint64_t volmode) { fstrans_cookie_t cookie = spl_fstrans_mark(); if (strchr(name, '@') != NULL) return; /* * It's unfortunate we need to remove minors before we create new ones: * this is necessary because our backing gendisk (zvol_state->zv_disk) * coule be different when we set, for instance, volmode from "geom" * to "dev" (or vice versa). * A possible optimization is to modify our consumers so we don't get * called when "volmode" does not change. */ switch (volmode) { case ZFS_VOLMODE_NONE: (void) zvol_remove_minor_impl(name); break; case ZFS_VOLMODE_GEOM: case ZFS_VOLMODE_DEV: (void) zvol_remove_minor_impl(name); (void) zvol_create_minor_impl(name); break; case ZFS_VOLMODE_DEFAULT: (void) zvol_remove_minor_impl(name); if (zvol_volmode == ZFS_VOLMODE_NONE) break; else /* if zvol_volmode is invalid defaults to "geom" */ (void) zvol_create_minor_impl(name); break; } spl_fstrans_unmark(cookie); } static zvol_task_t * zvol_task_alloc(zvol_async_op_t op, const char *name1, const char *name2, uint64_t value) { zvol_task_t *task; char *delim; /* Never allow tasks on hidden names. */ if (name1[0] == '$') return (NULL); task = kmem_zalloc(sizeof (zvol_task_t), KM_SLEEP); task->op = op; task->value = value; delim = strchr(name1, '/'); strlcpy(task->pool, name1, delim ? (delim - name1 + 1) : MAXNAMELEN); strlcpy(task->name1, name1, MAXNAMELEN); if (name2 != NULL) strlcpy(task->name2, name2, MAXNAMELEN); return (task); } static void zvol_task_free(zvol_task_t *task) { kmem_free(task, sizeof (zvol_task_t)); } /* * The worker thread function performed asynchronously. */ static void zvol_task_cb(void *param) { zvol_task_t *task = (zvol_task_t *)param; switch (task->op) { case ZVOL_ASYNC_CREATE_MINORS: (void) zvol_create_minors_impl(task->name1); break; case ZVOL_ASYNC_REMOVE_MINORS: zvol_remove_minors_impl(task->name1); break; case ZVOL_ASYNC_RENAME_MINORS: zvol_rename_minors_impl(task->name1, task->name2); break; case ZVOL_ASYNC_SET_SNAPDEV: zvol_set_snapdev_impl(task->name1, task->value); break; case ZVOL_ASYNC_SET_VOLMODE: zvol_set_volmode_impl(task->name1, task->value); break; default: VERIFY(0); break; } zvol_task_free(task); } typedef struct zvol_set_prop_int_arg { const char *zsda_name; uint64_t zsda_value; zprop_source_t zsda_source; dmu_tx_t *zsda_tx; } zvol_set_prop_int_arg_t; /* * Sanity check the dataset for safe use by the sync task. No additional * conditions are imposed. */ static int zvol_set_snapdev_check(void *arg, dmu_tx_t *tx) { zvol_set_prop_int_arg_t *zsda = arg; dsl_pool_t *dp = dmu_tx_pool(tx); dsl_dir_t *dd; int error; error = dsl_dir_hold(dp, zsda->zsda_name, FTAG, &dd, NULL); if (error != 0) return (error); dsl_dir_rele(dd, FTAG); return (error); } /* ARGSUSED */ static int zvol_set_snapdev_sync_cb(dsl_pool_t *dp, dsl_dataset_t *ds, void *arg) { char dsname[MAXNAMELEN]; zvol_task_t *task; uint64_t snapdev; dsl_dataset_name(ds, dsname); if (dsl_prop_get_int_ds(ds, "snapdev", &snapdev) != 0) return (0); task = zvol_task_alloc(ZVOL_ASYNC_SET_SNAPDEV, dsname, NULL, snapdev); if (task == NULL) return (0); (void) taskq_dispatch(dp->dp_spa->spa_zvol_taskq, zvol_task_cb, task, TQ_SLEEP); return (0); } /* * Traverse all child datasets and apply snapdev appropriately. * We call dsl_prop_set_sync_impl() here