1 /* 2 * Copyright (c) 2020 iXsystems, Inc. 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 14 * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND 15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 17 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE 18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 24 * SUCH DAMAGE. 25 * 26 */ 27 28 #include <sys/cdefs.h> 29 __FBSDID("$FreeBSD$"); 30 31 #include <sys/types.h> 32 #include <sys/param.h> 33 #include <sys/systm.h> 34 #include <sys/conf.h> 35 #include <sys/kernel.h> 36 #include <sys/lock.h> 37 #include <sys/malloc.h> 38 #include <sys/mutex.h> 39 #include <sys/proc.h> 40 #include <sys/errno.h> 41 #include <sys/uio.h> 42 #include <sys/buf.h> 43 #include <sys/file.h> 44 #include <sys/kmem.h> 45 #include <sys/conf.h> 46 #include <sys/cmn_err.h> 47 #include <sys/stat.h> 48 #include <sys/zfs_ioctl.h> 49 #include <sys/zfs_vfsops.h> 50 #include <sys/zfs_znode.h> 51 #include <sys/zap.h> 52 #include <sys/spa.h> 53 #include <sys/spa_impl.h> 54 #include <sys/vdev.h> 55 #include <sys/vdev_impl.h> 56 #include <sys/dmu.h> 57 #include <sys/dsl_dir.h> 58 #include <sys/dsl_dataset.h> 59 #include <sys/dsl_prop.h> 60 #include <sys/dsl_deleg.h> 61 #include <sys/dmu_objset.h> 62 #include <sys/dmu_impl.h> 63 #include <sys/dmu_tx.h> 64 #include <sys/sunddi.h> 65 #include <sys/policy.h> 66 #include <sys/zone.h> 67 #include <sys/nvpair.h> 68 #include <sys/mount.h> 69 #include <sys/taskqueue.h> 70 #include <sys/sdt.h> 71 #include <sys/fs/zfs.h> 72 #include <sys/zfs_ctldir.h> 73 #include <sys/zfs_dir.h> 74 #include <sys/zfs_onexit.h> 75 #include <sys/zvol.h> 76 #include <sys/dsl_scan.h> 77 #include <sys/dmu_objset.h> 78 #include <sys/dmu_send.h> 79 #include <sys/dsl_destroy.h> 80 #include <sys/dsl_bookmark.h> 81 #include <sys/dsl_userhold.h> 82 #include <sys/zfeature.h> 83 #include <sys/zcp.h> 84 #include <sys/zio_checksum.h> 85 #include <sys/vdev_removal.h> 86 #include <sys/dsl_crypt.h> 87 88 #include <sys/zfs_ioctl_compat.h> 89 #include <sys/zfs_context.h> 90 91 #include <sys/arc_impl.h> 92 #include <sys/dsl_pool.h> 93 94 95 /* BEGIN CSTYLED */ 96 SYSCTL_DECL(_vfs_zfs); 97 SYSCTL_NODE(_vfs_zfs, OID_AUTO, arc, CTLFLAG_RW, 0, "ZFS adaptive replacement cache"); 98 SYSCTL_NODE(_vfs_zfs, OID_AUTO, condense, CTLFLAG_RW, 0, "ZFS condense"); 99 SYSCTL_NODE(_vfs_zfs, OID_AUTO, dbuf, CTLFLAG_RW, 0, "ZFS disk buf cache"); 100 SYSCTL_NODE(_vfs_zfs, OID_AUTO, dbuf_cache, CTLFLAG_RW, 0, "ZFS disk buf cache"); 101 SYSCTL_NODE(_vfs_zfs, OID_AUTO, deadman, CTLFLAG_RW, 0, "ZFS deadman"); 102 SYSCTL_NODE(_vfs_zfs, OID_AUTO, dedup, CTLFLAG_RW, 0, "ZFS dedup"); 103 SYSCTL_NODE(_vfs_zfs, OID_AUTO, l2arc, CTLFLAG_RW, 0, "ZFS l2arc"); 104 SYSCTL_NODE(_vfs_zfs, OID_AUTO, livelist, CTLFLAG_RW, 0, "ZFS livelist"); 105 SYSCTL_NODE(_vfs_zfs, OID_AUTO, lua, CTLFLAG_RW, 0, "ZFS lua"); 106 SYSCTL_NODE(_vfs_zfs, OID_AUTO, metaslab, CTLFLAG_RW, 0, "ZFS metaslab"); 107 SYSCTL_NODE(_vfs_zfs, OID_AUTO, mg, CTLFLAG_RW, 0, "ZFS metaslab group"); 108 SYSCTL_NODE(_vfs_zfs, OID_AUTO, multihost, CTLFLAG_RW, 0, "ZFS multihost protection"); 109 