to set the value only on the toplevel * dataset and read the effective "snapdev" on every child in the callback * function: this is because the value is not guaranteed to be the same in the * whole dataset hierarchy. */ static void zvol_set_snapdev_sync(void *arg, dmu_tx_t *tx) { zvol_set_prop_int_arg_t *zsda = arg; dsl_pool_t *dp = dmu_tx_pool(tx); dsl_dir_t *dd; dsl_dataset_t *ds; int error; VERIFY0(dsl_dir_hold(dp, zsda->zsda_name, FTAG, &dd, NULL)); zsda->zsda_tx = tx; error = dsl_dataset_hold(dp, zsda->zsda_name, FTAG, &ds); if (error == 0) { dsl_prop_set_sync_impl(ds, zfs_prop_to_name(ZFS_PROP_SNAPDEV), zsda->zsda_source, sizeof (zsda->zsda_value), 1, &zsda->zsda_value, zsda->zsda_tx); dsl_dataset_rele(ds, FTAG); } dmu_objset_find_dp(dp, dd->dd_object, zvol_set_snapdev_sync_cb, zsda, DS_FIND_CHILDREN); dsl_dir_rele(dd, FTAG); } int zvol_set_snapdev(const char *ddname, zprop_source_t source, uint64_t snapdev) { zvol_set_prop_int_arg_t zsda; zsda.zsda_name = ddname; zsda.zsda_source = source; zsda.zsda_value = snapdev; return (dsl_sync_task(ddname, zvol_set_snapdev_check, zvol_set_snapdev_sync, &zsda, 0, ZFS_SPACE_CHECK_NONE)); } /* * Sanity check the dataset for safe use by the sync task. No additional * conditions are imposed. */ static int zvol_set_volmode_check(void *arg, dmu_tx_t *tx) { zvol_set_prop_int_arg_t *zsda = arg; dsl_pool_t *dp = dmu_tx_pool(tx); dsl_dir_t *dd; int error; error = dsl_dir_hold(dp, zsda->zsda_name, FTAG, &dd, NULL); if (error != 0) return (error); dsl_dir_rele(dd, FTAG); return (error); } /* ARGSUSED */ static int zvol_set_volmode_sync_cb(dsl_pool_t *dp, dsl_dataset_t *ds, void *arg) { char dsname[MAXNAMELEN]; zvol_task_t *task; uint64_t volmode; dsl_dataset_name(ds, dsname); if (dsl_prop_get_int_ds(ds, "volmode", &volmode) != 0) return (0); task = zvol_task_alloc(ZVOL_ASYNC_SET_VOLMODE, dsname, NULL, volmode); if (task == NULL) return (0); (void) taskq_dispatch(dp->dp_spa->spa_zvol_taskq, zvol_task_cb, task, TQ_SLEEP); return (0); } /* * Traverse all child datasets and apply volmode appropriately. * We call dsl_prop_set_sync_impl() here to set the value only on the toplevel * dataset and read the effective "volmode" on every child in the callback * function: this is because the value is not guaranteed to be the same in the * whole dataset hierarchy. */ static void zvol_set_volmode_sync(void *arg, dmu_tx_t *tx) { zvol_set_prop_int_arg_t *zsda = arg; dsl_pool_t *dp = dmu_tx_pool(tx); dsl_dir_t *dd; dsl_dataset_t *ds; int error; VERIFY0(dsl_dir_hold(dp, zsda->zsda_name, FTAG, &dd, NULL)); zsda->zsda_tx = tx; error = dsl_dataset_hold(dp, zsda->zsda_name, FTAG, &ds); if (error == 0) { dsl_prop_set_sync_impl(ds, zfs_prop_to_name(ZFS_PROP_VOLMODE), zsda->zsda_source, sizeof (zsda->zsda_value), 1, &zsda->zsda_value, zsda->zsda_tx); dsl_dataset_rele(ds, FTAG); } dmu_objset_find_dp(dp, dd->dd_object, zvol_set_volmode_sync_cb, zsda, DS_FIND_CHILDREN); dsl_dir_rele(dd, FTAG); } int zvol_set_volmode(const char *ddname, zprop_source_t source, uint64_t volmode) { zvol_set_prop_int_arg_t zsda; zsda.zsda_name = ddname; zsda.zsda_source = source; zsda.zsda_value = volmode; return (dsl_sync_task(ddname, zvol_set_volmode_check, zvol_set_volmode_sync, &zsda, 0, ZFS_SPACE_CHECK_NONE)); } void zvol_create_minors(spa_t *spa, const char *name, boolean_t async) { zvol_task_t *task; taskqid_t id; task = zvol_task_alloc(ZVOL_ASYNC_CREATE_MINORS, name, NULL, ~0ULL); if (task == NULL) return; id = taskq_dispatch(spa->spa_zvol_taskq, zvol_task_cb, task, TQ_SLEEP); if ((async == B_FALSE) && (id != TASKQID_INVALID)) taskq_wait_id(spa->spa_zvol_taskq, id); } void zvol_remove_minors(spa_t *spa, const char *name, boolean_t async) { zvol_task_t *task; taskqid_t id; task = zvol_task_alloc(ZVOL_ASYNC_REMOVE_MINORS, name, NULL, ~0ULL); if (task == NULL) return; id = taskq_dispatch(spa->spa_zvol_taskq, zvol_task_cb, task, TQ_SLEEP); if ((async == B_FALSE) && (id != TASKQID_INVALID)) taskq_wait_id(spa->spa_zvol_taskq, id); } void zvol_rename_minors(spa_t *spa, const char *name1, const char *name2, boolean_t async) { zvol_task_t *task; taskqid_t id; task = zvol_task_alloc(ZVOL_ASYNC_RENAME_MINORS, name1, name2, ~0ULL); if (task == NULL) return; id = taskq_dispatch(spa->spa_zvol_taskq, zvol_task_cb, task, TQ_SLEEP); if ((async == B_FALSE) && (id != TASKQID_INVALID)) taskq_wait_id(spa->spa_zvol_taskq, id); } int zvol_init(void) { int threads = MIN(MAX(zvol_threads, 1), 1024); int i, error; list_create(&zvol_state_list, sizeof (zvol_state_t), offsetof(zvol_state_t, zv_next)); mutex_init(&zvol_state_lock, NULL, MUTEX_DEFAULT, NULL); ida_init(&zvol_ida); zvol_taskq = taskq_create(ZVOL_DRIVER, threads, maxclsyspri, threads * 2, INT_MAX, TASKQ_PREPOPULATE | TASKQ_DYNAMIC); if (zvol_taskq == NULL) { printk(KERN_INFO "ZFS: taskq_create() failed\n"); error = -ENOMEM; goto out; } zvol_htable = kmem_alloc(ZVOL_HT_SIZE * sizeof (struct hlist_head), KM_SLEEP); if (!zvol_htable) { error = -ENOMEM; goto out_taskq; } for (i = 0; i < ZVOL_HT_SIZE; i++) INIT_HLIST_HEAD(&zvol_htable[i]); error = register_blkdev(zvol_major, ZVOL_DRIVER); if (error) { printk(KERN_INFO "ZFS: register_blkdev() failed %d\n", error); goto out_free; } blk_register_region(MKDEV(zvol_major, 0), 1UL << MINORBITS, THIS_MODULE, zvol_probe, NULL, NULL); return (0); out_free: kmem_free(zvol_htable, ZVOL_HT_SIZE * sizeof (struct hlist_head)); out_taskq: taskq_destroy(zvol_taskq); out: ida_destroy(&zvol_ida); mutex_destroy(&zvol_state_lock); list_destroy(&zvol_state_list); return (SET_ERROR(error)); } void zvol_fini(void) { zvol_remove_minors_impl(NULL); blk_unregister_region(MKDEV(zvol_major, 0), 1UL << MINORBITS); unregister_blkdev(zvol_major, ZVOL_DRIVER); kmem_free(zvol_htable, ZVOL_HT_SIZE * sizeof (struct hlist_head)); taskq_destroy(zvol_taskq); list_destroy(&zvol_state_list); mutex_destroy(&zvol_state_lock); ida_destroy(&zvol_ida); } /* BEGIN CSTYLED */ module_param(zvol_inhibit_dev, uint, 0644); MODULE_PARM_DESC(zvol_inhibit_dev, "Do not create zvol device nodes"); module_param(zvol_major, uint, 0444); MODULE_PARM_DESC(zvol_major, "Major number for zvol device"); module_param(zvol_threads, uint, 0444); MODULE_PARM_DESC(zvol_threads, "Max number of threads to handle I/O requests"); module_param(zvol_request_sync, uint, 0644); MODULE_PARM_DESC(zvol_request_sync, "Synchronously handle bio requests"); module_param(zvol_max_discard_blocks, ulong, 0444); MODULE_PARM_DESC(zvol_max_discard_blocks, "Max number of blocks to discard"); module_param(zvol_prefetch_bytes, uint, 0644); MODULE_PARM_DESC(zvol_prefetch_bytes, "Prefetch N bytes at zvol start+end"); module_param(zvol_volmode, uint, 0644); MODULE_PARM_DESC(zvol_volmode, "Default volmode property value"); /* END CSTYLED */ diff --git a/scripts/zloop.sh b/scripts/zloop.sh index a45c450247f9..afcd87f3626d 100755 --- a/scripts/zloop.sh +++ b/scripts/zloop.sh @@ -1,273 +1,273 @@ #!