SYSCTL_NODE(_vfs_zfs, OID_AUTO, prefetch, CTLFLAG_RW, 0, "ZFS prefetch"); 110 SYSCTL_NODE(_vfs_zfs, OID_AUTO, reconstruct, CTLFLAG_RW, 0, "ZFS reconstruct"); 111 SYSCTL_NODE(_vfs_zfs, OID_AUTO, recv, CTLFLAG_RW, 0, "ZFS receive"); 112 SYSCTL_NODE(_vfs_zfs, OID_AUTO, send, CTLFLAG_RW, 0, "ZFS send"); 113 SYSCTL_NODE(_vfs_zfs, OID_AUTO, spa, CTLFLAG_RW, 0, "ZFS space allocation"); 114 SYSCTL_NODE(_vfs_zfs, OID_AUTO, trim, CTLFLAG_RW, 0, "ZFS TRIM"); 115 SYSCTL_NODE(_vfs_zfs, OID_AUTO, txg, CTLFLAG_RW, 0, "ZFS transaction group"); 116 SYSCTL_NODE(_vfs_zfs, OID_AUTO, vdev, CTLFLAG_RW, 0, "ZFS VDEV"); 117 SYSCTL_NODE(_vfs_zfs, OID_AUTO, vnops, CTLFLAG_RW, 0, "ZFS VNOPS"); 118 SYSCTL_NODE(_vfs_zfs, OID_AUTO, zevent, CTLFLAG_RW, 0, "ZFS event"); 119 SYSCTL_NODE(_vfs_zfs, OID_AUTO, zil, CTLFLAG_RW, 0, "ZFS ZIL"); 120 SYSCTL_NODE(_vfs_zfs, OID_AUTO, zio, CTLFLAG_RW, 0, "ZFS ZIO"); 121 122 SYSCTL_NODE(_vfs_zfs_livelist, OID_AUTO, condense, CTLFLAG_RW, 0, 123 "ZFS livelist condense"); 124 SYSCTL_NODE(_vfs_zfs_vdev, OID_AUTO, cache, CTLFLAG_RW, 0, "ZFS VDEV Cache"); 125 SYSCTL_NODE(_vfs_zfs_vdev, OID_AUTO, file, CTLFLAG_RW, 0, "ZFS VDEV file"); 126 SYSCTL_NODE(_vfs_zfs_vdev, OID_AUTO, mirror, CTLFLAG_RD, 0, 127 "ZFS VDEV mirror"); 128 129 SYSCTL_DECL(_vfs_zfs_version); 130 SYSCTL_CONST_STRING(_vfs_zfs_version, OID_AUTO, module, CTLFLAG_RD, 131 (ZFS_META_VERSION "-" ZFS_META_RELEASE), "OpenZFS module version"); 132 133 extern arc_state_t ARC_anon; 134 extern arc_state_t ARC_mru; 135 extern arc_state_t ARC_mru_ghost; 136 extern arc_state_t ARC_mfu; 137 extern arc_state_t ARC_mfu_ghost; 138 extern arc_state_t ARC_l2c_only; 139 140 /* 141 * minimum lifespan of a prefetch block in clock ticks 142 * (initialized in arc_init()) 143 */ 144 145 /* arc.c */ 146 147 /* legacy compat */ 148 extern uint64_t l2arc_write_max; /* def max write size */ 149 extern uint64_t l2arc_write_boost; /* extra warmup write */ 150 extern uint64_t l2arc_headroom; /* # of dev writes */ 151 extern uint64_t l2arc_headroom_boost; 152 extern uint64_t l2arc_feed_secs; /* interval seconds */ 153 extern uint64_t l2arc_feed_min_ms; /* min interval msecs */ 154 extern int l2arc_noprefetch; /* don't cache prefetch bufs */ 155 extern int l2arc_feed_again; /* turbo warmup */ 156 extern int l2arc_norw; /* no reads during writes */ 157 158 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_write_max, CTLFLAG_RW, 159 &l2arc_write_max, 0, "max write size (LEGACY)"); 160 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_write_boost, CTLFLAG_RW, 161 &l2arc_write_boost, 0, "extra write during warmup (LEGACY)"); 162 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_headroom, CTLFLAG_RW, 163 &l2arc_headroom, 0, "number of dev writes (LEGACY)"); 164 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_feed_secs, CTLFLAG_RW, 165 &l2arc_feed_secs, 0, "interval seconds (LEGACY)"); 166 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_feed_min_ms, CTLFLAG_RW, 167 &l2arc_feed_min_ms, 0, "min interval milliseconds (LEGACY)"); 168 169 SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_noprefetch, CTLFLAG_RW, 170 &l2arc_noprefetch, 0, "don't cache prefetch bufs (LEGACY)"); 171 SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_feed_again, CTLFLAG_RW, 172 &l2arc_feed_again, 0, "turbo warmup (LEGACY)"); 173 SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_norw, CTLFLAG_RW, 174 &l2arc_norw, 0, "no reads during writes (LEGACY)"); 175 #if 0 176 extern int zfs_compressed_arc_enabled; 177 SYSCTL_INT(_vfs_zfs, OID_AUTO, compressed_arc_enabled, CTLFLAG_RW, 178 &zfs_compressed_arc_enabled, 1, "compressed arc buffers (LEGACY)"); 179 #endif 180 181 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_size, CTLFLAG_RD, 182 &ARC_anon.arcs_size.rc_count, 0, "size of anonymous state"); 183 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_metadata_esize, CTLFLAG_RD, 184 &ARC_anon.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, 185 "size of anonymous state"); 186 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_data_esize, CTLFLAG_RD, 187 &ARC_anon.arcs_esize[ARC_BUFC_DATA].rc_count, 0, 188 "size of anonymous state"); 189 190 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_size, CTLFLAG_RD, 191 &ARC_mru.arcs_size.rc_count, 0, "size of mru state"); 192 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_metadata_esize, CTLFLAG_RD, 193 &ARC_mru.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, 194 "size of metadata in mru state"); 195 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_data_esize, CTLFLAG_RD, 196 &ARC_mru.arcs_esize[ARC_BUFC_DATA].rc_count, 0, 197 "size of data in mru state"); 198 199 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_size, CTLFLAG_RD, 200 &ARC_mru_ghost.arcs_size.rc_count, 0, "size of mru ghost state"); 201 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_metadata_esize, CTLFLAG_RD, 202 &ARC_mru_ghost.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, 203 "size of metadata in mru ghost state"); 204 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_data_esize, CTLFLAG_RD, 205 &ARC_mru_ghost.arcs_esize[ARC_BUFC_DATA].rc_count, 0, 206 "size of data in mru ghost state"); 207 208 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_size, CTLFLAG_RD, 209 &ARC_mfu.arcs_size.rc_count, 0, "size of mfu state"); 210 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_metadata_esize, CTLFLAG_RD, 211 &ARC_mfu.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, 212 "size of metadata in mfu state"); 213 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_data_esize, CTLFLAG_RD, 214 &ARC_mfu.arcs_esize[ARC_BUFC_DATA].rc_count, 0, 215 "size of data in mfu state"); 216 217 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_size, CTLFLAG_RD, 218 &ARC_mfu_ghost.arcs_size.rc_count, 0, "size of mfu ghost state"); 219 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_metadata_esize, CTLFLAG_RD, 220 &ARC_mfu_ghost.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, 221 "size of metadata in mfu ghost state"); 222 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_data_esize, CTLFLAG_RD, 223 &ARC_mfu_ghost.arcs_esize[ARC_BUFC_DATA].rc_count, 0, 224 "size of data in mfu ghost state"); 225 226 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2c_only_size, CTLFLAG_RD, 227 &ARC_l2c_only.arcs_size.rc_count, 0, "size of mru state"); 228 229 static int 230 sysctl_vfs_zfs_arc_no_grow_shift(SYSCTL_HANDLER_ARGS) 231 { 232 int err, val; 233 234 val = arc_no_grow_shift; 235 err = sysctl_handle_int(oidp, &val, 0, req); 236 if (err != 0 || req->newptr == NULL) 237 return (err); 238 239 if (val < 0 || val >= arc_shrink_shift) 240 return (EINVAL); 241 242 arc_no_grow_shift = val; 243 return (0); 244 } 245 246 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_no_grow_shift, 247 CTLTYPE_INT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, NULL, sizeof (int), 248 sysctl_vfs_zfs_arc_no_grow_shift, "I", 249 "log2(fraction of ARC which must be free to allow growing)"); 250 251 int 252 param_set_arc_long(SYSCTL_HANDLER_ARGS) 253 { 254 int err; 255 256 err = sysctl_handle_long(oidp, arg1, 0, req); 257 if (err != 0 || req->newptr == NULL) 258 return (err); 259 260 arc_tuning_update(B_TRUE); 261 262 return (0); 263 } 264 265 int 266 param_set_arc_int(SYSCTL_HANDLER_ARGS) 267 { 268 int err; 269 270 err = sysctl_handle_int(oidp, arg1, 0, req); 271 if (err != 0 || req->newptr == NULL) 272 return (err); 273 274 arc_tuning_update(B_TRUE); 275 276 return (0); 277 } 278 279 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_min, 280 CTLTYPE_ULONG | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, 281 &zfs_arc_min, sizeof (zfs_arc_min), param_set_arc_long, "LU", 282 "min arc size (LEGACY)"); 283 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_max, 284 CTLTYPE_ULONG | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, 285 &zfs_arc_max, sizeof (zfs_arc_max), param_set_arc_long, "LU", 286 "max arc size (LEGACY)"); 287 288 /* dbuf.c */ 289 290 291 /* dmu.c */ 292 293 /* dmu_zfetch.c */ 294 SYSCTL_NODE(_vfs_zfs, OID_AUTO, zfetch, CTLFLAG_RW, 0, "ZFS ZFETCH (LEGACY)"); 295 296 /* max bytes to prefetch per stream (default 8MB) */ 297 extern uint32_t zfetch_max_distance; 298 SYSCTL_UINT(_vfs_zfs_zfetch, OID_AUTO, max_distance, CTLFLAG_RWTUN, 299 &zfetch_max_distance, 0, "Max bytes to prefetch per stream (LEGACY)"); 300 301 /* max bytes to prefetch indirects for per stream (default 64MB) */ 302 extern uint32_t zfetch_max_idistance; 303 SYSCTL_UINT(_vfs_zfs_zfetch, OID_AUTO, max_idistance, CTLFLAG_RWTUN, 304 &zfetch_max_idistance, 0, 305 "Max bytes to prefetch indirects for per stream (LEGACY)"); 306 307 /* dsl_pool.c */ 308 309 /* dnode.c */ 310 extern int zfs_default_bs; 311 SYSCTL_INT(_vfs_zfs, OID_AUTO, default_bs, CTLFLAG_RWTUN, 312 &zfs_default_bs, 0, "Default dnode block shift"); 313 314 extern int zfs_default_ibs; 315 SYSCTL_INT(_vfs_zfs, OID_AUTO, default_ibs, CTLFLAG_RWTUN, 316 &zfs_default_ibs, 0, "Default dnode indirect block shift"); 317 318 319 /* dsl_scan.c */ 320 321 /* metaslab.c */ 322 323 /* 324 * In pools where the log space map feature is not enabled we touch 325 * multiple metaslabs (and their respective space maps) with each 326 * transaction group. Thus, we benefit from having a small space map 327 * block size since it allows us to issue more