/bin/bash # # CDDL HEADER START # # 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. # # CDDL HEADER END # # # Copyright (c) 2015 by Delphix. All rights reserved. # Copyright (C) 2016 Lawrence Livermore National Security, LLC. # BASE_DIR=$(dirname "$0") SCRIPT_COMMON=common.sh if [ -f "${BASE_DIR}/${SCRIPT_COMMON}" ]; then . "${BASE_DIR}/${SCRIPT_COMMON}" else echo "Missing helper script ${SCRIPT_COMMON}" && exit 1 fi # shellcheck disable=SC2034 PROG=zloop.sh DEFAULTWORKDIR=/var/tmp DEFAULTCOREDIR=/var/tmp/zloop function usage { echo -e "\n$0 [-t ] [ -s ] [-c ]" \ "[ -- [extra ztest parameters]]\n" \ "\n" \ " This script runs ztest repeatedly with randomized arguments.\n" \ " If a crash is encountered, the ztest logs, any associated\n" \ " vdev files, and core file (if one exists) are moved to the\n" \ " output directory ($DEFAULTCOREDIR by default). Any options\n" \ " after the -- end-of-options marker will be passed to ztest.\n" \ "\n" \ " Options:\n" \ " -t Total time to loop for, in seconds. If not provided,\n" \ " zloop runs forever.\n" \ " -s Size of vdev devices.\n" \ " -f Specify working directory for ztest vdev files.\n" \ " -c Specify a core dump directory to use.\n" \ " -h Print this help message.\n" \ "" >&2 } function or_die { # shellcheck disable=SC2068 $@ # shellcheck disable=SC2181 if [[ $? -ne 0 ]]; then # shellcheck disable=SC2145 echo "Command failed: $@" exit 1 fi } # core file helpers origcorepattern="$(cat /proc/sys/kernel/core_pattern)" coreglob="$(egrep -o '^([^|%[:space:]]*)' /proc/sys/kernel/core_pattern)*" if [[ $coreglob = "*" ]]; then echo "Setting core file pattern..." echo "core" > /proc/sys/kernel/core_pattern coreglob="$(egrep -o '^([^|%[:space:]]*)' \ /proc/sys/kernel/core_pattern)*" fi function core_file { # shellcheck disable=SC2012 disable=2086 printf "%s" "$(ls -tr1 $coreglob 2> /dev/null | head -1)" } function core_prog { prog=$ZTEST core_id=$($GDB --batch -c "$1" | grep "Core was generated by" | \ tr \' ' ') # shellcheck disable=SC2076 if [[ "$core_id" =~ "zdb " ]]; then prog=$ZDB fi printf "%s" "$prog" } function store_core { core="$(core_file)" if [[ $ztrc -ne 0 ]] || [[ -f "$core" ]]; then df -h "$workdir" >>ztest.out coreid=$(date "+zloop-%y%m%d-%H%M%S") foundcrashes=$((foundcrashes + 1)) dest=$coredir/$coreid or_die mkdir -p "$dest" or_die mkdir -p "$dest/vdev" echo "*** ztest crash found - moving logs to $dest" or_die mv ztest.history "$dest/" or_die mv ztest.ddt "$dest/" or_die mv ztest.out "$dest/" or_die mv "$workdir/ztest*" "$dest/vdev/" or_die mv "$workdir/zpool.cache" "$dest/vdev/" # check for core if [[ -f "$core" ]]; then coreprog=$(core_prog "$core") corestatus=$($GDB --batch --quiet \ -ex "set print thread-events off" \ -ex "printf \"*\n* Backtrace \n*\n\"" \ -ex "bt" \ -ex "printf \"*\n* Libraries \n*\n\"" \ -ex "info sharedlib" \ -ex "printf \"*\n* Threads (full) \n*\n\"" \ -ex "info