I/O operations scattered 328 * around the disk. So a sane default for the space map block size 329 * is 8~16K. 330 */ 331 extern int zfs_metaslab_sm_blksz_no_log; 332 SYSCTL_INT(_vfs_zfs_metaslab, OID_AUTO, sm_blksz_no_log, CTLFLAG_RDTUN, 333 &zfs_metaslab_sm_blksz_no_log, 0, 334 "Block size for space map in pools with log space map disabled. " 335 "Power of 2 and greater than 4096."); 336 337 /* 338 * When the log space map feature is enabled, we accumulate a lot of 339 * changes per metaslab that are flushed once in a while so we benefit 340 * from a bigger block size like 128K for the metaslab space maps. 341 */ 342 extern int zfs_metaslab_sm_blksz_with_log; 343 SYSCTL_INT(_vfs_zfs_metaslab, OID_AUTO, sm_blksz_with_log, CTLFLAG_RDTUN, 344 &zfs_metaslab_sm_blksz_with_log, 0, 345 "Block size for space map in pools with log space map enabled. " 346 "Power of 2 and greater than 4096."); 347 348 /* 349 * The in-core space map representation is more compact than its on-disk form. 350 * The zfs_condense_pct determines how much more compact the in-core 351 * space map representation must be before we compact it on-disk. 352 * Values should be greater than or equal to 100. 353 */ 354 extern int zfs_condense_pct; 355 SYSCTL_INT(_vfs_zfs, OID_AUTO, condense_pct, CTLFLAG_RWTUN, 356 &zfs_condense_pct, 0, 357 "Condense on-disk spacemap when it is more than this many percents" 358 " of in-memory counterpart"); 359 360 extern int zfs_remove_max_segment; 361 SYSCTL_INT(_vfs_zfs, OID_AUTO, remove_max_segment, CTLFLAG_RWTUN, 362 &zfs_remove_max_segment, 0, "Largest contiguous segment ZFS will attempt to" 363 " allocate when removing a device"); 364 365 extern int zfs_removal_suspend_progress; 366 SYSCTL_INT(_vfs_zfs, OID_AUTO, removal_suspend_progress, CTLFLAG_RWTUN, 367 &zfs_removal_suspend_progress, 0, "Ensures certain actions can happen while" 368 " in the middle of a removal"); 369 370 371 /* 372 * Minimum size which forces the dynamic allocator to change 373 * it's allocation strategy. Once the space map cannot satisfy 374 * an allocation of this size then it switches to using more 375 * aggressive strategy (i.e search by size rather than offset). 376 */ 377 extern uint64_t metaslab_df_alloc_threshold; 378 SYSCTL_QUAD(_vfs_zfs_metaslab, OID_AUTO, df_alloc_threshold, CTLFLAG_RWTUN, 379 &metaslab_df_alloc_threshold, 0, 380 "Minimum size which forces the dynamic allocator to change it's allocation strategy"); 381 382 /* 383 * The minimum free space, in percent, which must be available 384 * in a space map to continue allocations in a first-fit fashion. 385 * Once the space map's free space drops below this level we dynamically 386 * switch to using best-fit allocations. 387 */ 388 extern int metaslab_df_free_pct; 389 SYSCTL_INT(_vfs_zfs_metaslab, OID_AUTO, df_free_pct, CTLFLAG_RWTUN, 390 &metaslab_df_free_pct, 0, 391 "The minimum free space, in percent, which must be available in a " 392 "space map to continue allocations in a first-fit fashion"); 393 394 /* 395 * Percentage of all cpus that can be used by the metaslab taskq. 396 */ 397 extern int metaslab_load_pct; 398 SYSCTL_INT(_vfs_zfs_metaslab, OID_AUTO, load_pct, CTLFLAG_RWTUN, 399 &metaslab_load_pct, 0, 400 "Percentage of cpus that can be used by the metaslab taskq"); 401 402 /* 403 * Max number of metaslabs per group to preload. 404 */ 405 extern int metaslab_preload_limit; 406 SYSCTL_INT(_vfs_zfs_metaslab, OID_AUTO, preload_limit, CTLFLAG_RWTUN, 407 &metaslab_preload_limit, 0, 408 "Max number of metaslabs per group to preload"); 409 410 /* refcount.c */ 411 extern int reference_tracking_enable; 412 SYSCTL_INT(_vfs_zfs, OID_AUTO, reference_tracking_enable, CTLFLAG_RDTUN, 413 &reference_tracking_enable, 0, 414 "Track reference holders to refcount_t objects, used mostly by ZFS"); 415 416 /* spa.c */ 417 extern int zfs_ccw_retry_interval; 418 SYSCTL_INT(_vfs_zfs, OID_AUTO, ccw_retry_interval, CTLFLAG_RWTUN, 419 &zfs_ccw_retry_interval, 0, 420 "Configuration cache file write, retry after failure, interval (seconds)"); 421 422 extern uint64_t zfs_max_missing_tvds_cachefile; 423 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, max_missing_tvds_cachefile, CTLFLAG_RWTUN, 424 &zfs_max_missing_tvds_cachefile, 0, 425 "allow importing pools with missing top-level vdevs in cache file"); 426 427 extern uint64_t zfs_max_missing_tvds_scan; 428 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, max_missing_tvds_scan, CTLFLAG_RWTUN, 429 &zfs_max_missing_tvds_scan, 0, 430 "allow importing pools with missing top-level vdevs during scan"); 431 432 /* spa_misc.c */ 433 extern int zfs_flags; 434 static int 435 sysctl_vfs_zfs_debug_flags(SYSCTL_HANDLER_ARGS) 436 { 437 int err, val; 438 439 val = zfs_flags; 440 err = sysctl_handle_int(oidp, &val, 0, req); 441 if (err != 0 || req->newptr == NULL) 442 return (err); 443 444 /* 445 * ZFS_DEBUG_MODIFY must be enabled prior to boot so all 446 * arc buffers in the system have the necessary additional 447 * checksum data. However, it is safe to disable at any 448 * time. 449 */ 450 if (!(zfs_flags & ZFS_DEBUG_MODIFY)) 451 val &= ~ZFS_DEBUG_MODIFY; 452 zfs_flags = val; 453 454 return (0); 455 } 456 457 SYSCTL_PROC(_vfs_zfs, OID_AUTO, debugflags, 458 CTLTYPE_UINT | CTLFLAG_MPSAFE | CTLFLAG_RWTUN, NULL, 0, 459 sysctl_vfs_zfs_debug_flags, "IU", "Debug flags for ZFS testing."); 460 461 int 462 param_set_deadman_synctime(SYSCTL_HANDLER_ARGS) 463 { 464 unsigned long val; 465 int err; 466 467 val = zfs_deadman_synctime_ms; 468 err = sysctl_handle_long(oidp, &val, 0, req); 469 if (err != 0 || req->newptr == NULL) 470 return (err); 471 zfs_deadman_synctime_ms = val; 472 473 spa_set_deadman_synctime(MSEC2NSEC(zfs_deadman_synctime_ms)); 474 475 return (0); 476 } 477 478 int 479 param_set_deadman_ziotime(SYSCTL_HANDLER_ARGS) 480 { 481 unsigned long val; 482 int err; 483 484 val = zfs_deadman_ziotime_ms; 485 err = sysctl_handle_long(oidp, &val, 0, req); 486 if (err != 0 || req->newptr == NULL) 487 return (err); 488 zfs_deadman_ziotime_ms = val; 489 490 spa_set_deadman_ziotime(MSEC2NSEC(zfs_deadman_synctime_ms)); 491 492 return (0); 493 } 494 495 int 496 param_set_deadman_failmode(SYSCTL_HANDLER_ARGS) 497 { 498 char buf[16]; 499 int rc; 500 501 if (req->newptr == NULL) 502 strlcpy(buf, zfs_deadman_failmode, sizeof (buf)); 503 504 rc = sysctl_handle_string(oidp, buf, sizeof (buf), req); 505 if (rc || req->newptr == NULL) 506 return (rc); 507 if (strcmp(buf, zfs_deadman_failmode) == 0) 508 return (0); 509 if (!strcmp(buf, "wait")) 510 zfs_deadman_failmode = "wait"; 511 if (!strcmp(buf, "continue")) 512 zfs_deadman_failmode = "continue"; 513 if (!strcmp(buf, "panic")) 514 zfs_deadman_failmode = "panic"; 515 516 return (-param_set_deadman_failmode_common(buf)); 517 } 518 519 520 /* spacemap.c */ 521 extern int space_map_ibs; 522 SYSCTL_INT(_vfs_zfs, OID_AUTO, space_map_ibs, CTLFLAG_RWTUN, 523 &space_map_ibs, 0, "Space map indirect block shift"); 524 525 526 /* vdev.c */ 527 int 528 param_set_min_auto_ashift(SYSCTL_HANDLER_ARGS) 529 { 530 uint64_t val; 531 int err; 532 533 val = zfs_vdev_min_auto_ashift; 534 err = sysctl_handle_64(oidp, &val, 0, req); 535 if (err != 0 || req->newptr == NULL) 536 return (SET_ERROR(err)); 537 538 if (val < ASHIFT_MIN || val > zfs_vdev_max_auto_ashift) 539 return (SET_ERROR(EINVAL)); 540 541 zfs_vdev_min_auto_ashift = val; 542 543 return (0); 544 } 545 546 int 547 param_set_max_auto_ashift(SYSCTL_HANDLER_ARGS) 548 { 549 uint64_t val; 550 int err; 551 552 val = zfs_vdev_max_auto_ashift; 553 err = sysctl_handle_64(oidp, &val, 0, req); 554 if (err != 0 || req->newptr == NULL) 555 return (SET_ERROR(err)); 556 557 if (val > ASHIFT_MAX || val < zfs_vdev_min_auto_ashift) 558 return (SET_ERROR(EINVAL)); 559 560 zfs_vdev_max_auto_ashift = val; 561 562 return (0); 563 } 564 565 SYSCTL_PROC(_vfs_zfs, OID_AUTO, min_auto_ashift, 566 CTLTYPE_U64 | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, 567 &zfs_vdev_min_auto_ashift, sizeof (zfs_vdev_min_auto_ashift), 568 param_set_min_auto_ashift, "QU", 569 "Min ashift used when creating new top-level vdev. (LEGACY)"); 570 SYSCTL_PROC(_vfs_zfs, OID_AUTO, max_auto_ashift, 571 CTLTYPE_U64 | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, 572 &zfs_vdev_max_auto_ashift, sizeof (zfs_vdev_max_auto_ashift), 573 param_set_max_auto_ashift, "QU", 574 "Max ashift used when optimizing for logical -> physical sector size on " 575 "new top-level vdevs. (LEGACY)"); 576 577 /* 578 * Since the DTL space map of a vdev is not expected to have a lot of 579 * entries, we default its block size to 4K. 580 */ 581 extern int zfs_vdev_dtl_sm_blksz; 582 SYSCTL_INT(_vfs_zfs, OID_AUTO, dtl_sm_blksz, CTLFLAG_RDTUN, 583 &zfs_vdev_dtl_sm_blksz, 0, 584 "Block size for DTL space map. Power of 2 and greater than 4096."); 585 586 /* 587 * vdev-wide space maps that have lots of entries written to them at 588 * the end of each transaction can benefit from a higher I/O bandwidth 589 * (e.g. vdev_obsolete_sm), thus we default their block size to 128K. 590 */ 591 extern int zfs_vdev_standard_sm_blksz; 592 SYSCTL_INT(_vfs_zfs, OID_AUTO, standard_sm_blksz, CTLFLAG_RDTUN, 593 &zfs_vdev_standard_sm_blksz, 0, 594 "Block size for standard space map. Power of 2 and greater than 4096."); 595 596 extern int vdev_validate_skip; 597 SYSCTL_INT(_vfs_zfs, OID_AUTO, validate_skip, CTLFLAG_RDTUN, 