threads" \ -ex "printf \"*\n* Backtraces \n*\n\"" \ -ex "thread apply all bt" \ -ex "printf \"*\n* Backtraces (full) \n*\n\"" \ -ex "thread apply all bt full" \ -ex "quit" "$coreprog" "$core" | grep -v "New LWP") # Dump core + logs to stored directory echo "$corestatus" >>"$dest/status" or_die mv "$core" "$dest/" # Record info in cores logfile echo "*** core @ $coredir/$coreid/$core:" | \ tee -a ztest.cores echo "$corestatus" | tee -a ztest.cores echo "" | tee -a ztest.cores fi echo "continuing..." fi } rngdpid="" function on_exit { if [ -n "$rngdpid" ]; then kill -9 "$rngdpid" fi } trap on_exit EXIT # parse arguments # expected format: zloop [-t timeout] [-c coredir] [-- extra ztest args] coredir=$DEFAULTCOREDIR basedir=$DEFAULTWORKDIR rundir="zloop-run" timeout=0 size="512m" while getopts ":ht:s:c:f:" opt; do case $opt in t ) [[ $OPTARG -gt 0 ]] && timeout=$OPTARG ;; s ) [[ $OPTARG ]] && size=$OPTARG ;; c ) [[ $OPTARG ]] && coredir=$OPTARG ;; f ) [[ $OPTARG ]] && basedir=$(readlink -f "$OPTARG") ;; h ) usage exit 2 ;; * ) echo "Invalid argument: -$OPTARG"; usage exit 1 esac done # pass remaining arguments on to ztest shift $((OPTIND - 1)) # enable core dumps ulimit -c unlimited if [[ -f "$(core_file)" ]]; then echo -n "There's a core dump here you might want to look at first... " core_file exit 1 fi if [[ ! -d $coredir ]]; then echo "core dump directory ($coredir) does not exist, creating it." or_die mkdir -p "$coredir" fi if [[ ! -w $coredir ]]; then echo "core dump directory ($coredir) is not writable." exit 1 fi or_die rm -f ztest.history or_die rm -f ztest.ddt or_die rm -f ztest.cores # start rngd in the background so we don't run out of entropy or_die read -r rngdpid < <(rngd -f -r /dev/urandom & echo $!) ztrc=0 # ztest return value foundcrashes=0 # number of crashes found so far starttime=$(date +%s) curtime=$starttime # if no timeout was specified, loop forever. while [[ $timeout -eq 0 ]] || [[ $curtime -le $((starttime + timeout)) ]]; do zopt="-VVVVV" # start each run with an empty directory workdir="$basedir/$rundir" or_die rm -rf "$workdir" or_die mkdir "$workdir" # switch between common arrangements & fully randomized if [[ $((RANDOM % 2)) -eq 0 ]]; then mirrors=2 raidz=0 parity=1 vdevs=2 else mirrors=$(((RANDOM % 3) * 1)) parity=$(((RANDOM % 3) + 1)) raidz=$((((RANDOM % 9) + parity + 1) * (RANDOM % 2))) vdevs=$(((RANDOM % 3) + 3)) fi align=$(((RANDOM % 2) * 3 + 9)) runtime=$((RANDOM % 100)) passtime=$((RANDOM % (runtime / 3 + 1) + 10)) zopt="$zopt -m $mirrors" zopt="$zopt -r $raidz" zopt="$zopt -R $parity" zopt="$zopt -v $vdevs" zopt="$zopt -a $align" zopt="$zopt -T $runtime" zopt="$zopt -P $passtime" zopt="$zopt -s $size" zopt="$zopt -f $workdir" # shellcheck disable=SC2124 cmd="$ZTEST $zopt $@" desc="$(date '+%m/%d %T') $cmd" echo "$desc" | tee -a ztest.history echo "$desc" >>ztest.out $cmd >>ztest.out 2>&1 ztrc=$? egrep '===|WARNING' ztest.out >>ztest.history - $ZDB -U "$workdir/zpool.cache" -DD ztest >>ztest.ddt + $ZDB -U "$workdir/zpool.cache" -DD ztest >>ztest.ddt 2>&1 store_core curtime=$(date +%s) done echo "zloop finished, $foundcrashes crashes found" #restore core pattern echo "$origcorepattern" > /proc/sys/kernel/core_pattern uptime >>ztest.out if [[ $foundcrashes -gt 0 ]]; then exit 1 fi