598 &vdev_validate_skip, 0, 599 "Enable to bypass vdev_validate()."); 600 601 602 /* vdev_cache.c */ 603 604 /* vdev_mirror.c */ 605 /* 606 * The load configuration settings below are tuned by default for 607 * the case where all devices are of the same rotational type. 608 * 609 * If there is a mixture of rotating and non-rotating media, setting 610 * non_rotating_seek_inc to 0 may well provide better results as it 611 * will direct more reads to the non-rotating vdevs which are more 612 * likely to have a higher performance. 613 */ 614 615 616 /* vdev_queue.c */ 617 #define ZFS_VDEV_QUEUE_KNOB_MIN(name) \ 618 extern uint32_t zfs_vdev_ ## name ## _min_active; \ 619 SYSCTL_UINT(_vfs_zfs_vdev, OID_AUTO, name ## _min_active, CTLFLAG_RWTUN,\ 620 &zfs_vdev_ ## name ## _min_active, 0, \ 621 "Initial number of I/O requests of type " #name \ 622 " active for each device"); 623 624 #define ZFS_VDEV_QUEUE_KNOB_MAX(name) \ 625 extern uint32_t zfs_vdev_ ## name ## _max_active; \ 626 SYSCTL_UINT(_vfs_zfs_vdev, OID_AUTO, name ## _max_active, CTLFLAG_RWTUN, \ 627 &zfs_vdev_ ## name ## _max_active, 0, \ 628 "Maximum number of I/O requests of type " #name \ 629 " active for each device"); 630 631 632 #undef ZFS_VDEV_QUEUE_KNOB 633 634 extern uint32_t zfs_vdev_max_active; 635 SYSCTL_UINT(_vfs_zfs, OID_AUTO, top_maxinflight, CTLFLAG_RWTUN, 636 &zfs_vdev_max_active, 0, 637 "The maximum number of I/Os of all types active for each device. (LEGACY)"); 638 639 extern int zfs_vdev_def_queue_depth; 640 SYSCTL_INT(_vfs_zfs_vdev, OID_AUTO, def_queue_depth, CTLFLAG_RWTUN, 641 &zfs_vdev_def_queue_depth, 0, 642 "Default queue depth for each allocator"); 643 644 /*extern uint64_t zfs_multihost_history; 645 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, multihost_history, CTLFLAG_RWTUN, 646 &zfs_multihost_history, 0, 647 "Historical staticists for the last N multihost updates");*/ 648 649 #ifdef notyet 650 SYSCTL_INT(_vfs_zfs_vdev, OID_AUTO, trim_on_init, CTLFLAG_RW, 651 &vdev_trim_on_init, 0, "Enable/disable full vdev trim on initialisation"); 652 #endif 653 654 655 /* zio.c */ 656 #if defined(__LP64__) 657 int zio_use_uma = 1; 658 #else 659 int zio_use_uma = 0; 660 #endif 661 662 SYSCTL_INT(_vfs_zfs_zio, OID_AUTO, use_uma, CTLFLAG_RDTUN, &zio_use_uma, 0, 663 "Use uma(9) for ZIO allocations"); 664 SYSCTL_INT(_vfs_zfs_zio, OID_AUTO, exclude_metadata, CTLFLAG_RDTUN, &zio_exclude_metadata, 0, 665 "Exclude metadata buffers from dumps as well"); 666 667 int 668 param_set_slop_shift(SYSCTL_HANDLER_ARGS) 669 { 670 int val; 671 int err; 672 673 val = *(int *)arg1; 674 675 err = sysctl_handle_int(oidp, &val, 0, req); 676 if (err != 0 || req->newptr == NULL) 677 return (err); 678 679 if (val < 1 || val > 31) 680 return (EINVAL); 681 682 *(int *)arg1 = val; 683 684 return (0); 685 } 686 687 int 688 param_set_multihost_interval(SYSCTL_HANDLER_ARGS) 689 { 690 int err; 691 692 err = sysctl_handle_long(oidp, arg1, 0, req); 693 if (err != 0 || req->newptr == NULL) 694 return (err); 695 696 if (spa_mode_global != SPA_MODE_UNINIT) 697 mmp_signal_all_threads(); 698 699 return (0); 700 } 701